Challenges and Solutions For Cellular Based V2X Communications

222 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 23, NO.
1, FIRST QUARTER 2021
Challenges and Solutions for Cellular

Based V2X Communications
Sohan Gyawali , Member, IEEE, Shengjie Xu , Member, IEEE,
Yi Qian , Fellow, IEEE, and Rose Qingyang Hu , Fellow, IEEE
Abstract—A wide variety of works have been conducted in and vehicle traffic optimization. In addition, V2X communi-
vehicle-to-everything (V2X) communications to enable a variety cations should support a variety of use cases like do not pass
of applications for road safety, traffic efficiency and passenger warning, forward collision warning, queue warning, parking
infotainment. Although dedicated short-range communications
(DSRC) based V2X is already in the deployment phase, cellular discovery, optimal speed advisory and curve speed warn-
based V2X is gaining more interest in academia and industry ing [1]. The V2X communications support various use cases
most recently. This article surveys the existing work and chal- by exchanging messages among infrastructure, vehicles and
lenges on LTE and 5G to support efficient V2X communications. pedestrian using various wireless communication technologies
First, we present the motivations for cellular based V2X com- such as DSRC and cellular network technologies.
munications. Second, we summarize the LTE V2X architecture
and operating scenarios being considered. Third, we discuss the DSRC technology supports a short exchange of information
challenges in existing LTE for supporting V2X communications among DSRC devices. DSRC devices are equipped with
such as physical layer structure, synchronization, multimedia 802.11p chip and mainly include onboard units (OBU), road-
broadcast multicast services (MBMS), resource allocation, secu- side units (RSU) and mobile devices carried by pedestrians.
rity and survey the recent solutions to these challenges. We To enable this technology, 75 MHz of the spectrum have
further discuss the challenges and possible solutions for 5G based
vehicular communications. Finally, we discuss the open research been allocated in a 5.9 GHz frequency band by U.S. Federal
issues and possible research directions in cellular based vehicular Communications Commission (FCC). Moreover, a set of
communications. services and interfaces have already been defined by IEEE
Index Terms—Cellular V2X, LTE vehicular communications, 802.11p and IEEE 1609 standards for Wireless Access for
5G vehicular communications, vehicle-to-vehicle communica- Vehicular Environment (WAVE) to be used in DSRC based
tions, V2X communication infrastructure, V2X security, V2X applications [2]. In addition, the U.S. National Highway
resource allocations. Traffic Safety Administration (NHTSA) worked with the U.S.
Department of Transportation to enable vehicular communi-
I. I NTRODUCTION cations capabilities in new light vehicles based on DSRC
HERE has been massive research from industry and technology [3]. DSRC based V2X communications provide
T other organizations to address communication capabili-
ties in vehicles and transportation infrastructures which mainly
a number of benefits such as low end-to-end latency, ad-hoc
communications and standardized protocols. However, it still
include vehicle-to-vehicle (V2V), vehicle-to-infrastructure faces a number of issues such as short-range, large channel
(V2I), vehicle-to-pedestrian (V2P) and vehicle-to-network access delay and huge capital investments. Thus, despite the
(V2N) communications collectively termed as V2X commu- deployment of DSRC based V2X prototype in the U.S., [4], the
nications. V2X communications can enhance the safety and inherent issues of DSRC and the recent growth in cellular tech-
efficiency of transportation systems. The V2X communications nologies have encouraged research and industry communities
along with existing vehicle-sensing capabilities provide sup- to investigate cellular technology based V2X communications.
port for enhanced safety applications, passenger infotainment, Cellular communications such as LTE provide ubiquitous
coverage, support very high mobility as well as the high
Manuscript received August 5, 2019; revised April 16, 2020 and July 20, number of vehicles in a cell. Moreover, the introduction
2020; accepted September 26, 2020. Date of publication October 8, 2020; date of device-to-device (D2D) communications further improved
of current version February 24, 2021. This work was supported by the National
Science Foundation under Grant EARS-1547312 and Grant EARS-1547330. spectrum utilization efficiency and system capacity of cellu-
(Corresponding author: Yi Qian.) lar systems [5], [6]. This motivated organizations like the 3rd
Sohan Gyawali is with the Department of Computer Science, University of Generation Partnership Project (3GPP) to study the feasibil-
Texas Permian Basin, Odessa, TX 79762 USA (e-mail: [email protected]).
Shengjie Xu is with the Beacom College of Computer and Cyber ity of LTE support for V2X communications [7]. 3GPP is
Sciences, Dakota State University, Madison, SD 57042 USA (e-mail: currently working on cellular technology based V2X services
[email protected]). and aims to provide a variety of V2X services [8]. Moreover,
Yi Qian is with the Department of Electrical and Computer Engineering,
University of Nebraska–Lincoln, Omaha, NE 68182 USA (e-mail: 3GPP has completed Releases 14 and 15 with LTE based
[email protected]). V2X services as one of the main features including other
Rose Qingyang Hu is with the Department of Electrical and Computer features like license assisted access, machine type communi-
Engineering, Utah State University, Logan, UT 84321 USA (e-mail:
[email protected]). cations, massive multiple input multiple output (MIMO) [9].
Digital Object Identifier 10.1109/COMST.2020.3029723 The Releases 14 and 15 specify highly reliable and real-time
1553-877X
c 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on September 05,2021 at 07:16:57 UTC from IEEE Xplore. Restrictions apply.
GYAWALI et al.: CHALLENGES AND SOLUTIONS FOR CELLULAR BASED V2X COMMUNICATIONS 223
TABLE I
communications for automotive safety use cases and will con- ACRONYMS AND C ORRESPONDING F ULL F ORMS
tinue to evolve to 5G to provide complementary and new
capabilities like sensor sharing while maintaining backward
compatibility [10]. In addition, there is also active research
being conducted in interworking between DSRC and cellular
technology to support efficient V2X communications [11].
Although LTE based V2X communications have gained a
lot of interest recently, there exist several challenges before the
LTE network can be massively exploited for V2X commu-
nications. Physical layer structure, synchronization, MBMS,
resource allocations and security are the main challenges for
LTE V2X communications. Existing physical layer design
of LTE cannot support high carrier frequency and vehicu-
lar UE velocity due to Doppler effects, so there is a need
for a new physical layer structure for LTE V2X communi-
cations. Similarly, high vehicular UE speed causes frequent
topology change and may cause the problem in the synchro-
nization between the vehicle user equipment and base station.
In addition, large and overlapped broadcast areas cause the
problem of deploying MBMS for broadcast communications.
Moreover, resource allocation is one of the main challenges
where resources used by the vehicular users should not conflict
with the resources being used by the cellular users. Security is
another main challenge in LTE V2X due to the broadcast and
un-encrypted nature of communications. Many projects that
have addressed these challenges are explained in detail in the
rest of this article.
Along with the above-mentioned projects in LTE V2X, tech-
nology organizations like 3GPP and Qualcomm have prepared
the roadmap towards 5G based V2X services. 3GPP is cur-
rently working towards the completion of Release 16 which
specifies various new services such as sharing of high through-
put sensor data, path planning, real-time location updates, and
coordinated driving [12]. The 5G wireless system is an excel-
lent candidate to enable high data rate applications in vehicular
environment [13]. The 5G wireless system is expected to
incorporate various emerging technologies such as device-to-
device communications, massive MIMO, full-duplex radios,
multi-radio access technology, cloud technologies, millime-
ter waves and software-defined networking (SDN) [14]. Each
of these technologies will bring various solutions and chal-
lenges to 5G based vehicular communications. Deployment of
heterogeneous and small cell networks might cause frequent
handover and increase the signal load in V2X communica-
tions. Moreover, there might be issues of short communication
range and high path loss in millimeter wave communications,
pilot contamination in massive MIMO, and resource utilization
in vehicular fog computing. In addition, there might be issues
such as control plane design in SDNs, content prefetching and
distribution in mobile edge computing, the configuration of
multiple slices in network slicing and spectrum efficiency in
dynamic spectrum sharing. There are several other challenges
in each of the above mentioned 5G technologies which are
discussed in detail in Section V.
Only a very few surveys have been published dis-
cussing LTE and 5G based V2X communications [15], [16]. and requirements, and different V2X access technologies.
Reference [15] provides a general overview of V2X Whereas [16] highlights the security requirements of V2X
access technologies with discussion on V2X applications and mainly discusses the authentication issues of V2X entities.
224 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 23, NO. 1, FIRST QUARTER 2021
Fig. 1. Outline of this article.
This survey is different from the above two in several ways. Section V discusses the open issues and solutions related to 5G
This survey highlights the major challenges in the physical based V2X communications. Section VI presents discussion
layer structure, synchronization, MBMS, resource allocations and future research directions. Finally, Section VII concludes
and security of LTE V2X communications and discusses in this article. For the readability, the outline of this article is
detail the prospective solutions for each of these challenges. shown in Fig. 1 and most used acronyms and their full forms
Moreover, benefits, challenges, and solutions for several key are shown in Table I.
technologies of 5G V2X communications are discussed in
detail along with the possible future work. Reference [15] II. M OTIVATIONS FOR C ELLULAR N ETWORK
provides a survey only on V2X access technologies and [16] BASED V2X C OMMUNICATIONS
provides a survey only on the authentication issues of cel-
The latest improvements in cellular technologies have
lular V2X communication. Whereas, our survey covers the
inspired industry and research communities to explore cellu-
wide area of V2X communications and provides comprehen-
lar network based V2X communications. This section mainly
sive information regarding major challenges, possible solutions
highlights the advantages and disadvantages of DSRC based
and future research work in various areas and technologies of
V2X communications and motivations for cellular network
evolving cellular based V2X communications.
based V2X communications.
The major contributions of this survey article can be
summarized as follows.
• Discussion on motivations for cellular based V2X com- A. DSRC Based V2X Communications
munications and detailed comparison of cellular and DSRC based V2X communications have been the subject of
DSRC based V2X communications. extensive standardization, field trial and product development
• Examination of LTE V2X architecture including both for almost two decades. Although a lot of projects have already
general LTE V2X communication model and LTE V2X been studied in DSRC, it still has some limitations from tech-
architecture from 3GPP. nical and business perspectives. From technical perspectives,
• Discussion on several operating scenarios of LTE V2X DSRC suffers from short range characteristics of around 200 to
communications. 400 meters and is limited to line of sight communications [17].
• Description of several main challenges in LTE based V2X Due to which vehicles will have intermittent connectivity
communications with prospective solutions. when moving with a high-speed [11]. Moreover, carrier-sense
• Discussion on benefits, challenges and solutions of vari- multiple access with collision avoidance (CSMA/CA) is used
ous technologies of 5G in V2X communications. as a channel access mechanism in DSRC. This mechanism
• Discussion on future research directions on cellular based results in large channel access delay in a high vehicular traffic
V2X communications. scenario due to the large intensity of channel conflict [18]. In
The rest of this article is organized as follows. Section II addition, there is a presence of a hidden node problem due
provides motivations for adopting cellular based V2X com- to the absence of a handshake and acknowledgment mecha-
munications. Section III presents the existing LTE based V2X nism. This hidden node problem results in packet collisions,
architecture and different operating scenarios for LTE based poor link performance and unreliable broadcast service in high
V2X communications. Section IV presents various challenges vehicle density scenarios. From business perspectives, huge
and potential solutions for LTE based V2X communications. investment is required for the deployment of DSRC based
TABLE II
C OMPARISON OF IEEE 802.11 P BASED DSRC AND LTE V2X C OMMUNICATIONS [21], [22], [23]
V2X communications due to network backbone infrastructure

like RSUs [19]. Capital expenditure and operating expenses
(CAPEX/OPEX) will be very high as the large number of
RSUs needs to be deployed along the roads.
B. Cellular Network Based V2X Communications

Cellular network based V2X communications such as LTE
has several advantages as compared to DSRC based V2X com-
munications from both technical and business perspectives.
From technical perspectives, the LTE network provides ubiq-
uitous coverage for V2I/V2N services and also supports very
high mobility of vehicles up to 350 km/hr [19]. Moreover, the
hidden node problem is prevented in LTE with the enhance-
ment of slotted MAC. In addition, MBMS of LTE technology
can be used for efficient safety message dissemination [11].
Likewise, LTE provides much better performance in the non-
line-of-sight (NLOS) environment and supports a high data
rate of up to 1 Gbps and a high number of vehicles in a
cell [20]. From business perspectives, LTE based V2X commu- Fig. 2. General LTE V2X communication model.
nications can reduce the cost of mass production by utilizing
the common hardware platform of LTE. A brief comparison article mainly surveys and discusses solutions to various chal-
of IEEE 802.11p based DSRC and LTE V2X communications lenges of cellular V2X which are presented in detail in
is shown in Table II. Sections IV and V.
Although there are many advantages of using LTE for V2X,
there are also some challenges for LTE V2X. LTE has a cen- III. LTE V2X A RCHITECTURE AND O PERATING
tralized nature due to which vehicular data needs to be passed S CENARIOS
through the base station which can raise the issue of latency.
Recently, various LTE V2X architectures and operating
It has also been shown that existing beaconing capabilities
scenarios have been proposed by 3GPP and non-3GPP organi-
of LTE to support vehicular safety application is poorer as
zations. This section highlights the general LTE V2X commu-
compared to 802.11p/WAVE [23]. Moreover, LTE base sta-
nication model, LTE V2X architecture and LTE V2X operating
tion coverage is much larger than a zone of the relevance of
scenarios.
safety messages due to which many vehicles can receive irrele-
vant messages [11]. These irrelevant messages can be reduced
with the help of multicast service by broadcasting messages A. General LTE V2X Communication Model
to the multicast group only. However, control signaling over- LTE V2X communications mainly involve V2V, V2I, V2N
head and latency will be high for multicast service due to the and V2P communications, as shown in Fig. 2. Vehicles within
join and leave process of evolved MBMS [24]. In addition, proximity of each other can exchange safety or infotainment
the performance evaluation of cellular network traffic should related information either directly or with the help of infras-
be considered while using LTE for V2X. tructures like evolved node-B (eNB) or RSUs. RSUs can
These various challenges of LTE or cellular V2X should exchange unicast information with vehicles using a V2I appli-
be addressed to gain full advantages of cellular V2X. This cation. Moreover, RSUs can broadcast information related to
Fig. 4. eMBMS and LTE-Uu based LTE V2X architecture, described by [25].
information related to V2X and provides UE authorization sta-

tus to e-UTRAN. The V2X application server is used for both
collecting the uplink data from UE and delivering data to UE
in a destination area. S-GW routes and forwards the data as
well as perform charging with the help from policy and charg-
Fig. 3. PC5 and LTE-Uu based reference architecture for V2X, described ing rules function (PCRF). P-GW is responsible for providing
by [25].
communications between IP and circuit-switched networks.
2) eMBMS and LTE-Uu Based V2X Architecture: In an
an emergency scenario or traffic condition to a group of user evolved MBMS (eMBMS) and LTE-Uu based V2X archi-
equipment (UE). V2N communications are also introduced in tecture, as shown in Fig. 4, there are two additional
which both parties - vehicular UE and serving entity, support functional entities broadcast multicast service center (BM-
V2N applications and communicate with the help of the LTE SC) and MBMS gateway (MBMS-GW). BM-SC receives
network. In addition, the evolved packet core (EPC) of the LTE local MBMS information from the V2X application server
network may connect to an intelligent transportation system and sends local MBMS information to the MBMS-GW. If
(ITS) server for various vehicular services. V2P supports the MBMS-GW receives local MBMS information from BM-
exchange of messages between vehicles and pedestrians. These SC, it skips the allocation procedure for IP multicast
messages are transmitted by vehicular UE to pedestrian UE or distribution.
by pedestrian UE to vehicle UE.
C. Operating Scenarios
B. LTE V2X Architecture From 3GPP There are several operating scenarios in spectrum usage for
3GPP document [25] describes PC5 and LTE Uu based, LTE V2X. In [27], the authors presented four different oper-
and evolved MBMS (eMBMS) and LTE-Uu based V2X ating scenarios, as shown in Fig. 5. In scenario 1, there is no
architecture. dedicated spectrum for V2V and each UE, i.e., cellular and
1) PC5 and LTE-Uu Based V2X Architecture: In PC5 and V2V UE, utilizes the spectrum of their own operator. In this
LTE-Uu based architecture, as shown in Fig. 3, there are 8 scenario, if required, the same spectrum can be used for both
reference points. The reference point between the V2X appli- links. However, there might be an issue of providing appro-
cation server and the V2X application is V1 while the refer- priate quality-of-service (QoS) for inter-operator operation. In
ence point between the V2X control function in the operator’s scenario 2, the dedicated spectrum is allocated to V2V. In this
network and the V2X application server is V2. The reference scenario, due to the dedicated spectrum for V2V operation,
point between the V2X enabled UE and the V2X control func- the inter-operator issue is restricted only to cellular operation.
tion in UE’s home public land mobile network (PLMN) is V3. In scenario 3, a single operator manages LTE V2X operation
Similarly, the reference point between the home subscriber for certain areas. Due to the dedicated spectrum for an entire
server (HSS) and the V2X control function in the operator’s area, capacity and QoS for V2X operations can be greatly
network is V4. The reference point between the V2X applica- improved. In scenario 4, V2V links operate without any help
tions is V5 while the reference point between the V2X control from eNB. This scenario is applicable to the areas without any
functions is V6. The reference point between the V2X enabled network coverage.
UE and the evolved universal terrestrial radio access network There are also other scenarios being discussed. In 3GPP
(E-UTRAN) is LTE-Uu and the reference point between the document [7], three different scenarios are considered. In the
V2X enabled UE for V2V, V2I and V2P services is V8. scenario I, only one operator has eNBs in a specific area and
In LTE V2X, functional entities include V2X control func- other operators share these eNBs for all services including
tion, mobility management entity (MME), V2X application V2X. In scenario II, only one operator owns the dedicated
server, service gateway (S-GW) and packet gateway (P-GW). V2X spectrum in a specific area and other operators share
The V2X control function provision the UE with neces- this spectrum for V2X service. In scenario III, both operators
sary parameters required for V2X communications when have eNBs in a specific area and a V2X server distributes the
served or not served by e-UTRAN. MME obtains subscription V2X message to both operator’s networks.
Fig. 5. Operating scenarios for LTE V2X, described by [27].
Fig. 6. Illustration of (a) existing LTE physical layer structure and (b) physical layer enhancement to track Doppler case.
IV. C HALLENGES AND S OLUTIONS IN LTE BASED However, for wider frequency allocation range, the LTE V2X
V2X C OMMUNICATIONS system may be required to support very high carrier frequency
There are several challenges before LTE can be mas- up to 6 GHz. Moreover, vehicles may achieve high relative
sively exploited for vehicular communications. Standards velocity when driving at high speed in the opposite direc-
organizations including 3GPP have actively carried out tion causing Doppler effects. Doppler effects can introduce
research to address these challenges. This section explores interference among carriers. In addition, short coherence time
some of the major challenges and possible solutions in LTE due to Doppler effects results in an inaccurate channel estima-
based V2X communications. tion [27]. If the existing physical layer design of LTE is used
for LTE V2X, then the time interval of reference signal will be
higher than the coherence time, as a result, the performance
A. Physical Layer Structure for demodulation of data will strongly fall [32]. In current
1) Challenges: Existing LTE physical layer design cannot LTE physical layer design, at max 1 kHz frequency offset can
support high carrier frequency as well as high UE velocity. be corrected. However, the maximum frequency error between
UE in the neighboring cell can be more than 2.2 kHz [32]. In LTE V2X, the TTI can be reduced by using two approaches:
Thus, there should be some enhancement in the physical layer lowering the number of symbols in a frame with fixed sub-
structure of LTE for V2X communications. carrier spacing or by increasing the subcarrier spacing and
2) Solutions: 3GPP is considering enhancing the reference reducing the symbol duration. For these two approaches, the
signal structure by reducing the time interval between refer- authors have considered a symbol-wise frame structure and a
ence signals. To reduce this time interval, they are considering self-contained frame structure. In a symbol-wise frame struc-
a sub-frame of 1 ms in which the four reference signal symbols ture, 3 OFDM symbols are used and the subcarrier spacing
are consistently located [7], [26], as shown in Fig. 6. Other is fixed to 15 kHz. In a self-contained frame structure, sub-
techniques are also being considered in which for a single ref- carrier spacing is set to 60 kHz, the guard period is set to
erence signal symbol, the high-frequency offset is estimated 20.33 µs with 1 symbol for uplink and 12 symbols for down-
by comparing the phase of the first and second half of each link. The symbol-wise frame structure has a scattered pilot
reference signal [27]. pattern and a self-contained frame structure uses a preamble
In [28], the authors proposed a frame structure for LTE- pilot pattern. Authors have done simulations and have shown
V2V and compared the performance of LTE-V2V with IEEE that the symbol-wise frame structure is more robust compared
802.11p. In the proposed scheme, a frame of length 10 ms to a self-contained frame structure in terms of the Doppler
is considered with 10 subframes of each length of 1 ms. A shift. The authors also discuss that faster feedback is possible
slot of 0.5 ms is considered, i.e., each subframe consist of 2 in a self-contained frame structure compared to symbol-wise
slots. The subcarrier spacing of 15 kHz is considered with a as channel estimation can only be done at the end of the frame
variable number of subcarriers in a signal. In the time domain, in case of a symbol-wise frame structure.
the smallest unit that can be scheduled is 0.5 ms and in the In [31], the authors discuss the agile frame structure based
frequency domain, the smallest unit is 12 subcarriers. The on NR Release 16 [12] and compares the new frame struc-
structure of frame and subframes affects the coding rate as ture with the LTE frame structure. For latency and in-band
well as the equivalent data rate in LTE-V2V as well as IEEE interference reductions in V2X communications, shorter TTI
802.11 p. The authors have done simulations and have shown and larger SCS are desired. In LTE, fixed 1 ms length TTI and
that the QoS setting impacts slightly in LTE-V2V whereas static SCS are used whereas in NR scalable SCS and TTI are
heavily in IEEE 802.11p, LTE-V2V is better at longer dis- used for growing V2X use cases. NR Release 16 supports TTI
tances communications and IEEE 802.11p is robust at limited duration of 1 ms to 31.25 µs, i.e., in frequency domain SCS
or shorter distances. settings from 15 to 480 kHz. Release 16 NR C-V2X provides
In [29], the authors discuss the physical layer structure for enhancements in both Uu and PC5 interfaces of LTE V2X.
3GPP new radio (NR) based side link transmissions. As dif- NR C-V2X supports variable subcarriers spacing for various
ferent countries have different bandwidth allocation policies, V2X applications. In addition, in NR C-V2X transmissions
orthogonal frequency division multiplexing (OFDM) with dif- are not bound to the entire subframe duration due to the use
ferent subcarrier spacing (SCR) is desired. To combat with of mini-slots. For a small amount of data, NR C-V2X sup-
different levels of inter-symbol interference, different cyclic ports less than 14 OFDM symbols to occupy only the required
prefix (CP) lengths are associated with different inter sym- number of symbols called mini-slots. In NR C-V2X, 5.9 GHz
bol interference levels in NR such as normal CP (NCP) spectrum supports SCS of 15, 30 and 60 kHz whereas spec-
and extended CP (ECP). For side link transmission in NR trum higher than 6 GHz supports higher frequency bands up
which is mainly used for V2V communications, CP-OFDM to 480 kHz. In NR C-V2X, due to the allocation of wider
is supported. In 3GPP NR side link transmission based V2X SCS, frequency offset and Doppler Shift are handled better
communications, the length of the frame is 10 ms and the sin- compared to existing LTE V2X. Similarly, due to shorter TTI,
gle frame contains 10 subframes. A slot contains a varying channel variation within TTI is smaller needing fewer DMRS
number of OFDM symbols which is equal to 12 OFDM sym- signals compared to LTE V2X. Authors have done extensive
bols in case of ECP and 14 OFDM symbols in case of NCP. simulations and have compared the existing LTE-V2X of 1 ms
The length of the slot is based on the value of SCS and for 15 TTI and SCS of 15 kHz with NR C-V2X of flexible numerol-
kHz SCS, the length of a slot is 1 ms. In addition, for uplink ogy in 5.9 GHz band. In terms of packet reception probability
and downlink transmission in 3GPP based NR, mini-slot is (PRR), the NR scheme achieves better performances compared
used to further reduce the latency. However, mini-slots are not to the LTE-V2X as a single cooperative awareness message
supported for NR side link transmissions. In this work, the can be accommodated in a shorter TTI of 0.5 ms instead of 1
authors have done simulations and have considered 4 differ- ms in LTE-V2X. Similarly, in terms of update delay, the NR
ent cases with different numbers of symbols per side - link C-V2X schemes outperform the legacy LTE due to a longer
control information (SCI), the different moving speed of vehi- burst of errors in existing LTE.
cles and different payload sizes per SCI. Authors have shown Table III shows the qualitative comparison of the differ-
that the impact of error propagation is limited when the 1st- ent schemes regarding the physical layer structure of V2X
stage and 2nd-stage SCI are correlated and there is 0.5 dB communications.
performance loss in case of high speed.
Transmission time interval (TTI), i.e., physical layer struc-
ture affects the end-to-end latency of LTE V2X communica- B. Synchronization
tions. In [30], the authors discuss the frame structure to meet 1) Challenges: High vehicular UE speed causes a frequent
the strict latency requirements of LTE V2X communications. topology change due to which UE needs to continuously
TABLE III
C OMPARISON OF P HYSICAL L AYER S TRUCTURE FOR V2X C OMMUNICATIONS
Fig. 8. Side link establishment for synchronization, described by [34].
Fig. 7. Time Synchronization using GNSS and eNB. adjust their time difference on the basis of other users weighted
mean timing. In the case when more than one synchronization
source is available, mobile users can prioritize the synchro-
update the reference point for synchronization. However, nization source as eNB signals, GNSS signals followed by side
V2X communications need to be robust to this synchroniza- link (SL) signals. In addition, in this article, the authors have
tion source change as a result of frequent topology change. presented the method for synchronization between side link
Moreover, in the legacy LTE system, the neighboring vehi- users attached to unsynchronized BS. In the proposed algo-
cle’s relative timing offset cannot be assumed to be within a rithm, UE2 attached to eNB2 sends a request to UE1 attached
cyclic prefix, if cellular coverage is not present or if eNBs are to eNB1, as shown in Fig. 8.
not synchronized [32]. If UE1 rejects the request, no side link is established. If UE1
2) Solutions: To address this synchronization problem, accepts the request, the side link is established only if a rel-
the synchronization source can be chosen as a global nav- ative timing offset lies below a certain level. However, if this
igation satellite system (GNSS) [33]. Accurate timing and offset exceeds the label, side link establishment and synchro-
frequency reference can be provided by utilizing the GNSS. nization procedure depend on whether UE2 is out of coverage
As shown in Fig. 7, both GNSS and eNB signal may be or attached to eNB2. If the UE2 is out of the network coverage,
available within cellular coverage. Thus, in the cellular cover- UE2 uses the time reference of UE1 for side link establishment
age region, prioritization among these two systems should be and then UE1 informs eNB1. If UE2 is attached to eNB2, then
considered [32]. UE1 should request its serving eNB1 for the synchronization
In [34], the authors proposed a different synchronization procedure. If eNB1 approves the request of UE1, UE1 receives
solution so that the cellular links and inter-cellular side link one of the following synchronization instructions from eNB1.
can co-exist without any interference. In LTE V2X com- The first instruction is to connect to UE2 via side link after
munications, if side link and uplink transmission are not disconnecting from eNB1 and to utilize the time reference of
synchronized then inter-symbol and inter-carrier interference UE2. The second instruction is to be handed over to eNB2 and
may arise due to the relative timing offset of uplink and to use side link under the control of eNB2. The third instruc-
side link transmission. This synchronization issue can be tion is to utilize GNSS for side links and to request UE2 to
addressed using GNSS as an external synchronization source. use GNSS as well.
There should be a synchronization within out of coverage In [29], the authors discuss synchronization procedure for
mobile users equipped or not equipped with GNSS and cellu- 3GPP NR based side link communications. If eNB or next-
lar network. In the proposed scheme, synchronization signals generation Node B (gNB) is present then, vehicle, eNB or gNB
are exchanged among out of coverage users in the side link. all follows the same timing reference. However if there is a
Other side link users detect these synchronization signals and lack of coverage or if eNB or gNB is not deployed, the timing
TABLE IV
C OMPARISON OF S YNCHRONIZATION P RIORITY AND E RROR IN V2X C OMMUNICATIONS
reference of all vehicles should be aligned with each other. In [30], the authors discuss the multi-link synchronization
GNSS can be used for synchronization of out of coverage in cellular-based V2X communications. Multiple synchroniza-
vehicles. Thus, GNSS based synchronization and eNB/gNB tion sources are available in vehicular communications due
based synchronization can be used for NR based vehicular to which different vehicular UE may be driven by different
communications. synchronization source. In case of partial and out of cover-
a) eNB/gNB based synchronization: When the eNB or age scenarios, mutual user time synchronization can be used
gNB is present, vehicles receive synchronization signal block similar to [34]. In [30], the authors have used the concept
(SSB) from eNB or gNB. The vehicle onboard unit contin- similar to [34] and have done simulations to analyze mutual
ues to receive SSB and remains synchronized with eNB or and mean absolute time offset in out-of-coverage and partial
gNB. If eNB or gNB synchronization signals are not present, coverage scenario. With simulations, the authors have shown
vehicle UE synchronizes to SSB based on the priority. The that with some out-of-coverage users equipped with GNSS
first priority is given to vehicle UE directly synchronized to and some in-coverage cellular users, the relative time error
eNB or gNB, the second priority is given to UE indirectly for partial coverage is below 0.2 µs. Similarly, the authors
synchronized to eNB or gNB, the third priority is given to have shown that the absolute time error for partial coverage is
GNSS, fourth priority is given to UE directly synchronized to 0.3µs. Moreover, the authors have shown that absolute time
GNSS followed by UE indirectly synchronized to GNSS and offset error for out of coverage is high due to the time drift
last priority is given to remaining UEs. UE acts as a synchro- from radio wave propagation time. This shows that the scheme
nization source for other UEs whose synchronization signals similar to [34] works perfectly in a partial coverage scenario.
are derived from the source with low priority. If a vehicle UE Table IV compares the synchronization priority and error in
cannot receive any synchronization signal from any source V2X communications.
then UE uses its local source as the synchronization source.
b) GNSS based synchronization: In GNSS based syn- C. MBMS
chronization, vehicle UE searches for GNSS signal as soon Due to the broadcast nature of MBMS, it can be uti-
as it is powered on. UE becomes a synchronization source lized for safety-related applications. In addition, the signal
by transmitting SSB for other vehicle UE whose timing refer- strength of the message can be reinforced by utilizing a multi-
ence is derived from an eNB or gNB. Similarly, UE becomes cell broadcast of MBMS [27]. However, there exist several
a synchronization source for other vehicle UE whose timing issues with MBMS. This section highlights some of the key
reference is derived neither from eNB nor GNSS. If GNSS issues and solutions related to MBMS service for vehicular
synchronization signals are not present, vehicle UE synchro- communications.
nizes to SSB based on the priority. The first priority is given to 1) Challenges:
vehicle UE directly synchronized to GNSS, the second prior- a) Large broadcast areas: For a specific V2X message,
ity is given to UE indirectly synchronized to GNSS, the third the broadcast area may cover one or more cells. However, with
priority is given to eNB or gNB, fourth priority is given to UE the use of MBMS, information may be broadcast on more than
directly synchronized to eNB or gNB followed by UE indi- required cells. Thus, existing LTE based MBMS can decrease
rectly synchronized to eNB or gNB and last priority is given to the efficiency of traffic flow information and is not directly
remaining UEs. Similarly, UE acts as a synchronization source applicable in V2X communications.
for other UEs whose synchronization signals are derived from b) Small, variable and overlapped broadcast areas:
the source with low priority. If a vehicle UE cannot receive There are various discussions on 3GPP document [7] to sup-
any synchronization signal from any source then UE uses its port small and variable areas in V2X using MBMS. 3GPP has
local source as the synchronization source. identified two main issues to support small and variable areas
in V2X. The first issue is to determine the V2X broadcast area can deploy localized MBMS in the radio access network for
and the second issue is to broadcast different V2X messages V2X service whereas non-V2X MBMS service may still use
in different areas, especially in overlapping areas. MBMS-GW and BM-SC in the core network. The current
c) Latency: In the existing system, MME, BM-SC, standard does not support the UE to use the local BM-SC for
MBMS, and MBMS-GW are placed in the core network to V2X MBMS service, and macro BM-SC for non-V2X MBMS
reduce the backhaul delay between BM-SC and the eNB while services at the same time.
delivering V2X messages. Latency will be one of the important In addition, in [36], the authors evaluated the efficiency
factors while using MBMS for LTE V2X services. of MBMS and showed that performance can be significantly
2) Solutions: enhanced by using a MIMO system. Mainly, the authors
a) Large broadcast areas: To solve the issue of large proposed and analyzed the use of diversity-oriented schemes
broadcast areas, geo-casting can be used with the help of as well as multiple stream schemes to enhance the spectral effi-
a special purpose back-end server. This back-end server ciency of MBMS. The authors advocate the use of multiple
can intercept and process traffic before redistributing it to antennas both at the base station and the UE to reduce the
concerned vehicles in a given geographical area [35]. latency.
b) Small, variable and overlapped broadcast areas: For In [37], the authors proposed a scheme to reduce the end-
the first issue of determining the V2X broadcast area, a spe- to-end latency in V2X MBMS broadcasting by performing
cial purpose V2X server can be used which can broadcast the processing at the layer-2 of the base station and the
the information based on the V2X UE location. For the sec- use of geographic radio network temporary identifiers (Geo-
ond issue of broadcasting different V2X messages in different RNTI). In Geo-RNTI, different RNTI values are assigned
areas, 3GPP has identified two solutions based on a single for different geographical regions. These Geo-RNTIs values
temporary mobile group identity (TMGI) and multiple TMGI. are pre-configured per service and per area and avoid the
Single TMGI Based: There are several options based on need for communication with application servers. Geo-RNTIs
single TMGI: are managed by the mobility engine and session manager.
• New ID to Differentiate the Flows: In existing MBMS, Mobility engine updates Geo-RNTIs based on UE position
flow id doesn’t allow different data in overlapping areas and the session manager allocates TMGIs and stores a map-
but allow the network to provide different data in differ- ping information betweens associated Geo-RNTIs and TMGIs.
ent areas. To transfer different data in overlapping areas, a Moreover, the authors performed the simulations and have
new id named ‘x’ is invented. V2X server provides data shown that the proposed layer-2 based technique reduces the
for each ‘x’ id of the network and MBMS-GW trans- latency of the existing LTE MBMS system and satisfies the
fers the corresponding data of different ‘x’ id bearer to cooperative collision avoidance (CCA) latency requirements.
the different multicast IP address. In this scenario, eNB Table V lists the challenges, solutions, limitations, and feasi-
identifies multiple ‘x’ ids for the same TMGI. Moreover, bility of MBMS in V2X communications. Feasibility indicates
it combines the data of different ‘x’ ids and provides in the suitability for deployment which is decided based on the
the Uu interface utilizing the same TMGI. complexity of the proposed solution and changes in the exist-
• Use Non-Overlapped MBMS Service Areas: An operator ing architecture. The term very high in the feasibility column
can arrange small non-overlapped MBMS service areas means very high suitability for deployment followed by the
for V2X. term high, whereas the term low means the low suitability for
• User Plane Enhancement Solution: In this solution, the deployment.
MBMS bearer is pre-established by the V2X server with a
specific TMGI for a V2X service in large broadcast areas. D. Resource Allocations
Broadcast area information is added to the SYNC header 1) Challenges: Due to a large number of concurrent trans-
by BM-SC after decoding the V2X message. Generated mission in dense vehicular traffic scenario, it is essential to
MBMS packets are then sent by BM-SC to MBMS-GW efficiently utilize the available resources. In LTE D2D, for
and eNB through the pre-established MBMS bearer. The centralized scheme downlink control information is used for
eNB decodes the new SYNC header in the arrived MBMS resource management whereas, for the distributed scheme, the
packets, to identify the V2X broadcast area and then random resource selection method is used [32]. In LTE D2D,
determine whether to broadcast the MBMS packet or not. a number of concurrent transmissions are low compared to
This solution affects eNB, BM-SC and SYNC packet. vehicular traffic scenarios so LTE D2D resource allocation
Multiple TMGI Based: In this solution, to transmit different may not be applicable for V2X traffic. Further, due to highly
V2X messages in different overlapped areas, different TMGI mobile and dense vehicular users in LTE V2X communica-
can be applied to overlapped MBMS service areas. tions, D2D centralized resource allocation may result in extra
3) Latency: There are two options to reduce latency while signaling overhead as UE needs to frequently connect to eNB
using MBMS to deliver V2X messages. The first option is that whereas random resource allocation may result in resource
MBMS core network functions can be moved closer to eNB collision due to unilateral bad decisions of vehicular UE [32].
or even collocated in the eNB. The second option is to move 2) Solutions: 3GPP has proposed several schemes to
the user plane of MBMS core network functions close to eNB improve D2D resource allocation that can be used in the
or even collocated in the eNB. There are still some issues vehicular scenario. Improvements include a semi-persistent
with localized MBMS based implementation. An operator resource allocation scheme for centralized resource allocation
TABLE V
C HALLENGES AND S OLUTIONS OF MBMS IN V2X C OMMUNICATIONS
and techniques such as sensing or detecting other UE trans- can only be shared with a single device user equipment which
mission for distributed resource allocation. Enhancement of is one of the main limitations of this article as it leads to poor
MBMS for communications in high traffic demand is also spectrum utilization.
being considered [27]. In [39], the authors designed a scheme to maximize the
Resource allocation can be broadly divided into central- minimum capacity of cellular users while guaranteeing the
ized resource allocation and distributed/autonomous resource reliability of vehicular users. In addition to the reliability
allocation. In centralized resource allocation, eNB sends the requirement of vehicular users, i.e., block error rate, the
control signal for resource allocation whereas, in distributed authors have considered the latency of vehicular users. Due to
resource allocation, resource allocation is conducted indepen- the difficulty in the computation of the latency for vehicular
dently without any help from eNB. users, the authors transformed the latency constraint into data
a) Centralized resource allocations: In literature, there rate constraints. The authors have proposed an algorithm based
are several schemes proposed to efficiently utilize resources on Lagrange dual decomposition and binary search method
with the help of eNB. Most of the proposed schemes exploit for radio resource management. Similar to [38], the authors
the uplink band as of cellular users for V2X traffic. For the have considered that the resource of one cellular user can
reader convenience, we have categorized centralized resource only be shared with a single vehicular user which leads to
allocations based on the application domain and technique poor spectrum utilization.
employed. Each of the categories is explained in detail below. In [40], the authors proposed a method for joint channel and
General Schemes: This category includes the schemes that power allocation in which channel state information of vehic-
employ a general system model for resource management in ular links are reported to the base station on a periodic basis.
vehicular communications. The main goal of this scheme is to maximize the capacity of
In [38], the authors proposed a resource allocation scheme the V2I link. In this scheme, at first, the power allocation is
for vehicular communications utilizing slow and large scale performed considering only single V2I and V2V pairs. Based
fading information of wireless channels. In the proposed on the optimum power allocation for single V2I links and V2V
scheme, a high information rate is desired for V2I links pairs, optimum V2I capacity is obtained. The authors have
whereas high reliability is desired for V2V links. In this arti- used the Hungarian algorithm to solve the proposed scheme.
cle, the authors have proposed two algorithms. In the first As in this scheme, multiple vehicular users can share the
algorithm, the authors formulated the objective function to same V2I link, it leads to better spectrum utilization compared
maximize the sum information rate of all V2I links. Whereas, to [38] and [39].
in the second algorithm, the authors formulated the objective In [41], the authors transformed the reliability and latency
function to maximize the minimum information rate of V2I requirements of vehicular networks into optimization con-
links so that the uniform performance can be achieved across straints. In this scheme, the authors proposed a SOLEN
all V2I links. The authors have only considered slow and algorithm. In the first stage of the SOLEN algorithm, the chan-
large scale fading information instead of instantaneous channel nel allocation problem is considered as the maximum weight
state information for resource management. The authors have matching problem for a bipartite graph to allocate channels
considered that the resource of one cellular user equipment to both V2V and cellular user equipment. Second, based on
channel allocation results from the first stage, power control expected reward so that more video requests of the vehicles
is performed for each vehicular and cellular user equipment. can be admitted while assuring QoS for cellular users.
Similar to [38] and [39], the proposed scheme leads to the In [44], the authors have considered resource allocations
poor spectrum utilization as a spectrum of one cellular user in heterogeneous vehicular networks with cognitive radio
equipment can be shared with the only single device user technology. In the proposed scheme, the available spectrum
equipment. is identified by cognitive radio enabled RSUs by perform-
In [42], the authors proposed graph theory based resource ing spectrum sensing in surrounding environments. Incentive
allocations for the connectivity optimization problem. In the mechanism is used to encourage base stations for sharing
proposed scheme, multiple vehicular user equipment shares a the available spectrum resource with RSUs. In addition, the
single cellular user resource leading to network interference. authors have formulated the problem of resource allocation as
In this scheme, the authors utilized the interference graph to the n-person game with regret matching based strategy selec-
obtain the number of resources that guarantee the connectivity tion. Moreover, the authors derived an equilibrium solution for
of the network. The minimum spanning tree method is used the proposed non-cooperative game and used graph coloring
to model and solve the proposed resource allocation scheme. to form the strategy set.
Moreover, the authors have performed simulation and shown Clustering or Group Based Schemes: This category dis-
that the proposed algorithm improves network connectivity. cusses the schemes that use clusters or groups of V2V links
Heterogeneous and Cognitive Radio Based Schemes: This for resource allocations.
category discusses the resource allocation schemes for vehic- In [45], the authors proposed a dynamic resource allocation
ular communications based on heterogeneous and cognitive algorithm for vehicular scenarios based on traffic flow. In the
radio based systems. proposed scheme, V2V nodes and cellular UE share the same
In [20], the authors proposed a cognitive radio based resource block in one LTE cell. A physical resource block is
resource allocation policy. This article assumes that resources shared by multiple V2V links without lowering the communi-
are centrally managed by eNB with the cellular user as a pri- cation quality of cellular users. In addition, in this article, the
mary user and V2X network as a secondary user. Moreover, authors have analyzed the connection between capacity utility
this article assumes that the maximum allowable transmit value and the number of V2V grouping. They showed that the
power for vehicles is computed by eNB. In the proposed number of V2V grouping should increase with the increase
scheme, the V2V network utilizes the most appropriate LTE- in the number of V2V links. Further, based on vehicle speed,
A uplink band for V2X traffic. When vehicular UE utilizes appropriate traffic flow is identified and then the number of
the same frequency band as of cellular users, it should not V2V links for that traffic flow is selected. Once the V2V links
interfere with the cellular user while maintaining the required are selected, suitable resources are assigned to each group of
power for QoS. Due to the correlation of average path loss V2V communication links. The authors further presented the
between uplink and downlink signals, the proposed scheme simulation results showing better performance than greedy and
senses the path loss in the downlink signal. In this article, random resource allocation algorithms in terms of runtime and
sensing algorithm is proposed which has two stages, cognitive spectrum efficiency.
sensing stage, and resource allocation stage. In cognitive sens- In [46], the authors transformed the reliability and latency
ing stage, interfering and non-interfering list is obtained by requirements of V2V communications into outage probability.
comparing the V2V nodes transmitting power with the non- In the proposed scheme, multiple vehicular user equipment
interfering rule. In resource allocation stage, eNB utilizes a is allowed to share the common resource block. In addition,
non-interfering V2V list to allocate resources to V2V nodes the authors proposed a three-stage radio resource management
and cellular users. In addition, eNB allocates unutilized white algorithm called CROWN. In the first stage, the vehicular
space in the uplink band to V2V nodes in the interfering list. user equipment is grouped in the cluster and in the second
In [43], the authors considered resource allocations to facil- stage resource block sharing strategy is performed. Vehicles
itate video streaming in cellular and cognitive radio enabled belonging to different clusters can share the same resource
vehicular communications. The main goal of the proposed block whereas vehicles in the same cluster cannot share the
scheme is to improve the video quality of vehicular users same resource block. After the resource block assignment, the
without affecting cellular users. The authors have used a semi- feasibility of the resource assignment is studied. If resource
Markov decision process (SMDP) for resource allocations. In management is feasible then power allocation is performed in
this scheme, the state space is defined on the basis of the the third stage.
number of admitted background cellular users and the spa- In [47], the authors presented a graph theory based
tial distribution of admitted vehicle users. The action of RSU resource allocations. In the proposed scheme, multiple V2V
or base station is to accept or reject the vehicle video request links can share the same V2I link. The authors have used
based on the available spectrum. After adopting the action, the a max n-cut algorithm to partition interfering V2V links
new state is defined by state transition probability. Moreover, into different clusters. The authors only considered large
the reward between two decision epochs is described as the scale fading information to partition V2V links and form
product of the number of admitted vehicle users and the a cluster. After forming the cluster, the resource allocation
expected time duration between two epochs. In the proposed problem is designed as a weighted three-dimensional match-
scheme, an optimization problem is designed to maximize the ing problem with vertices as V2I links, resource block and
V2V clusters. To solve a weighted three-dimensional match- Nash equilibrium for the proposed game by employing
ing problem, the authors have adapted the high-performance the Karush-Kuhn-Tucker (KKT) condition into non-linear
approximation algorithm. In addition, the proposed scheme optimization. In addition, the authors have exploited the
results in a higher spectrum utilization as a single cellular software-defined network to address the complex and het-
user equipment resource is shared by multiple device user erogeneous vehicular network. Moreover, they achieve global
equipment. optimization result with the help of the software-defined
In [48], the authors used dynamic programming techniques network.
and greedy techniques for resource allocations. The dynamic NOMA Based Schemes: NOMA based schemes include
programming technique solves the resource allocations using schemes that employ non-orthogonal multiple access tech-
optimal clustering whereas the greedy technique solves the niques for resource management.
resource allocations with low computing complexity. In the Non-orthogonal multiple access (NOMA) technique can be
proposed method, vehicles in a cluster can reuse the same used in LTE based vehicular network to reduce the resource
uplink frequency resource to communicate directly via a D2D collisions between different V2X applications and to support
mode. SINR criteria are used to compute the maximum trans- massive connectivity by using code domain or power domain
mission range for vehicles while sharing the resources of multiplexing. In [51], the authors proposed resource alloca-
cellular users. The maximum transmission range is determined tion for NOMA based vehicular system, in which channel
by the traffic density and distribution of cellular users. In allocation is performed by the base station and power con-
addition, different transmission directions may induce different trol is performed by V2V users. At the beginning of each
D2D transmission distances. In each cluster, vehicles share the semi-persistent scheduling (SPS) period, the base station allo-
same resource so the objective of the resource allocation algo- cates the channel on the basis of position information of each
rithm is to minimize the number of clusters. In the proposed vehicle. After channel allocations, dynamic power control is
algorithm, if a current vehicle uses resources different from performed by the transmitter or V2V users in each SPS period.
that of a former vehicle, then the number of resources is added Due to the use of the NOMA technique, a lot of interference
by one and traditional cellular resources can be used by vehi- is produced at the receiver side. Thus, for distributed power
cle only when formerly allocated resources are not used by control and for successive interference cancellation at the
the vehicle. Moreover, the authors have shown with simula- receiver side, information is exchanged between transmitter
tion results that when vehicle density increases, a few numbers and receiver users.
of clusters are needed in D2D mode and the unidirectional Similar to [51], the authors in [52] designed a scheme
transmission has fewer clusters as compared to bidirectional in which the base station performs channel allocations in a
transmissions. semi-persistent manner and vehicles perform distributed power
Cloud Computing Based Schemes: This category includes control. In the proposed scheme, the transmitter-receiver is
the schemes that consider resource management in a central- selected by base-station in a centralized manner and packet
ized vehicular cloud platform. reception probability maximization problem is considered
In [49], the authors designed a scheme to maximize the rev- for resource allocations. The authors considered vehicular
enue of the vehicular cloud computing (VCC) system. In the users and channels as two disjoint sets of objects to be
proposed scheme, moving vehicles constitute a dynamic vehic- matched together. For this matching problem, the authors
ular cloud. Each vehicle has one basic computation resource have employed a multi-dimensional stable roommate matching
unit and can save the processing speed and energy by send- technique. In addition, the authors proposed their own rotation
ing the task request to VCC. Whenever a request arrives, the matching algorithm which after a limited number of iterations
vehicular cloud either accepts it or transfers it to the remote converges to stable matching.
cloud. In this scheme, the system state is defined by several In [53], the authors proposed NOMA based resource allo-
parameters like vehicles, the number of resource units, the cations to maximize the minimum capacity of the cellulars.
unoccupied resource in the vehicular cloud and the event of In addition, they imposed minimum SINR for V2V users as
requests. Whereas, the action reflects whether to accept the the constraints while formulating the problem. The authors
request or to transfer it to the remote cloud and reward is the solve the proposed resource allocation problem by decoupling
revenue of the system which is calculated based on the income the optimization problem into three stages. In the first stage,
and the cost of the VCC system. Moreover, the authors have power is allocated for cellular user equipment whereas in the
performed a simulation to compare the proposed scheme with second stage power allocation and sub-carrier assignment is
other schemes and found a significant reward gain compared performed for the vehicular user equipment. In the final stage,
to other schemes. the cellular user equipment is clustered and the subcarrier is
In [50], the authors have considered enhanced cloud radio assigned for the cellular user equipment. Moreover, the authors
access network (C-RAN), D2D communications and SDN have presented an algorithm that integrates these three stages
technologies to provide large bandwidth and high comput- to obtain a joint solution. The authors have shown with sim-
ing capability for vehicular communications. The authors ulation results that the cellular user equipment with diverse
have proposed a matrix game approach for resource allo- channel gains occupies the same cluster. In addition, they
cation and management of cloudlets. The authors formu- have shown that minimum cellular capacity decreases with
lated resource management to enhance the resource uti- the increase in the minimum SINR threshold and the number
lization of each cloudlet. The authors have obtained a of V2V links.
TABLE VI
C OMPARISON OF C ENTRALIZED R ESOURCE A LLOCATIONS FOR V2X C OMMUNICATIONS
3GPP Based Solutions: This category discusses ongoing and solutions to reduce MBSFN/SC-PTM latency are being
discussions and projects in 3GPP for centralized resource considered. Supporting inter-operator deployment is also being
allocations. discussed so that UE is allowed to receive the downlink broad-
3GPP document [7] discusses the downlink and uplink cast of another operator. Similarly, 3GPP is also considering
enhancement of the Uu interface for centralized resource uplink enhancement for LTE V2X communications. For uplink
allocation. For downlink enhancement, 3GPP is currently con- enhancement, semi-persistent scheduling (SPS) can be used to
sidering the use of both multicast-broadcast single frequency reduce the uplink overhead. For the SPS method, it is desir-
network (MBSFN) and single-cell point to multipoint (SC- able to use a long SPS period between 100 ms and 1 s. In
PTM). Improving MBSFN/SC-PTM on the basis of UE geo- addition, UE assistance on periodicity and timing can be pro-
graphical location is being considered so that the information vided to eNB and also UE can inform the network when SPS
can be broadcast only on the target area. Moreover, the need resources are not used.
If vehicle densities are different for vehicles moving in

opposite directions, service quality can be relaxed in the direc-
tion of high vehicle density and high service quality can be
applied in the direction of low vehicle density. Thus, different
resource pools can be assigned to vehicles moving in opposite
directions to provide different QoS.
In addition, the 3GPP document [56] explains resource
allocation based on position and heading. By assigning a
set of resources to a geographical zone, the effect of hid-
den node problem and the near-far effect caused by in-band
emissions can be reduced. Two approaches used for defining
geographical zones are:
• Position Based Geographical Partitioning: The
performance of this scheme relies on having actual
location information of each UE and an appropriate
dividing of the geographical region into a set of
Fig. 9. Resource pool partition in intersection area, described by [55].
zones/grids. But this scheme results in a large amount of
signaling between eNBs and UE as a new set of zones
has to be formed each time a UE leaves the previous
Table VI compares the above discussed centralized resource zone. Position based zoning mechanism can be more
allocations on the basis of spectrum/channel sharing, objec- useful in rural or highway scenarios however a great
tive, the approach employed, performance metric, spectrum amount of tasks has to be conducted to optimize zones
utilization, and performance comparison/evaluation. for complex urban areas.
b) Distributed resource allocations: In distributed • Heading Based Geographical Partitioning: In most of
resource allocation, vehicular UE allocates the resource inde- the urban scenarios, only two zones, each correspond-
pendently without any help from eNB. In D2D mode 2 ing to one direction, i.e., East-West and North-South are
resource allocation, UE arbitrarily selects the resource for data required. Significant performance gain can be achieved
transmission but for V2X, a random resource selection scheme by using heading based resource allocation as compared
is not sufficient. Distributed resource allocation algorithm to random resource allocation. In addition, heading based
has to deal with three types of effects: interference between partitioning can greatly simplify the signaling and provi-
UE due to the utilization of the same resource, interference sioning in complex urban areas and can be more bene-
due to in-band emissions (IBE) by other UE transmitting ficial compared to the position based partitioning. Thus,
in the neighboring band and packet loss due to half-duplex the combination of both position and heading based geo-
constraints [54]. The best way to increase the efficiency of graphic partitioning can be used where position based
distributed resource utilization is to make sure that nearby partitioning can be used for rural areas and heading based
transmitters are co-scheduled for transmission at the same partitioning can be used for urban areas.
time resources. In addition, there are various methods to effi- Sensing Based Distributed Resource Allocations: 3GPP
ciently allocate the resources in a distributed scenario which document [54] discusses sensing based resource allocations.
is described with various categories of distributed resource Sensing based resource allocations can reduce the possibility
allocations in the following paragraphs. of resource collision. UE can identify the occupied resource
Location Based Distributed Resource Allocations: 3GPP by energy-sensing or reading other UE scheduling assignment
document [55] discusses resource allocation based on geo- (SA) broadcasting. In an energy-sensing method, a UE senses
graphic partitioning. Geographic partitioning is crucial for the received signal on the resource block to detect whether a
distributed resource allocations as various parameters of V2V resource block is occupied or not. In this method, a threshold
communications such as transmitter power, carrier sensing value can be used to determine whether the resource block is
threshold are location dependent. In V2V communications, occupied or not. Energy sensing helps to determine how much
different interference levels may arise across different direc- severe interference is on each resource block but this scheme
tions and lanes if the vehicle speed is different across different cannot provide information on UE resource occupation period
directions and lanes. If vehicles approaching the intersection and duration. Whereas, in reading other UE SA/data broad-
in the perpendicular road section are using the same resource casting method, resource block usage of UE is broadcast by
pool, then a significant reception power difference will be the scheduling assignment or together with data broadcasting.
observed between vehicles. Due to the large difference in UE can obtain additional information like period, frequency
reception power, the weak signals coming from different group domain resource usage but UE may not have the information
UE can suffer from in-band emission interference of the strong on the interference level on each occupied resource. Therefore,
signals coming from the same group UE. To address this it is beneficial to use a combination of both energy sensing
issue, time division multiplexed resource pool partitioning and reading other UE SA/data broadcasting method.
between different UE groups approaching the intersection can In addition, the 3GPP document [57] discusses several con-
be considered, as shown in Fig. 9. ditions that should be met in order for collision avoidance or
sensing-based scheme to work efficiently. The first condition multiplexing manner to prevent IBE interference. After group-
is that the channel conditions and interference environment ing, resource prioritization is conducted in which UE selects
should vary slowly so that the sensing method can precisely resources from the subset of free resources. If there is not an
predict the propagation environment during actual transmis- adequate resource, then UE selects a resource from decoded
sions. The second condition is that the sensing based collision side-link control information and if that is also not suffi-
scheme should be preserved with low to medium system load- cient then UE randomly selects the resource from the resource
ing to properly handle the collision. The final condition is that pool.
the target communication range should not be larger than the c) Hybrid centralized-distributed resource allocations:
sensing range. In addition, a sensing based collision avoid- Hybrid schemes address resource allocations in both central-
ance scheme should address different challenges like mobility, ized and distributed scenarios. In [60], the authors presented a
hidden node, in-band emission and half-duplex constraint and hybrid resource allocation and considered ad-hoc and assisted
latency. D2D mode. Outside the network coverage, devices can operate
Enhanced Resource Randomization Based Distributed only in ad-hoc mode whereas inside the network cover-
Resource Allocations: 3GPP document [55] discusses age devices can operate in both ad-hoc and assisted mode.
enhanced resource randomization technique. In Release 12, The authors proposed a unified radio frame structure which
time and frequency resources are randomly selected by UE includes sub-frames for supporting both ad-hoc and assisted
which is not applicable for V2V communications. In addition, D2D mode. In unified radio frame, control and data chan-
for mode 2, there are 108 time resource patterns (T-RPTs) nels of ad-hoc and assisted D2D mode are separated in the
in frequency division duplexing but these resources are not time domain. In the assisted control channel, only an assisted
enough for PC5 based V2V as there is a large number of UE device can transmit and in the ad-hoc control channel, only
and packet size is also large in case of V2V. It is possible an ad-hoc device can transmit. However, both ad-hoc and
to increase the number of T-RPTS but this will cause signal- assisted D2D mode device can transmit in the same data chan-
ing overhead for each UE and also new T-RPT design needs nel resource. Moreover, the authors proposed a unified D2D
to be created. Thus, a more appealing solution is the use of MAC protocol in which location-aware reservation Aloha is
randomized T-RPT. used. The authors have performed simulations and have shown
Moreover, 3GPP document [58] discusses randomization that the proposed protocol can achieve high reliability and low
technique. This document explains how the data pool structure latency, and can be applied for both in and out of coverage
of Release 12 D2D can be reused with some modifications like scenarios.
SA pool frequency division multiplexed with the data pool. For In [61], the authors have proposed hybrid centralized-
enhanced randomization, T-RPT hoping can also be used. If distributed resource allocations in which channel is allocated
a T-RPT index is randomly hoped then more patterns can be by the base station in a centralized manner whereas power is
created which provide enhanced randomization. allocated by vehicular UE in a distributed manner. The authors
Hybrid Distributed Resource Allocations: There is a dis- have considered a bipartite graph for channel allocation and
cussion going on in 3GPP whether operating a combination multi-agent learning for power control. Channel assignment is
of these principals provide more gain then operating individ- performed in two stages, channel assignment for the cellular
ual principle. For example, a combination of location based user and channel assignment for vehicular users. For channel
scheme and sensing based scheme can be used in which sub- assignment of cellular users, the bipartite graph is designed
sets of resources are associated with sets of UE location and with two groups as a set of cellular users and a set of channels.
the UE can perform sensing based schemes on these sub- Hungarian algorithm is used to solve the designed weighted
sets associated with UE’s current location. Inspired by these bipartite graph. For channel assignment for vehicular users, the
discussions, the authors in [59] proposed a scheme in which estimated network graph is constructed. For power control, a
resource partitioning is studied based on the heading direction multi-agent game with incomplete information is considered.
of vehicle UE and sensing based collision avoidance mecha- The authors have used Q-learning to obtain the power selection
nism. In this scheme, a set of resources is reserved only for strategy in a distributed manner.
V2V communications. In the resource pool, one subset is for Hybrid schemes discussed above are more applicable in
transmission of side-link control information whereas another the vehicular scenario than the individual centralized and
subset is for associated data. In urban areas due to the pres- distributed resource allocations. In centralized resource allo-
ence of the building and other large structures, sensing based cations, the base station collects CSI of all links in a system.
approach will not perform perfectly. To address this issue, However, in vehicular communications, it is very difficult to
the authors proposed a two-stage resource allocation scheme. accurately measure the CSI of all links in a system due to
In the first stage, time-orthogonal resources are allocated to the high mobility. For accurate CSI estimation, channel states
UE moving along orthogonal directions in intersections and can be frequently fed back from the vehicle to the base sta-
in the second stage in-band emission aware sensing approach tion. However, such schemes incur a lot of signaling overhead.
is used. In sensing based stage, interfering vehicular UE is Whereas distributed resource allocations are more applicable
identified by decoding side-link control information. After in out of network coverage scenarios and are not efficient
detecting interfering vehicular UE, they are classified as a for in coverage scenarios. Thus, an elegant way of resource
nearby and distant vehicular UE. Nearby interfering vehicular allocations is Hybrid schemes which take advantage of both
UE is arranged into the same subframe in a frequency division centralized and distributed resource allocations.
E. Security
Security is one of the main challenges in LTE based
V2X communications due to the broadcast nature of wire-
less communications. The traffic-related and safety-critical
information transferred in the V2X network may be repro-
duced, falsify or modified by an attacker causing a disas-
trous effect. Moreover, an attacker may mimic an autho-
rized entity such as RSU or eNB and may distribute wrong
information to nearby vehicles causing a traffic jam and acci-
dents [62]. Thus, LTE V2X communications must satisfy the
fundamental security requirements and requirements specific
to LTE V2X in order to provide defense against various
attacks.
1) LTE V2X Security Requirements: The basic security ser-
vice requirements of LTE V2X are discussed in this section, Fig. 10. PC5 and LTE-Uu based security architecture for V2X, described
by [64].
as outlined in [16], [63] and [64].
a) Authentication: Authentication in LTE V2X helps to
differentiate the legitimate entities with the malicious ones. VCF is mainly responsible for authentication and autho-
Authentication in LTE V2X includes both user authentica- rization of vehicular UE (V-UE). Temporary ID Management
tion and message authentication. User authentication helps to Function (TIMF) is mainly responsible for distributing tem-
guarantees the authenticity of the vehicles whereas message porary IDs and credentials to V-UE and V2X data source
authentication guarantees the authenticity of the messages. By accountability. V2X service operator may either choose V2X
using both user and message authentications, each vehicle can KMS or V2X CA and the interface between TIMF and V2X
confirm whether the message was sent by authorized vehicles CA/V2X KMS is referred to as V7. V8 is a reference point
or not and also find out whether the message was modified between TIMF and VCF and this reference point is used by
unexpectedly or not during the transit. TIMF to forward the authentication request to VCF.
b) Authorization: In LTE V2X, different entities such as Some of the key security challenges and their potential solu-
vehicles, RSUs, eNB and certificate authority can have differ- tions in LTE V2X communications, mainly mentioned in the
ent levels of access. There should be a provision of rules and 3GPP document [64], [65], are described below.
policies so that each legitimate entity gets the right level of 3) Challenges:
access without getting access to unauthorized content. a) Authentication: In V2X communications, an attacker
c) Data confidentiality: In LTE V2X, data confidential- may modify, reply, or forge the broadcast messages caus-
ity is not that of concern for basic public safety messages as ing traffic jams and road accidents. Thus, authentication is
they are not encrypted. However, there exist several other V2X essential to identify the authenticity of the sender’s vehi-
messages such as private vehicle-to-vehicle messages where cle. LTE V2X communications can adopt the existing LTE
data confidentiality is required. In addition, the privacy of a security mechanism for authentication. However, there exist
vehicular user is of utmost importance and is one of the main security vulnerabilities related to pre-authentication, authen-
security concern in LTE V2X. tication, and re-authentication of LTE which are described
d) Availability: Availability in LTE V2X ensures that below.
the vehicular messages are processed and made available in • Pre-Authentication: In LTE, user equipment arbitrarily
a timely manner to all V2X entities. Enforcing a heavy and accesses the cell during the initial cell selection pro-
strong cryptographic algorithm may delay the availability of cess. This arbitrary access may lead to a control channel
the vehicular messages. Thus, lightweight and low-cost cryp- spoofing attack. The control channel spoofing attack can
tographic algorithms are required so that vehicular messages restrict the vehicles to associate with authorized eNB and
can be processed in a timely manner and are available to most receive V2X services [16]. In the cell selection process,
of the V2X entities. UE selects the strongest signal cell and obtains timing
e) Data integrity: Data integrity ensures that the vehic- and frequency synchronization through the primary syn-
ular messages are not modified or deleted by malicious users chronization signal (PSS) and secondary synchronization
during the transit. A digital signature can be utilized to pro- signal (SSS) [66]. However, attackers may setup fake
vide the integrity of the messages. In addition, the freshness eNB and transmit signals at higher power levels. In addi-
of the messages can be used to ensure that messages are not tion, attackers may spoof PSS and SSS and perform
been replayed by malicious users. PHY layer signaling attack. When UE receives spoofed
2) LTE V2X Security Architecture: 3GPP document [64] PSS and SSS, UE reports to the radio resource control
has described the LTE V2X security architecture, as shown in (RRC) layer and then RRC will expect to receive a master
Fig. 10. Main functional entities of LTE V2X security archi- information block (MIB) message. As the fake cell can-
tecture include V2X control function (VCF), Temporary ID not transmit MIB, the RRC will be in idle mode. When
Management Functions (TIMF) and V2X Key Management RRC is in idle mode, UE treats this cell as barred and
Server (KMS) or V2X Certificate Authority (CA). performs barring.
• Authentication: Lightweight authentication schemes are authors in [66] have proposed modifications to the LTE cell
desired for LTE V2X so that the messages can selection process. In this modification, UE should be allowed
be processed in a timely manner. Public key based to select the second strongest cell within the same frequency
authentication scheme has heavy computation whereas if the strongest cell is considered as barred. In this modifica-
symmetric-based solutions don’t provide non-repudiation. tion, there will not be any uplink interference as no UE will
• Re-Authentication: High speed of vehicles causes fre- be attached to the strongest cell considered as barred. This
quent handover between MMEs or eNBs [16]. In han- modification is effective in some kind of channel spoofing
dover between two eNBs or MMEs, the current key attack.
is used to extract a new key between UE as well as Lightweight authentication is desired for LTE V2X com-
the new key for source eNB and target eNB. This munications. Identity-based authentication can be used which
key-management process can be exploited to perform a can lower the communication overhead as they do not use
de-synchronization attack which halts the forward key any certificates. But identity based encryption schemes are
separation employing the fake eNB [68]. public key based and still slower compared to the symmet-
b) V2X one-to-many communications security: Existing ric key. A hybrid authentication scheme can be used that can
LTE security system provides credentials only for point- lower the computation overhead and provide secure authentica-
to-point communications and is not suitable for LTE V2X tion in the meantime. However, existing hybrid authentication
broadcast communications. D2D proximity services (ProSe) schemes are complex and are not directly applicable to V2X
security for group communications is also not suitable as a communications.
group of vehicle UE cannot easily be defined. Moreover, in Root key can be periodically updated during re-
ProSe group communications, as any member can deduce any authentication to avert from the de-synchronization attack.
other member’s traffic key, it does not provide any traceability During periodic root key updates, additional signaling over-
of UE. head may occur due to additional key updates. In [67], the
c) Vehicular UE privacy: Privacy is one of the most authors proposed a scheme to jointly minimize two func-
important requirements in the LTE V2X system. Privacy will tions: the expected volume of exposed packets after the
ensure the high participation rate of the vehicular UE. There de-synchronization attack and the expected signaling overhead
are many security threats concerning the privacy of vehic- due to key updates. The authors considered objective functions
ular UE. In a PC5 mode, some identifiable information is as a linear weighted sum of two functions in which network
included in the periodic message which may reveal the private operators can select the suitable value of weight to balance the
information of the user. Whenever the vehicular UE attaches tradeoff of two functions regarding computational complexity.
to the network, a mobile network operator (MNO) has the Moreover, the authors have done simulations results and have
ability to track the UE. In addition, whenever V2X data is shown that their simulation results match with their analytical
sent across the network in the Uu interface, the V2X appli- results.
cation can track the UE by linking together V2X data such b) V2X one-to-many communications security: Security
as IP address. Hence in both PC5 and Uu mode, there is a requirements applicable to V2X communications for one-to-
threat that identity will be revealed based on data transmitted many communications can be satisfied by employing applica-
by vehicular UE. tion layer security from other software-defined organizations
d) Authorization and accountability: There exist differ- (SDO) such as IEEE 1609.2 standard, which is an open stan-
ent levels of access for different entities such as a vehicle, dard for IEEE 802.11p based communications. In addition, the
RSU, certificate authority in V2X communications. Rules and security guidelines of DSRC based scheme [69] can be ana-
policies should be clearly defined so that each legitimate entity lyzed and applied in LTE V2X one-to-many communications
gets the right level of access without getting access to unau- security.
thorized content. In addition, LTE V2X messages should be c) Vehicular UE privacy: There are several security
processed and made available in a timely manner to all V2X requirements currently being discussed in 3GPP such as vehic-
entities. Heavy and strong cryptographic algorithms will intro- ular UE pseudonymity, untraceability of a user by a single
duce delay and affect accountability due to the dynamic nature entity. Some of the proposed solutions are described below.
of vehicular communications. Moreover, if many UEs try to • Attach Identifier Obfuscation: In order to prevent MNO
request radio resources at the same time, it would cause the from tracking the vehicular UE, UE identities used for
exhaustion of radio resources. V2X communications can be managed separately by a
e) UE to V2X control function interface: An attacker can third-party server such as original equipment manufac-
pretend to be a V2X control function and may maliciously turer (OEM). In the proposed scheme in [64], [70],
manipulate or modify the data. As there is no shared chan- vehicular UE obtains temporary key Kperiod by request-
nel between V2X control function and V-UE, V-UE will not ing to MNO. MNO will send a mega-pool of pseudonym
be able to differentiate legitimate and malicious V2X control mobile subscriber identity (PMSI) pairs, each encrypted
functions. with a key Kperiod , to pseudonym CA or OEM. Vehicular
4) 3GPP Based Solutions: UE establishes a secure link with OEM and requests
a) Authentication: Various schemes have been proposed PMSI pairs by sending its international mobile sub-
to tackle the authentication issues in cellular V2X communi- scriber identity (IMSI). OEM sends sub pool of (PMSI,
cations. To prevent the control channel spoofing attack, the KPMSI ) pairs to vehicular UE. Vehicular UE decrypts
each received (PMSI, KPMSI ) pairs with Kperiod . After • V2X UE Tracking Based on PC5 Autonomous Mode: In
vehicular UE has several (PMSI, KPMSI ) pairs, it can use the region where strong privacy rules apply, it is proposed
them to attach with MNO. MNO will know that PMSI to operate V2V/V2I communications solely over PC5
was authorized but will not be able to recognize the IMSI Interface in autonomous mode.
of UE. Both OEM and MNO should collude with each • Homomorphic Encryption: In the proposed scheme,
other in order to track the user equipment. MNO generates the PMSI pair pool and sends it to the
• Privacy From MNO Based on Attached Data: LTE pseudonym certificate authority (PCA). V-UE sends an
network should be prevented from tracking the UE by identity request to PCA and asks for PMSI pair. PCA
using data of UE attached to it. There are several solutions chooses encrypted PMSI pairs from the pool and sends
being discussed to prevent this tracking as simultane- an identity response message together with this pool.
ous re-attach with new identities. All UEs under an eNB V-UE chooses blinding parameters and blind encrypted
should detach completely and then re-attach simultane- PMSI pairs. V-UE sends a decryption request to MNO
ously at the same time. To spread the load, all UE can for decryption. MNO decrypts the ciphertext and gets the
attach simultaneously at each eNBs but at different times blinded PMSI pairs and MNO sends it to V-UE. V-UE
from neighboring eNB. For this, UE needs to know the recovers the plaintext using the blind parameters.
re-attach period and next re-attach boundary time which d) Authorization and accountability: For authorization
can be obtained from eNB or can be obtained as a system and accountability, MNO can check authorization information
parameter during authorization time. Detach and re-attach of the vehicular UE before allocating radio resources. In the
is triggered in UE whenever current time value matches proposed scheme, V-UE establishes a secure channel to the
a boundary time instance for that eNB and UE reat- relevant trusted traffic authority (TTA) and runs an authoriza-
tach timer has a large enough value less than reattach tion or enrollment procedure and then V-UE obtains evidence
period or current UE reattach timer has exceeded a certain from TTA upon successful authorization. During V-UE autho-
maximum value. rization with the V2X control function, the evidence from TTA
• Privacy Based on Data Traversing the Network: In order is presented. In order to transmit revocation information from
to prevent MNO from tracking V-UE, it is proposed that the TTA domain to the MNO domain, some form of additional
when UE changes its application layer identifier, it also communications would be required between the TTA domain
re-attaches to the network to refresh all the lower layer and the V2X control function. In addition, TTA authorization
identities that will be visible to the V2X application evidence could be required to be unique per V-UE. The evi-
server. dence binds the V-UE to a specific long-term certificate in the
• Encrypted International Mobile Subscriber Identity TTA domain and specific subscription in the MNO domain and
(IMSI): Encrypted IMSI is being considered for LTE V2X can be used to identify malicious V-UE and thus confirming
communications. The proposed solution for LTE V2X accountability.
communications mainly addresses three changes from the e) UE to V2X control function interface: To solve the
existing LTE attachment procedure. The first change is problem, the V2X control function (VCF) and V-UE can use
to encrypt IMSI used at attaching, in which IMSI is a shared key for mutual authentication and establish a secure
encrypted with a key that is itself encrypted with the channel based on the pre-shared key. The pre-shared key could
public key of HSS. The second change is to encrypt an be derived using the AKA procedure when V-UE attaches to
authentication vector (AV) in which UE sends an encryp- the network through the Uu interface. Here, a new branch
tion key along with encrypted IMSI to HSS in the attach of key derivation is that V-UE and HSS derive a root key
procedure and HSS will encrypt AV using this key. The KV2X and then V-UE and VCF derive a pre-shared key KPSK
final change is to provide an IMSI from the HSS to use from KV2X . V-UE and VCF could utilize the KPSK and PSK
during the attach procedure. This scheme can provide Transport Layer Security (TLS) protocol to perform mutual
some sort of privacy of UE from MNO. authentication and establish a secure channel.
• Hiding UE Identity From Other V2X UE and the Serving Table VII presents challenges, solutions, mechanisms, and
Network: A pseudonym can be used to hide the UE iden- limitations of security in V2X communication networks.
tity from all network elements except HSS. Pseudonym 5) Non-3GPP Based Solutions: In addition to 3GPP
called P-IMSI can be generated following the standard projects in security, there are a few other papers address the
IMSI format with a mobile station identification num- security issues related to cellular V2X communications and
ber (MSIN) part replaced by a mobile station pseudonym they are discussed below.
(MSPN). Since mobile country code (MCC) and mobile The authors in [71] proposed a privacy-preserving LTE V2X
network code (MNC) are the same in both P-IMSI and security scheme. In the proposed scheme, transport author-
IMSI, P-IMSI is processed by all the network nodes ity (TA) is considered fully trusted and is responsible for
like IMSI. So there is no impact on the legacy LTE the management of V2X security services. In addition, vehi-
network. On the user side, V2X capable USIM should be cle pseudonym distributors (VPD) are considered as another
installed for generating the MSPN part for the pseudonym trusted entity that is responsible to distributes pseudonyms to
and on the network side, HSS needs to be prepared for vehicles. In the proposed scheme, at first, there is a domain
maintaining records for V2X enabled UE and generating initialization phase. In this phase, the domain parameters and
pseudonyms. master secret are created by TA. Moreover, long-term and
TABLE VII
C HALLENGES , S OLUTIONS , M ECHANISMS , AND L IMITATIONS OF S ECURITY IN V2X C OMMUNICATION N ETWORKS
short-term keys are created by TA by using VPD id and mas- with a certain threshold value and revokes the vehicle if it
ter secret. Public keys and domain parameters of VPD is then is greater than a certain value. The authors also performed a
provided to VCF. The second phase is the V2X domain reg- security analysis of the proposed scheme which is described
istration phase, in which long-term keys and signature are below.
loaded by TA into OBU of the vehicle. The third phase is the • Privacy Attack: For privacy protection, short term keys
V2X service registration phase, in which a V-UE performs ser- are frequently changed by V-UE. Even in the worst case
vice registration with VCF. The final phase is the pseudonym when VPD is compromised, the attacker can only get
requesting phase in which pseudonym seed is requested from long term public key and cannot get V-UE real identity
VPD by V-UE. In addition, the authors discussed privacy and and actual IMSI by listening to V2X service registration.
revocation procedure. Privacy is addressed by the frequent • Traceability: A UE may use other’s identity to hide their
change of short-term keys. The revocation procedure is ini- own trace. However, in this scheme, if an attacker cap-
tiated by the vehicle by reporting the rogue V-UE to VPD. tures another vehicle’s short term public key and attaches
TA with the help of VPD compares the revocation parameter to its message, then message verification will fail.
• Frameability: Legit V-UE might get revoked due to a lot the traffic information in less than 1 minute without violating
of fake reports from malicious V-UE. However, in this the privacy of V-UE.
scheme, fake reports from the malicious vehicle using In [74], the authors have evaluated and proposed a secu-
the same short-term key are treated as a single report rity scheme related to resource allocation. In LTE V2X, PC5
lowering the probability for false frameability. communications with semi-persistent resource allocations may
• Attack From Compromised User: Compromised V-UE leak the location privacy of vehicular users. In this article, the
may retrieve pseudonyms by sending pseudonym requests authors have addressed this location privacy issues consider-
to VPD. However, they cannot extract the seed from ing four different message types and have employed different
an encrypted message. In addition, it is difficult to find resource allocation procedure related to different message
long term private key of V-UE which are stored in the types. In the proposed scheme, the type 1 message is related to
tamper-resistant OBU of the vehicle. periodic information from vehicles such as location, speed, and
• Attack From Compromised VPD: A VPD may be compro- direction. This message can be sent to vehicular application
mised to create pseudonyms from long term public keys servers that disseminate messages to the target area through
to mimic V-UE. However, compromised VPD cannot MBMS. Network supported LTE-Uu links can be used for
mimic V-UE due to the unavailability of the short- location privacy. Type 3 message is related to global event
term private key which can only be extracted from the information such as road construction and lane closure and
long-term private key. is sent through LTE-Uu link and location privacy is provided
In [72], the authors proposed and analyzed security scheme by the LTE-Uu link. Type 4 message is related to a message
for V2X networks in crowdsensing applications. In the crowd- from fire-truck, ambulance, and other road signal modules.
sensing applications, vehicles share information from the sur- As in type-4 message, location privacy of special entity is
rounding environment and local sensors to improve road traffic not that of a concern, semi-persistent resource allocation can
safety. In this article, the authors have evaluated two specific be used. Thus, type 1, type 2, and type 3 messages do not
attacks, platoon disruption attack, and perception data falsifica- violate privacy if used according to 3GPP Rel. 14 standard
tion attack. In platoon disruption attack, an attacker repeats the policy. However, type 2 message is related to localized event-
information such as to accelerate or decelerate from the lead triggered information sent through the PC5 link and it reveals
vehicle to deviate the following vehicles from the platoon. In the location information if semi-persistent resource allocation
perception data falsification attack, an attacker or selfish vehi- is used. The authors have proposed random access with sta-
cle inserts or maliciously modify video frames to deceive the tus feedback for type 2 messages which will provide location
other vehicles to make a wrong decision. To prevent these two privacy. In the proposed scheme, all vehicular user equipment
attacks, the authors advocate non-cryptographic methods in allocates data resource block in a random fashion for every
which a vehicle frequently collects the information from other sidelink control period due to which the position of allocated
vehicles in the same platoon in addition to the information resource will not leak the privacy information.
from the lead vehicle. The collected data from lead and other In [75], the authors have investigated group based vehic-
vehicles in the same platoon provide the receiving vehicle with ular communications in the next-generation wireless system.
the capability to identify whether the received information is The authors have addressed two major security challenges.
coherent with the next step operation of the whole platoon. The first challenge is the secure and dynamic management for
However, there might be larger communication overhead and distributed networks and the second challenge is the mobil-
detection delay which are the main limitations of this article. ity and secure group access management for the centralized
In [73], the authors proposed a model where the vehicles can network. The authors have proposed software-defined network
immediately send the videos of a traffic accident to the near- management protocol and have enforced simple network man-
est official vehicles. The proposed protocol takes advantage agement protocol (SNMP)and OpenFlow in base stations. The
of 5G technology such as D2D communications, mmWave authors have presented asymmetric and symmetric key based
communications, and cloud platforms for high-speed com- group handover authentication scheme. Moreover, the authors
munications and low latency. In the proposed protocol, the have proposed an efficient and secure group mobility manage-
traffic accident video is encrypted and sent by the participating ment framework. The authors have applied the group handover
vehicles to the cloud and the cloud send the obtained video authentication protocols to the group mobility management
to nearby law enforcement agency or traffic authority. The framework for fast key agreement and access to the Internet
proposed scheme achieves both conditional anonymity and pri- while moving across heterogeneous networks.
vacy, and traceability. For conditional anonymity and privacy,
the pseudonyms authentication scheme is used. For traceabil-
ity, cooperation between different entities is allowed to trace V. 5G BASED V EHICULAR C OMMUNICATIONS
the actual identity of the vehicle. In the proposed scheme, 5G wireless networks will incorporate various emerging
only TA can identify the relationships between pseudonyms technologies such as massive MIMO, device-to-device com-
certificate and 5G ID of the vehicle. After identifying the 5G munications, multi-radio access technology, full-duplex radios,
ID, TA can send the 5G ID of the vehicle to find the actual millimeter waves, cloud technologies and SDN, as shown in
identity of the vehicle. In addition, the authors have shown Fig. 11. Each of these technologies will bring various solu-
with simulation results that the proposed method can report tions and challenges to 5G based vehicular communications.
Fig. 11. 5G based vehicular communication model.
This section analyzes these key technologies and techniques of communications capacity and handoff rate in cooperative small
5G and their associated challenges and solutions in vehicular cell network [79]. Thus, there exists a tradeoff between vehic-
communications. ular handoff rate and communications capacity and should be
balanced in cooperative transmissions.
A. Heterogeneous and Small Cell Networks In addition, for a smooth handover, content can be cached at
1) Challenges: 5G is expected to be composed of heteroge- candidate cells using mobility and handover predictions [80].
neous networks with macro, micro, and nano cell along with Self-sustaining caching stations can be used to realize this
other access technologies. In addition, 5G is envisioned to mobility and handover aware caching where the content can
adopt small cells to increase network capacity. However, such be pre-cached in the next station before a vehicle con-
schemes face difficulties in providing user mobility due to fre- ducts handover [81]. Frequent content caching is required
quent handovers and increased signal load on the network. for mobility aware caching which introduces a lot of com-
Most of the existing handover schemes are applicable for munication overhead. In addition, popular content and cache
centralized architecture and are not suitable for distributed size should be identified for designing mobility aware caching
and heterogeneous vehicular architecture. Moreover, the han- schemes.
dover decision is mainly determined by a single threshold Cell association optimization technique can also be used
value. This threshold value depends on various parameters to reduce the handover rate [82] in which serving stations
such as network load, channel conditions and received sig- are selected among all available base stations. Existing cell
nal strength [76]. However, it is very difficult to map all of association techniques rely on the transmit power, chan-
these parameters into a single threshold value due to the lack nel and traffic-load, and cause excessive handovers in the
of a suitable model. vehicular scenario. New cell associations are required that
2) Solutions: Efficient vehicular handover scheme is perform cell handover accounting mobility issues, handover
needed for the envisioned ultra-dense distribution of small frequency, handover delays, backhaul variation, and network
cells in 5G communications. Handover performance can be dynamics [83].
enhanced by implementing macro assisted small cells network 5G-enabled smart collaborative network can be used to
and dividing the user and control plane of the network among improve handover issues in heterogeneous vehicular commu-
small and macro cells. Moreover, dual connectivity can be nications [76]. In such networks, the handover decision is
used to reduce the handover complexity where vehicles are performed by an onboard multi-access manager. The multi-
allowed to be concurrently associated with macro and small access manager observes vehicular links and predicts the
base stations [78]. Cooperative transmissions are also rec- handover among cells. In addition, the multi-access manager
ommended for vehicles in cooperative small cell networks. with the help of a ground network can allow communications
The cooperative transmission scheme increases the vehicular between wireless heterogeneous networks.
B. Millimeter Wave Communications the availability of accurate channel state information is one of
1) Challenges: Millimeter wave communications have been the major requirements to avoid multi-user interference and
envisioned for 5G V2X to provide high data rate connec- to achieve high spectral efficiency gain. SINR and achiev-
tions between vehicles and nearby roadside infrastructures. able rate in such systems is limited by pilot contamination
However, the millimeter wave faces several challenges. On which is more pronounced in high mobility situations. In high
the physical layer, the millimeter wave is vulnerable to shad- mobility situations, the length of the pilot signals is restricted
owing. This shadowing effect limits millimeter waves for a due to short channel coherence time. Due to this restric-
point-to-point line of sight connections up to a few hundred tion on the length of the pilot signals, pilot reuse distance
meters. In addition, due to the high operating frequency of between interfering base stations is reduced [77]. In addition,
60 GHz, Doppler spread will be more pronounced in millime- the network performance is significantly impacted by chan-
ter wave communications than in the 2-6 GHz band. Thus, nel estimation and beamforming which are the key issues in
there is a need for special frame design and numerology for a massive MIMO system [89].
millimeter wave communications [30]. 2) Solutions: Several solutions have been suggested
Moreover, delay and reliability, security, and broadcasting to tackle some issues of massive MIMO in vehicular
issues may arise in millimeter wave connections [84]. In the communications. In [90], the authors proposed a model
millimeter wave, there might be a delay while generating for three-dimension MIMO techniques based on vehicular
a physical layer key as well as during the beam search at communications. The proposed model increases the overall
the initial access. In addition, the millimeter wave increases network efficiency and is applicable for various communica-
randomness and may generate the key with a shorter delay. tions environments such as pico, micro and macro cell. In [91],
This quick key generation complicates the information recon- the authors optimized the design of the superimposed pilot in
ciliation process due to the generation of different keys in massive MIMO systems. In the proposed scheme, the specific
two communicating vehicles. Broadcasting is another issue sequences for different cells are first designed based on the
in the millimeter wave due to beamforming communications. Grassmannian link packing (GLP) approach and then extended
In millimeter wave, messages are transmitted by beamform- to form the superimposed pilot matrices for the users in dif-
ing towards a particular group of users which cause problem ferent cells. In [92], the authors proposed a scheme in which
while broadcasting the messages towards vehicles located in pilots are reused among device user equipment to shorten
an angular direction. the pilot overhead in a single cell. In the proposed scheme,
2) Solutions: Location based improvement techniques like pilot contamination due to pilot reuse is reduced with the
iterative search can be used to reduce the delay associated with help of the revised graph coloring algorithm. In addition, the
beam search at the initial access in millimeter wave commu- authors employed iterative power control algorithms for the
nications. The problem associated with broadcasting messages minimization of the data transmit power of D2D links.
towards vehicles located in an angular direction can be solved In [88], the authors recommend the use of different modu-
with the use of in-vehicle transceiver with the combination of lation in different areas according to the channel properties for
wider beams broadcasting the same information in a different enhancing the throughput in the MIMO system. The authors
direction. suggested using transmit-diversity with quadrature phase-shift
In [85], the authors stated that beamforming and directional keying (QPSK) modulation in the area with a very bad channel
transmission can be used in millimeter wave to prevent high condition. Whereas, the authors suggested the use of spatial-
path loss. In addition, ultra-dense deployment of base stations multiplexing with 64-quadrature amplitude modulation (QAM)
or RSUs can be used to tackle the short communication range in the area with a very good channel condition. In the areas
of millimeter wave. In [86], the authors proposed a contex- with moderate channel conditions, the authors recommended
tual online learning algorithm. The authors have applied the using a mixture of both together with 16-QAM modulation to
online learning algorithm to identify data blockages for each increase the throughput in the MIMO system.
vehicle based on aggregated received data and user location Table VIII, IX and X compare benefits, challenges, solu-
information. In addition, the proposed algorithm automati- tions, limitations, and feasibility of various technologies in
cally provides larger coverage in areas with heavy traffic to 5G based vehicular communications. Feasibility indicates the
maximize the system capacity. suitability for deployment which is decided based on the com-
plexity of the proposed solution and changes in the existing
architecture. The term very high in the feasibility column
C. Massive MIMO means very high suitability for deployment followed by the
1) Challenges: Massive MIMO which employs hundreds or term high, whereas the term low means the low suitability for
even thousands of antennas at the base stations provides the deployment.
high magnitude of spectral efficiency gains and can satisfy the
escalating traffic demands of next-generation wireless commu- D. Vehicular Cloud and Fog Computing
nication networks [87]. In [88], the authors have proposed a Vehicular cloud computing (VCC) can provide realistic
reproducible test method for the evaluation of MIMO anten- cloud services for vehicular communications. By collecting
nas in V2X communications and have shown that high data and mining the data from various sensors within a vehicle
rates up to 40 Mbps can be achieved at the real test drive and roadside infrastructures, different applications can be pro-
using MIMO antennas. However, in massive MIMO systems, vided for customers. However, due to the huge volume of
TABLE VIII
B ENEFITS , C HALLENGES , S OLUTIONS , L IMITATIONS , AND F EASIBILITY IN 5G BASED V EHICULAR C OMMUNICATIONS
data coming from various sources, providing realistic vehicular Several challenges of VCC and VFC are highlighted below.
applications has been a challenging issue in vehicular cloud 1) Challenges:
computing [93], [94]. To some extent, these challenges can a) Mobility model: An appropriate and accurate mobility
be solved with the use of vehicular fog computing (VFC). model is required to predict and evaluate the movement of
VFC utilizes vehicles as infrastructures for communications surrounding vehicles so that vehicles can connect with each
and computation. VFC can accommodate the growth of mobile other and share data [97].
traffic in vehicular communications. In addition, it can reduce b) Resource utilization and management: Resource uti-
the latency of traditional vehicular cloud computing. In fog lization and management should be addressed to take full
computing, services are hosted on the fog close to the end advantage of VFC. There exist various challenges for resource
users to eliminate the network hops [95]. Moreover, fog management in vehicular cloud computing. There is a require-
computing reduces the risk of data exposure due to the ment of an efficient protocol to discover and management of
short distances between end devices and fog nodes. Vehicular the resources. In addition, suitable architectures and schemes
crowdsensing based on vehicular fog computing is another are required to access V2X resources from the vehicular cloud
emerging area in which vehicles collect and share location and in heterogeneous wireless networks [101].
environment-specific data for the common interest [96]. Most c) Virtualized network functions: In cloud and fog com-
of the challenges and solutions in vehicular fog computing are puting, virtualized network functions (VNF) runs on the vehi-
also applicable to vehicular crowdsensing. cle hardware and configures security and operation of network
TABLE IX
functions such as load balancing, packet inspection and fire- of the difficult challenges to be addressed in vehicular cloud
walls. However, there are issues associated with virtualization computing.
management, selection and deployment of VNFs. 2) Solutions:
d) Security and incentives: Due to the sharing of content a) Mobility model: Vehicular mobility models such as
and data between vehicles in the fog computing environ- in [98] can be utilized to predict the movement of surrounding
ment, there can be several security attacks such as information vehicles. In [98], the authors compared the unit disc and log-
stealing attacks, hostile attacks and virus spreading attacks. normal shadowing models with obstacle-based channel model
In addition, when vehicles are acquiring navigation results and have shown that the obstacle channel model provides real-
and reporting traffic information in vehicular crowdsensing istic vehicle mobility and topology characteristics. Moreover,
applications, privacy is exposed [96]. Thus, there is a need the authors have used a matching mechanism to tune vari-
to carefully designed security mechanisms for vehicular fog ous parameters of the lognormal model to provide a good
computing. Moreover, in vehicular fog computing, as vehi- match with the obstacle-based channel model. This kind of
cles rent out their resources, there should be a provision mechanism can be employed as a mobility model in 5G based
of some kind of incentives to vehicle owners. There exist vehicular networks.
several challenges for designing incentive models in vehicle b) Resource utilization and management: In [99], the
cloud computing. Appealing compensation or incentive mod- authors discussed resource management strategies in a fog
els should be proposed to stimulate vehicle to share their computing environment to enhance the real-time service qual-
resources as vehicle owners may incur extra operational costs ity of vehicular services. In this article, the authors mainly
while sharing their resources or data. Moreover, the assurance discussed resource reservation and allocation strategies for
of the privacy and security of the involved vehicles is one fog. In these strategies, real-time vehicular services are given
TABLE X
higher priority over other services so that real-time vehicular price of sensing resources to be more accurate and mutual
users can access the services within a single hop. benefit for both parties. In addition, to prevent the privacy
c) Virtualized network functions: In [100], the authors exposure of the vehicle and to achieve the mutual authentica-
proposed the VNFs selection strategy by identifying traffic tion between vehicles and vehicle cloud servers, the authors
packets. In vehicular cloud networks, VNFs are sequentially have employed signcryption technique. The authors have done
deployed according to the order in a specific service func- simulations and have shown that both vehicles and vehicle
tion chain (SFC). The authors have combined the traffic cloud servers receive desired utility while preserving privacy.
packet identification at the source vehicles with the VNFs In [103] the authors have employed partial prepayment strat-
selection strategy. For traffic identification, the authors have egy in which the vehicles co-operating to download a file will
considered different machine learning models including deep receive partial payment before the requesting vehicles receive
neural network and gcForest. For VNFs selection, the network the full packet. This kind of virtual payments or virtual checks
is divided into layers according to the order of VNFs in will stimulate more vehicles to participate in vehicular cloud
SFC. Transmission priority and network congestion are jointly computing.
considered for the scheduling of different layers. Moreover, In [104], the authors proposed a privacy-preserving scheme
the authors have done simulations and have shown that the for real-time navigation systems by utilizing vehicular crowd-
proposed traffic identification and VNFs selection strategy sensing. In the proposed scheme, a vehicle requests navigation
reduces the number of packets failing to reach the destination information from nearby vehicular fog nodes. Group signature
by 30% to 50%. is sent along with the navigation request to preserve the privacy
d) Security and incentives: In [102], the authors of a requesting vehicle. Moreover, crowdsensing vehicles use a
proposed a privacy-preserving incentive mechanism to stim- randomized signature while sending traffic and safety-related
ulate vehicles to share their sensing resources. The authors information to fog nodes to prevent exposure of their iden-
have considered resource sensing evaluation mechanisms to tity. Thus, the proposed scheme provides privacy to vehicular
guarantee fairness evaluation. Moreover, the authors have con- user equipment while sending and obtaining information from
sidered a role based bargain game model to guarantee the vehicular fog nodes.
E. SDN Wi-Fi networks. Authors have done extensive simulations

SDNs have been conceived as a novel technology to pro- and have shown that the low overhead is introduced by the
vide network resilience and flow programmability in 5G proposed platform with heterogeneous services.
networks [105]. SDN enabled 5G vehicular networks con- Control plane can be designed effectively to tackle the
sist of macro and small cells including base stations and issues related to SDN in heterogeneous and integrated vehic-
access points [106]. In SDN vehicular networks, a central- ular networks. In hybrid DSRC and cellular network based
ized controller with OpenFlow protocol controls macro cell V2X, SDN can utilize both ad-hoc and cellular networks to
base stations and access points. In addition, SDN relocates the control the network. However, the hybrid control plane can
control logic from the infrastructure to the control layer to host bring some difficulty in communications between controllers
various applications and provide new functions. Moreover, and vehicles [110]. Thus, there is a need for an efficient con-
the SDN controller is responsible for access networks includ- trol plane scheduler to schedule different SDN events with a
ing authentication, traffic management, and other issues while different priority to different heterogeneous links.
base stations constitute the data plane of the SDN and imple-
ment the controller defined policies. Thus, SDNs can help in F. Mobile Edge Computing
reducing the signaling overhead of the vehicular network with
Mobile edge computing has been envisioned for future
improved communication quality.
5G vehicular communications to provide low latencies. In
In addition, SDNs can be used for inter-networking and
such system, some core network functionalities are moved
exploiting resources in heterogeneous networks in an agile and
to the edge of the network, that is, nearer to the consumer.
flexible manner. Through inter-networking, advantages from
Network function virtualization (NFV) can be used for mobile
different segments and technologies can be utilized to support
edge computing to provide different services. In NFV, core
various vehicular services in an efficient and flexible man-
network functionalities are placed in the eNB and RSUs
ner [107]. Moreover, SDN enabled vehicular networks can be
with storage and processing capacities [111]. In addition,
composed of cloud computing centers, SDN controllers and
mobile edge computing can be utilized to bolster the con-
fog computing clusters [108]. In such networks, SDN con-
trol of 5G-SDN based vehicular networks [112]. Moreover,
trollers transfer the state information from fog clusters to cloud
mobile edge computing servers can administrate the vehicular
centers. These fog computing clusters or fog cells include RSU
group in 5G-SDN based vehicular networks. SDN’s control
centers, RSUs, base stations, users and vehicles and are posi-
plane and mobile edge computing servers can be positioned
tioned at the edge of 5G-SDN based vehicular networks. Fog
in the same place so that mobile edge computing servers
cells based on 5G-SDN can use multi-hop relays to lower
can act as local controllers to localize the control plane of
the frequency of handover between RSUs and vehicles [108].
5G-SDN based vehicular networks. In addition, improved
In addition, they provide higher throughput than traditional
network and resource management, and sustainable develop-
transportation management systems.
ment are some other advantages of mobile edge computing
However, there are several challenges for software-defined
with SDN.
vehicular networks [107].
There are several challenges for edge computing assisted
1) Challenges: 5G vehicular networks due to huge data volume, dynamic
• Inter-Operation: SDN enabled 5G vehicular networks can topology changes and uneven traffic distributions. Moreover,
be used for inter-operation between various technologies. moving the core network functionalities to RSU and eNB will
However, due to different communications standards and increase the cost for large scale deployment. In addition, there
communications links in various networking paradigms, will be a challenge for providing high computing capability
inter-operability may be limited. and storage capacity as well as security in those edge systems.
• Network Management: In SDN networks, there are dif- In [113], the authors discussed challenges and solutions related
ferent hardware and software specifications with various to content sharing in edge computing assisted 5G vehicular
services. Such networks will be more complex to man- networks which are described below.
age. Thus, dynamic and flexible reconfiguration cannot 1) Challenges:
be performed in such networks. • Content Pre-Fetching: Content pre-fetching can be used
• QoS Provisioning: Due to the dynamic topology in a to lower the network access delay and the consump-
vehicular network, resource availability may vary over tion of resources. But there is a challenge in content
time. In addition, this problem is worsened due to the pre-fetching due to the vehicle’s mobility and traffic dis-
high heterogeneity in SDN based vehicular networks. tributions. Thus, for unbalanced and different dwell time,
2) Solutions: In [109], the authors have proposed an open- flexible and network-wide content pre-fetching strategy
source SDN for inter-operation and network management is required.
in heterogeneous networks. The proposed SDN consists of • Content Distribution: After the pre-fetching of the con-
an application program interface (API) for radio access that tent at the desired nodes, content distribution should
clearly separates the user plane and control plane. Proposed be scheduled by macro base stations or roadside units
SDN supports active RAN slicing as well as load balancing according to the requests of the vehicle. The macro base
and mobility management. The proposed platform simultane- station should try to meet the requests of all vehicles by
ously manage heterogeneous networks such as cellular and proper scheduling and by considering link capacity.
2) Solutions: vehicular application. In addition, there should be a proper

• Content Pre-Fetching Based on Machine Learning: guideline to determine whether the network functions need to
Machine learning can be used for predicting driving trace, be centralized or sliced.
traffic situation and the congestion of the base station. 2) Solutions: Some literatures have tried to address issues
Using machine learning, data from the congested base of network slicing in vehicular communications. In [115],
station can be offloaded to un-congested ones. Part of the authors introduced network slicing for vehicular networks
data can also be pre-fetched into the vehicles when it based on air-ground integrated architecture to support specific
enters the congested area using machine learning and then vehicular applications while guaranteeing the quality of ser-
vehicles can serve the content request to other vehicles vice. In the proposed architecture, three slices are introduced.
through DSRC. The first slice is a high definition map slice used for navi-
• Content Distribution Based on Graph Theory: Graph- gating vehicles. The second slice is a file of common interest
based approach can be used to efficiently schedule data slice for distributing popular content. The third slice is on-
dissemination. The neighbor graph can be obtained at the demand transmission slice for real-time interactive sessions
macro base station by aggregating received information and voice calls. Each of these slices supports specific appli-
and then the neighbor graph can be converted into a cations and guarantees the quality of service in integrated
directed matched graph. A maximum weighted indepen- vehicular network architecture.
dent set problem is then formulated in the constructed
conflict graph for content sharing. H. Dynamic Spectrum Sharing
1) Challenges: Licensed and unlicensed spectrum such as
G. Network Slicing cellular, DSRC and millimeter wave spectrum can be dynami-
cally shared to provide a high quality of user experience in 5G
Network slicing is a partition of network functions into vehicular communications [116]. Multi-radio technology can
various virtual network slices and parameter configurations be used to dynamically share and to enhance the utilization
with each slice supporting particular functions and market of these spectrum resources. However, high vehicular mobil-
requirements [114]. Different network slices can be oper- ity limits the availability of 5G spectrum resources. Moreover,
ated in parallel to provide real-time services. Thus, network spectrum efficiency is greatly reduced due to spectrum hetero-
slicing aims to isolate network functions and resources of geneity. Thus, vehicular users need to collaborate with each
a single network infrastructure. Slicing logically divides the other to dynamically share different spectrum resources and
network by splitting a single control plane into multiple con- to enhance spectrum utilization.
trol planes and it affects several control function functionalities 2) Solutions: In [117], the authors studied the vehicular
such as authentication, mobility and session management. In network where cellular V2X and DSRC users coexist in the
addition, it affects user plane functionalities such as being unlicensed spectrum. This unlicensed spectrum can be uti-
programmable and auto-reconfigurable. lized for supporting V2X services. In addition, the authors
1) Challenges: Different network slices can be used such as designed a scheme in which cellular users and DSRC users
safety application network slice for applications that required share the spectrum based on energy-sensing. The proposed
low latency and infotainment application network slice for method allows cellular and DSRC users to coexist in the
applications that required high bandwidth [111]. However, same unlicensed spectrum without data transmission collision
network slicing required several capabilities [114] such as: between them.
• Slice Description: Service level agreement (SLA) require-
ments are captured by a slice description. This slice VI. D ISCUSSION AND F UTURE R ESEARCH D IRECTIONS
description should be adapted by network elements.
• Slice Instantaneous: Common virtual network functions,
Vehicular communications are evolving rapidly with the
control plane and user plane architecture and interface emergence of novel technologies in cellular communications.
should be identified for slice instantaneous. There are various benefits and challenges due to the latest tech-
• Slice Life Cycle Management: Proper configuration, adap-
nological concepts. There are still many areas to be explored in
tion and monitoring of network elements are required for the domain of the novel concepts to ensure efficient vehicular
slice life cycle management. communication networks. Few insights on some of the impor-
There are several challenges for network slicing assisted 5G tant evolving issues, based on lessons learned from existing
vehicular networks. concepts, are highlighted below.
• Slice manager should configure multiple slices per device
simultaneously. A. Physical Layer Structure
• Network slicing can emerge security threats as different Various works have already been done to modify the frame
security services may be required for different slices. structure or physical layer structure of cellular V2X commu-
• Trust relationships are required between the slice man- nications. As discussed before, 3GPP has considered agile
ager and provider owning to the infrastructure. frame structure on NR Release 16 [12]. Shorter TTI and
Thus, for network slicing to be effective, vehicular appli- larger SCS are desired in cellular based V2X communica-
cations need to be translated to technical specifications and tions to reduce the Doppler effect. Scalable SCS as discussed
slice manager should be able to configure multiple slices per in [31] can be used for growing V2X use cases. Although
wider SCS can better handle frequency offset and Doppler Further research is needed to analyze delays and other issues
shift, inter-symbol interference will be high. For reducing associated with Deep Q-learning for resource allocations in
the inter-symbol interference extended CP can be used which vehicular communications.
in turn reduces the OFDM efficiency. Thus, further research
is required to address the issue of inter-symbol interference
in scalable SCS in NR Release 16. Moreover, a unified E. Security
frame structure with a suitable number of DMRS symbols Most of the existing cryptographic schemes as discussed in
is required to address the issue of highly dynamic vehicular this survey can provide protection from external attacks but are
communications. found to be less effective against the attacks generated within
the network. A misbehavior detection system (MDS) can be
B. Synchronization deployed to prevent such attacks. Misbehavior or anomaly
Synchronization issues will be more exacerbated in out of detection systems are complicated computing systems for
network and in partial coverage areas in V2X networks. GNSS monitoring and detecting malicious behaviors [119]–[123].
can be used as a synchronization source as explained in [29], In vehicular communications, machine learning based MDS
however, there still exist problems with the absolute time offset can be used to detect several attacks such as false alert
error in partial coverage and out of coverage. Absolute time attack and position falsification attack [124], [125]. Thus,
offset error is more pronounced for out of coverage vehicu- it is essential to address both cryptographic and MDS
lar users due to the radio wave propagation time from GNSS. based schemes for strengthening the security of LTE V2X
Further research should try to minimize the absolute time off- communications. Future research direction in V2X security
set error for out of coverage users and should design a protocol should address both cryptographic and misbehavior detection
that will allow vehicle UE to easily switch the synchronization issues.
source. 5G V2X will incorporate SDN, mobile edge computing
and network function virtualization which will bring addi-
C. MBMS tional security challenges. In SDN based V2X networks,
The use of MBMS has a great advantage in cellular based the attackers may attack the northbound and southbound
V2X communications. However, determining the V2X broad- application program interface (API) and can have control
cast area and latency are still major challenges. Arranging over the SDN network through the controller. Attackers will
small non-overlapped MBMS service areas is a challenging then be able to create their own policies and disrupt the
task. In addition, it is difficult to transmit V2X messages in V2X services. SDN based privacy-preserving and mutual
different overlapped areas. To transmit messages in overlapped authentication framework can be used for simplifying the
areas, TMGI, as discussed in [35], can be used. However, network management and enhancing privacy [127], [128].
TMGI management in overlapped areas is complicated and However, these schemes lack real test-beds implementation.
should be addressed in future research. For the latency issues Future work should address the authentication and privacy
in MBMS, localized MBMS can be used. However, the cur- issues of V2X communications using real test-beds imple-
rent architecture of cellular V2X does not support the joint mentation. Moreover, network function virtualization will
local and centralized MBMS service and should be addressed allow the movement of security functions throughout the
in future research. network. However, there might be issues while moving secu-
rity functions throughout the network [129]. In addition,
D. Resource Allocations edge computing will bring security services closer to the
Resource allocation is one of the most researched areas user which will introduce additional challenges related to
in cellular based vehicular communications. However, there security.
still exists a lot of challenges in resource allocations. NOMA
based techniques are envisioned in 5G based vehicular
communications which can provide various V2X services F. Heterogenous Networks
and supports massive connectivity [126]. However, resource Macro, micro, and pico cell will be the components of het-
allocation using the NOMA based technique is challeng- erogeneous networks in 5G V2X. Especially in micro and
ing due to a lot of interference produced at the receiver pico cells, there arise issues of handover. As discussed ear-
side. Information can be exchanged between receiver and lier various works have been proposed to address the issue of
sender to cancel successive interference as discussed in [51]. handover in highly dynamic vehicular communications. The
However, a lot of communication overhead is introduced due promising solutions are cell co-operation and dual connectiv-
to information exchanged. Thus, future research should try ity [78], [79]. However, there might be high backhaul signaling
to address the interference cancellation issue without caus- overhead. Future work should address the issue related to sig-
ing a lot of communication overhead. Another promising naling overhead in co-operative handover schemes. Another
area is the reinforcement learning based resource allocations promising solution is the identification of candidate cells to
in which reinforcement or Q-learning is used for the dis- pre-fetch the content during handover. However, there is a
tributed power selection. For large state state-action pairs, need for reliable prediction schemes such as machine learning
graph and Deep Q-learning based resource allocations as based prediction to identify the candidate cell for the handover
proposed in [118] can be used for distributed power allocation. process.
G. Millimeter Wave Communications with accurately identifying suitable vehicles for content pre-
Millimeter wave communications will provide a high fetching. The mobility model along with a trust model com-
data rate in 5G based V2X communications. Broadcast bined with the machine learning model can be used to select
communication and directional transmission for V2X use cases a suitable vehicle for content pre-fetching. Similarly, graph
remain the challenging problems in millimeter wave com- and mobility based vehicle selection should be performed for
munications. Contextual online learning [86] is a promising content distribution.
solution to identify the V2X application area and beam-
forming. However, contextual online learning requires a huge L. Network Slicing
amount of periodic data and location information. Future Network slices can be tailored to specific V2X use cases to
research should address the challenges associated with the data provide real time V2X applications. However, there exists a
gathering for contextual online learning in millimeter wave slice configuration and security or trust issues. A trust model
communications. can be developed between the slice manager and provider own-
ing to the infrastructure. The dynamic trust model based on
H. Massive MIMO reinforcement learning can be used to establish trust between
Pilot contamination and CSI estimation are the major chal- network slice manager and vehicles. Moreover, SDNs should
lenges for the massive MIMO system in V2X communications. be integrated into V2X networks which can be virtualized and
Promising solutions such as Block-diagonal [90] and graph sliced network functions according to V2X use cases.
based scheme [91] have been proposed to address the issues
related to Massive MIMO. However, these schemes have M. Dynamic Spectrum Sharing
issues related to obtaining specific sequences and iterative
Dynamic spectrum sharing in multi-radio technology can
power control algorithms. Future research should clearly
be used to enhance the spectrum utilization in V2X networks.
address the issues related to obtaining sequences in differ-
V2X users should dynamically share the spectrum with other
ent cells and reducing pilot contamination. Graph based or
users to enhance V2X capabilities. The co-existing of V2X and
machine learning based scheme can be used to reduce pilot
DSRC users in the unlicensed spectrum as discussed in [117]
contamination.
can be considered. However, interference will be the main
I. Vehicular Cloud or Fog Computing issue in such schemes. A complex channel estimation method
can be used for avoiding the interference during dynamic spec-
Resource management, security and incentives are major
trum sharing. Future works should consider the graph and
challenges in vehicular cloud computing. As discussed vari-
machine learning based channel estimation method in dynamic
ous solutions have been proposed to address these challenges.
spectrum sharing applications.
However, delay and inaccurate predictions are the main issues
in most of the proposed solutions [100], [102]. For privacy-
VII. C ONCLUSION
preserving and incentive schemes, delay plays an important
factor. There is a tradeoff between privacy and delay and it In this article, we provide a comprehensive survey of the
should be balanced according to V2X applications. Future recent development in LTE and 5G to support V2X com-
works in vehicular fog computing should try to address the munications. We show how strong aspects of LTE such as
issues related to delay for resource management, privacy and wide coverage, high capacity and high penetration paved the
incentives solutions. path for LTE based V2X communications. We also show that
several challenges lie ahead before LTE can be massively
J. SDN utilized for vehicular communications. Physical layer struc-
ture and synchronization are some of the main challenges
SDNs based vehicular communications will reduce the sig- for LTE V2X communications due to high relative velocity,
naling overhead and allows to access heterogeneous resources high frequency and frequent handover in V2X communica-
in real-time. However, inter-operation network management tions. Resource allocation will be another challenge where the
using SDN in heterogeneous networks is still an open chal- resources being used by the vehicular system should not con-
lenge. Dynamic and flexible configuration systems in het- flict with the resources being used by cellular users. Moreover,
erogeneous environments as proposed in [109] can be used. the security and MBMS system of LTE needs to be redesigned
However, such schemes introduce load balancing issues due and analyzed for V2X communications. We also discuss var-
to the attachment of different types of loads in heterogeneous ious 3GPP and non-3GPP based solutions to address these
V2X networks. Future works should clearly address the issues challenges. In addition, the incorporation of various emerg-
related to load balancing and slicing various V2X applications. ing technologies such as massive MIMO, device-to-device
communications, multi-radio access technology, full-duplex
K. Mobile Edge Computing radios, millimeter waves, cloud technologies and SDN in 5G
Mobile edge computing will provide low latency and will bring various challenges to V2X communications. There
bring various V2X use cases close to the user. Content pre- are active researches going on to address these challenges.
fetching and content distribution are the major challenges Furthermore, a cross-industry organization such as 5GAA is
in edge computing. The machine learning algorithm can be working to develop end-to-end solutions for 5G based V2X
used for content pre-fetching. However, there is an issue communications.
R EFERENCES [26] R. Molina-Masegosa and J. Gozalvez, “LTE-V for sidelink 5G V2X

vehicular communications: A new 5G technology for short-range
[1] Optimizing 5G for V2X—Requirements, Implications and Challenges, vehicle-to-everything communications,” IEEE Veh. Technol. Mag.,
CISCO, San Jose, CA, USA. Sep. 2014. vol. 12, no. 4, pp. 30–39, Dec. 2017.
[2] J. B. Kenney, “Dedicated short-range communications (DSRC) stan- [27] H. Seo, K. D. Lee, S. Yasukawa, Y. Peng, and P. Sartori, “LTE evolu-
dards in the United States,” Proc. IEEE, vol. 99, no. 7, pp. 1162–1182, tion for vehicle-to-everything services,” IEEE Commun. Mag., vol. 54,
Jul. 2011. no. 6, pp. 22–28, Jun. 2016.
[3] J. Harding et al., Vehicle-to-Vehicle Communications: Readiness of V2V [28] A. Bazzi, B. M. Masini, A. Zanella, and I. Thibault, “On the
Technology for Application, U.S. Nat. Highway Traffic Safety Admin., performance of IEEE 802.11p and LTE-V2V for the cooperative aware-
Washington, DC, USA, Aug. 2014. ness of connected vehicles,” IEEE Trans. Veh. Technol., vol. 66, no. 11,
[4] K. Gay and V. Kniss, “Safety pilot model deployment: Lessons learned pp. 10419–10432, Nov. 2017.
and recommendations for future connected vehicle activities,” U.S. [29] S. Lien et al., “3GPP NR sidelink transmissions toward 5G V2X,”
Dept. Transp., Washington, DC, USA, Rep. FHWA-JPO-16-363, 2018. IEEE Access, vol. 8, pp. 35368–35382, 2020.
[Online]. Available: https://ntl.bts.gov/lib/59000/59300/59361/FHWA- [30] M. Boban, K. Manolakis, M. Ibrahim, S. Bazzi, and W. Xu, “Design
JPO-16-363.pdf aspects for 5G V2X physical layer,” in Proc. IEEE Conf. Stand.
[5] S. Gyawali, S. Xu, F. Ye, R. Q. Hu, and Y. Qian, “A D2D based Commun. Netw. (CSCN), Berlin, Germany, 2016, pp. 1–7.
clustering scheme for public safety communications,” in Proc. IEEE [31] C. Campolo, A. Molinaro, F. Romeo, A. Bazzi, and A. O. Berthet, “5G
87th Veh. Technol. Conf. (VTC Spring), 2018, pp. 1–5. NR V2X: On the impact of a flexible numerology on the autonomous
[6] H. Ding, S. Ma, and C. Xing, “Feasible D2D communication distance sidelink mode,” in Proc. IEEE 2nd 5G World Forum (5GWF), Dresden,
in D2D-enabled cellular networks,” in Proc. IEEE Int. Conf. Commun. Germany, 2019, pp. 102–107.
Syst., 2014, pp. 1–5. [32] S. H. Sun, J. L. Hu, Y. Peng, X. M. Pan, L. Zhao, and J. Y. Fang,
[7] “3rd generation partnership project; technical specification group radio “Support for vehicle-to-everything services based on LTE,” IEEE
access network; study on LTE-based V2X services, rel-14 V0.5.0,” Wireless Commun., vol. 23, no. 3, pp. 4–8, Jun. 2016.
3GPP, Sophia Antipolis, France, Rep. TR 36.885, Feb. 2016. [33] L. Gallo and J. Haerri, “Unsupervised long-term evolution device-to-
[8] “Technical specification group services and system aspects; study on device: A case study for safety-critical V2X communications,” IEEE
LTE support for vehicle to everything (V2X) services, rel. 14,” 3GPP, Veh. Technol. Mag., vol. 12, no. 2, pp. 69–77, Jun. 2017.
Sophia Antipolis, France, Rep. TR 22.885, Sep. 2015. [34] K. Manolakis and W. Xu, “Time synchronization for multi-link
[9] C. Hoymann et al., “LTE release 14 outlook,” IEEE Commun. Mag., D2D/V2X communication,” in Proc. IEEE 84th Veh. Technol. Conf.
vol. 54, no. 6, pp. 44–49, Jun. 2016. (VTC-Fall), Montreal, QC, Canada, 2016, pp. 1–6.
[10] The Path to 5G: Cellullar Vehicel to Everything (C-V2X), [35] G. Araniti, C. Campolo, M. Condoluci, A. Iera, and A. Molinaro, “LTE
QUALCOMM, San Diego, CA, USA, Feb. 2017. for vehicular networking: A survey,” IEEE Commun. Mag., vol. 51,
[11] K. Abboud, H. A. Omar, and W. Zhuang, “Interworking of DSRC and no. 5, pp. 148–157, May 2013.
cellular network technologies for V2X communications: A survey,” [36] A. Khina, T. Philosof, and M. Laifenfeld, “Multicast MIMO enhance-
IEEE Trans. Veh. Technol., vol. 65, no. 12, pp. 9457–9470, Dec. 2016. ment for V2X over LTE,” in Proc. IEEE Int. Conf. Microw. Commun.
[12] “3rd generation partnership project; technical specification group Antennas Electron. Syst. (COMCAS), 2015, pp. 1–5.
services and system aspects; release 16 description; summary of rel-16 [37] S. Roger et al., “Low-latency layer-2-based multicast scheme for local-
work items, rel. 16,” 3GPP, Sophia Antipolis, France, Rep. TR 21916, ized V2X communications,” IEEE Trans. Intell. Transp. Syst., vol. 20,
Mar. 2020. no. 8, pp. 2962–2975, Aug. 2019.
[13] J. G. Andrews et al., “What will 5G be?” IEEE J. Sel. Areas Commun., [38] L. Liang, G. Y. Li, and W. Xu, “Resource allocation for D2D-enabled
vol. 32, no. 6, pp. 1065–1082, Jun. 2014. vehicular communications,” IEEE Trans. Commun., vol. 65, no. 7,
[14] A. Gupta and R. K. Jha, “A survey of 5G network: Architecture and pp. 3186–3197, Jul. 2017.
emerging technologies,” IEEE Access, vol. 3, pp. 1206–1232, 2015. [39] J. Mei, K. Zheng, L. Zhao, Y. Teng, and X. Wang, “A latency and
[15] Z. MacHardy, A. Khan, K. Obana, and S. Iwashina, “V2X access reliability guaranteed resource allocation scheme for LTE V2V com-
technologies: Regulation, research, and remaining challenges,” IEEE munication systems,” IEEE Trans. Wireless Commun., vol. 17, no. 6,
Commun. Surveys Tuts., vol. 20, no. 3, pp. 1858–1877, 3rd Quart., pp. 3850–3860, Jun. 2018.
2018. [40] L. Liang, J. Kim, S. C. Jha, K. Sivanesan, and G. Y. Li, “Spectrum
[16] M. Muhammad and G. A. Safdar, “Survey on existing authentica- and power allocation for vehicular communications with delayed CSI
tion issues for cellular-assisted V2X communication,” Veh. Commun., feedback,” IEEE Wireless Commun. Lett., vol. 6, no. 4, pp. 458–461,
vol. 12, pp. 50–65, Feb. 2018. Aug. 2017.
[17] Z. H. Mir and F. Filali, “Simulation and performance evaluation of [41] W. Sun, E. G. Ström, F. Brännström, K. C. Sou, and Y. Sui, “Radio
vehicle-to-vehicle (V2V) propagation model in urban environment,” resource management for D2D-based V2V communication,” IEEE
in Proc. 7th Int. Conf. Intell. Syst. Model. Simulat. (ISMS), Bangkok, Trans. Veh. Technol., vol. 65, no. 8, pp. 6636–6650, Aug. 2016.
Thailand, 2016, pp. 394–399. [42] Y. Meng, Y. Dong, X. Liu, and Y. Zhao, “An interference-aware
[18] Z. H. Mir and F. Filali, “LTE and IEEE 802.11p for vehicu- resource allocation scheme for connectivity improvement in vehicular
lar networking: A performance evaluation,” EURASIP J. Wireless networks,” IEEE Access, vol. 6, pp. 51319–51328, 2018.
Commun. Netw., vol. 2014, no. 1, p. 89, Dec. 2014. [43] H. He, H. Shan, A. Huang, and L. Sun, “Resource allocation for video
[19] S. Chen, J. Hu, Y. Shi, and L. Zhao, “LTE-V: A TD-LTE-based V2X streaming in heterogeneous cognitive vehicular networks,” IEEE Trans.
solution for future vehicular network,” IEEE Internet Things J., vol. 3, Veh. Technol., vol. 65, no. 10, pp. 7917–7930, Oct. 2016.
no. 6, pp. 997–1005, Dec. 2016. [44] Z. Xiao et al., “Spectrum resource sharing in heterogeneous vehic-
[20] S. Mumtaz et al., “Cognitive vehicular communication for 5G,” IEEE ular networks: A noncooperative game-theoretic approach with cor-
Commun. Mag., vol. 53, no. 7, pp. 109–117, Jul. 2015. related equilibrium,” IEEE Trans. Veh. Technol., vol. 67, no. 10,
[21] X. Wang, S. Mao, and M. X. Gong, “An overview of 3GPP cellular pp. 9449–9458, Oct. 2018.
vehicle-to-everything standards,” Mobile Comput. Commun., vol. 21, [45] G. Li, Z. Yang, S. Chen, Y. Li, and P. Yuan, “A traffic flow-based and
no. 3, p. 154, Sep. 2017. dynamic grouping-enabled resource allocation algorithm for LTE-D2D
[22] Z. Xu, X. Li, X. Zhao, M. Zhang, and Z. Wang, “DSRC versus 4G- vehicular networks,” in Proc. IEEE/CIC Int. Conf. Commun. China
LTE for connected vehicle applications: A study on field experiments (ICCC), Chengdu, China, 2016, pp. 1–6.
of vehicular communication performance,” J. Adv. Transp., vol. 2017, [46] W. Sun, D. Yuan, E. G. Ström, and F. Brännström, “Cluster-based
Aug. 2017, Art. no. 2750452. radio resource management for D2D-supported safety-critical V2X
[23] A. Vinel, “3GPP LTE versus IEEE 802.11p/WAVE: Which technology communications,” IEEE Trans. Wireless Commun., vol. 15, no. 4,
is able to support cooperative vehicular safety applications?” IEEE pp. 2756–2769, Apr. 2016.
Wireless Commun. Lett., vol. 1, no. 2, pp. 125–128, Apr. 2012. [47] L. Liang, S. Xie, G. Y. Li, Z. Ding, and X. Yu, “Graph-based resource
[24] Multimedia Broadcast/Multicast Service (MBMS); Architecture and sharing in vehicular communication,” IEEE Trans. Wireless Commun.,
Functional Description (Release 13), V13.0.0, 3GPP Standard TS vol. 17, no. 7, pp. 4579–4592, Jul. 2018.
23.246, 2015. [48] W. Xing, N. Wang, C. Wang, F. Liu, and Y. Ji, “Resource allocation
[25] Universal Mobile Telecommunications System (UMTS); LTE; schemes for D2D communication used in VANETs,” in Proc. IEEE
Architecture Enhancements for V2X Services, Rel-14 V14.2.0, 3GPP 80th Veh. Technol. Conf. (VTC-Fall), Vancouver, BC, Canada, 2014,
Standard TS 33.885, May 2017. pp. 1–6.
[49] K. Zheng, H. Meng, P. Chatzimisios, L. Lei, and X. Shen, “An SMDP- [72] K. Bian, G. Zhang, and L. Song, “Toward secure crowd sens-
based resource allocation in vehicular cloud computing systems,” IEEE ing in vehicle-to-everything networks,” IEEE Netw., vol. 32, no. 2,
Trans. Ind. Electron., vol. 62, no. 12, pp. 7920–7928, Dec. 2015. pp. 126–131, Mar./Apr. 2018.
[50] R. Yu, J. Ding, X. Huang, M. Zhou, S. Gjessing, and Y. Zhang, [73] M. H. Eiza, Q. Ni, and Q. Shi, “Secure and privacy-aware cloud-
“Optimal resource sharing in 5G-enabled vehicular networks: A assisted video reporting service in 5G-enabled vehicular networks,”
matrix game approach,” IEEE Trans. Veh. Technol., vol. 65, no. 10, IEEE Trans. Veh. Technol., vol. 65, no. 10, pp. 7868–7881, Oct. 2016.
pp. 7844–7856, Oct. 2016. [74] K. J. Ahmed and M. J. Lee, “Secure resource allocation for LTE-
[51] B. Di, L. Song, Y. Li, and Z. Han, “V2X meets NOMA: Non- based V2X service,” IEEE Trans. Veh. Technol., vol. 67, no. 12,
orthogonal multiple access for 5G-enabled vehicular networks,” IEEE pp. 11324–11331, Dec. 2018.
Wireless Commun., vol. 24, no. 6, pp. 14–21, Dec. 2017. [75] C. Lai, H. Zhou, N. Cheng, and X. S. Shen, “Secure group commu-
[52] B. Di, L. Song, Y. Li, and G. Y. Li, “Non-orthogonal multiple nications in vehicular networks: A software-defined network-enabled
access for high-reliable and low-latency V2X communications in 5G architecture and solution,” IEEE Veh. Technol. Mag., vol. 12, no. 4,
systems,” IEEE J. Sel. Areas Commun., vol. 35, no. 10, pp. 2383–2397, pp. 40–49, Dec. 2017.
Oct. 2017. [76] P. Dong, T. Zheng, S. Yu, H. Zhang, and X. Yan, “Enhancing vehicu-
[53] H. Zheng, H. Li, S. Hou, and Z. Song, “Joint resource allocation lar communication using 5G-enabled smart collaborative networking,”
with weighted max–min fairness for NOMA-enabled V2X commu- IEEE Wireless Commun., vol. 24, no. 6, pp. 72–79, Dec. 2017.
nications,” IEEE Access, vol. 6, pp. 65449–65462, 2018. [77] S. Schwarz and M. Rupp, “Society in motion: Challenges for LTE and
[54] Discussion on Resource Allocation Enhancement for PC5 Based beyond mobile communications,” IEEE Commun. Mag., vol. 54, no. 5,
V2V Communications, document R1-157438, 3GPP TSG RAN WG1 pp. 76–83, May 2016.
Meeting #83, Anaheim, CA, USA, Nov. 2015. [78] Y. Wu, K. Guo, J. Huang, and X. S. Shen, “Secrecy-based energy-
[55] Discussion on Enhancement for PC5 Based V2V Resource Allocation, efficient data offloading via dual connectivity over unlicensed spec-
document R1-157435, 3GPP TSG RAN WG1 Meeting #83, LG trums,” IEEE J. Sel. Areas Commun., vol. 34, no. 12, pp. 3252–3270,
Electronics, Anaheim, CA, USA, Nov. 2015. Dec. 2016.
[56] Geo-Location Resource Allocation Based on Zones and Headings, doc- [79] X. Ge, H. Cheng, G. Mao, Y. Yang, and S. Tu, “Vehicular commu-
ument R2-164097, 3GPP TSG RAN WG2 Meeting #94, InterDigital, nications for 5G cooperative small-cell networks,” IEEE Trans. Veh.
Nanjing, China, May 2016. Technol., vol. 65, no. 10, pp. 7882–7894, Oct. 2016.
[57] Sensing Based Collision Avoidance for V2V Communication, docu- [80] R. Wang, X. Peng, J. Zhang, and K. B. Letaief, “Mobility-aware
ment R1-160432, 3GPP TSG RAN WG1 Meeting #84, 3GPP, Sophia caching for content-centric wireless networks: Modeling and method-
Antipolis, France, Feb. 2016. ology,” IEEE Commun. Mag., vol. 54, no. 8, pp. 77–83, Aug. 2016.
[58] Discussion on Resource Pool Structure and Control Signalling for PC5-
[81] S. Zhang, N. Zhang, X. Fang, P. Yang, and X. S. Shen, “Self-
Based V2V, document R1-156892, 3GPP TSG RAN WG1 Meeting
sustaining caching stations: Toward cost-effective 5G-enabled vehic-
#83, 3GPP, Sophia Antipolis, France, Nov. 2015.
ular networks,” IEEE Commun. Mag., vol. 55, no. 11, pp. 202–208,
[59] J. Yang, B. Pelletier, and B. Champagne, “Enhanced autonomous Nov. 2017.
resource selection for LTE-based V2V communication,” in Proc. IEEE
[82] S. Sadr and R. Adve, “Handoff rate and coverage analysis in multi-
Veh. Netw. Conf. (VNC), Columbus, OH, USA, 2016, pp. 1–6.
tier heterogeneous networks,” IEEE Trans. Wireless Commun., vol. 14,
[60] L. Hu, J. Eichinger, M. Dillinger, M. Botsov, and D. Gozalvez, “Unified
no. 5, pp. 2626–2638, May 2015.
device-to-device communications for low-latency and high reliable
[83] L. P. Qian, Y. Wu, H. Zhou, and X. Shen, “Non-orthogonal multiple
vehicle-to-X services,” in Proc. IEEE 83rd Veh. Technol. Conf. (VTC
access vehicular small cell networks: Architecture and solution,” IEEE
Spring), Nanjing, China, 2016, pp. 1–7.
Netw., vol. 31, no. 4, pp. 15–21, Jul./Aug. 2017.
[61] S. Maghsudi and S. Stanczak, “Hybrid centralized distributed resource
allocation for device-to-device communication underlaying cellular [84] F. J. Martin-Vega, M. C. Aguayo-Torres, G. Gomez,
networks,” IEEE Trans. Veh. Technol., vol. 65, no. 4, pp. 2481–2495, J. T. Entrambasaguas, and T. Q. Duong, “Key technologies,
Apr. 2016. modeling approaches, and challenges for millimeter-wave vehicular
communications,” IEEE Commun. Mag., vol. 56, no. 10, pp. 28–35,
[62] H. La Vinh and A. R. Cavalli, “Security attacks and solutions in vehicu-
Oct. 2018.
lar ad hoc networks: A survey,” Int. J. Ad Hoc Netw. Syst., vol. 4, no. 2,
pp. 1–20, 2014. [85] S. Rangan, T. S. Rappaport, and E. Erkip, “Millimeter-wave cellular
[63] J. Cao, M. Ma, H. Li, Y. Zhang, and Z. Luo, “A survey on security wireless networks: Potentials and challenges,” Proc. IEEE, vol. 102,
aspects for LTE and LTE-A networks,” IEEE Commun. Surveys Tuts., no. 3, pp. 366–385, Feb. 2014.
vol. 16, no. 1, pp. 283–302, 1st Quart., 2014. [86] G. H. Sim, S. Klos, A. Asadi, A. Klein, and M. Hollick, “An
[64] “3rd generation partnership project; technical specification group online context-aware machine learning algorithm for 5G mmWave
services and system aspects; study on security aspects for LTE support vehicular communications,” IEEE/ACM Trans. Netw., vol. 26, no. 6,
of V2X services, Rel-14 V14.1.0,” 3GPP, Sophia Antipolis, France, pp. 2487–2500, Dec. 2018.
Rep. TR 33.885, Sep. 2017. [87] E. G. Larsson, O. Edfors, F. Tufvesson, and T. L. Marzetta, “Massive
[65] 3rd Generation Partnership Project; Technical Specification Group MIMO for next generation wireless systems,” IEEE Commun. Mag.,
Services and System Aspects; Security Aspects for LTE Support of V2X vol. 52, no. 2, pp. 186–195, Feb. 2014.
Services, Rel-15 V15.0.0, 3GPP Standard TS 33.185, Jun. 2018. [88] M. Almarashli and S. Lindenmeier, “Evaluation of vehicular 4G/5G-
[66] M. Labib, V. Marojevic, J. H. Reed, and A. I. Zaghloul, “How to MIMO antennas via data-rate measurement in an emulated urban test
enhance the immunity of LTE systems against RF spoofing,” in Proc. drive,” in Proc. 48th Eur. Microw. Conf. (EuMC), Madrid, Spain, 2018,
Int. Conf. Comput. Netw. Commun. (ICNC), Kauai, HI, USA, 2016, pp. 300–303.
pp. 1–5. [89] T. E. Bogale and L. B. Le, “Massive MIMO and mmWave for 5G
[67] P. Agarwal, D. E. Thomas, and A. Kumar, “Security analysis of wireless HetNet: Potential benefits and challenges,” IEEE Veh. Technol.
LTE/SAE networks under de-synchronization attack for hyper-erlang Mag., vol. 11, no. 1, pp. 64–75, Mar. 2016.
distributed residence Time,” IEEE Commun. Lett., vol. 21, no. 5, [90] H. Jiang, Z. Zhang, J. Dang, and L. Wu, “A novel 3-D massive MIMO
pp. 1055–1058, May 2017. channel model for vehicle-to-vehicle communication environments,”
[68] Q. Vien, T. Le, X. Yang, and T. Duong, “On the handover security key IEEE Trans. Commun., vol. 66, no. 1, pp. 79–90, Jan. 2018.
update and residence management in LTE networks,” in Proc. IEEE [91] X. Jing, M. Li, H. Liu, S. Li, and G. Pan, “Superimposed pilot
Wireless Commun. Netw. Conf. (WCNC), 2017, pp. 1–6. optimization design and channel estimation for multiuser mas-
[69] Y. Qian and N. Moayeri, “Design of secure and application-oriented sive MIMO systems,” IEEE Trans. Veh. Technol., vol. 67, no. 12,
VANETs,” in Proc. IEEE Veh. Technol. Conf. (VTC Spring), Singapore, pp. 11818–11832, Dec. 2018.
2008, pp. 2794–2799. [92] H. Xu, N. Huang, Z. Yang, J. Shi, B. Wu, and M. Chen, “Pilot allo-
[70] A. Daniel, A. Kiening, and F. Stumpf, “PAL—Privacy augmented LTE: cation and power control in D2D underlay massive MIMO systems,”
A pricacy-preserving scheme for vehicular LTE communication,” in IEEE Commun. Lett., vol. 21, no. 1, pp. 112–115, Jan. 2017.
Proc. 10th ACM Int. Workshop Veh. Inter Netw. Syst. Appl. (ACM [93] M. Tao, J. Li, J. Zhang, X. Hong, and C. Qu, “Vehicular data cloud plat-
VANET), 2013, pp. 1–9. form with 5G support: Architecture, services, and challenges,” in Proc.
[71] K. J. Ahmed and M. J. Lee, “Secure, LTE-based V2X service,” IEEE IEEE Int. Conf. Comput. Sci. Eng. (CSE) IEEE Int. Conf. Embedded
Internet Things J. , vol. 5, no. 5, pp. 3724–3732, Oct. 2018. Ubiquitous Comput. (EUC), Guangzhou, China, 2017, pp. 32–37.
[94] E. Skondras, A. Michalas, and D. D. Vergados, “A survey on medium [117] P. Wang, B. Di, H. Zhang, K. Bian, and L. Song, “Cellular V2X
access control schemes for 5G vehicular cloud computing systems,” in communications in unlicensed spectrum: Harmonious coexistence with
Proc. Global Inf. Infrastruct. Netw. Symp. (GIIS), Thessaloniki, Greece, VANET in 5G systems,” IEEE Trans. Wireless Commun., vol. 17,
2018, pp. 1–5. no. 8, pp. 5212–5224, Aug. 2018.
[95] M. Sookhak et al., “Fog vehicular computing: Augmentation of fog [118] S. Gyawali, Y. Qian, and R. Q. Hu, “Resource allocation in vehic-
computing using vehicular cloud computing,” IEEE Veh. Technol. Mag., ular communications using graph and deep reinforcement learning,”
vol. 12, no. 3, pp. 55–64, Sep. 2017. in Proc. IEEE Global Commun. Conf. (GLOBECOM), Waikoloa, HI,
[96] J. Ni, A. Zhang, X. Lin, and X. S. Shen, “Security, privacy, and fairness USA, 2019, pp. 1–6.
in fog-based vehicular crowdsensing,” IEEE Commun. Mag., vol. 55, [119] S. Xu, Y. Qian, and R. Q. Hu, “Data-driven network intelli-
no. 6, pp. 146–152, Jun. 2017. gence for anomaly detection,” IEEE Netw., vol. 33, no. 3, pp. 88–95,
[97] X. Hou, Y. Li, M. Chen, D. Wu, D. Jin, and S. Chen, “Vehicular fog May/Jun. 2019.
computing: A viewpoint of vehicles as the infrastructures,” IEEE Trans. [120] S. Xu, Y. Qian, and R. Q. Hu, “A data-driven preprocessing scheme
Veh. Technol., vol. 65, no. 6, pp. 3860–3873, Jun. 2016. on anomaly detection in big data applications,” in Proc. IEEE Conf.
[98] N. Akhtar, S. C. Ergen, and O. Ozkasap, “Vehicle mobility and com- Comput. Commun. Workshops (INFOCOM WKSHPS), Atlanta, GA,
munication channel models for realistic and efficient highway VANET USA, 2017, pp. 814–819.
simulation,” IEEE Trans. Veh. Technol., vol. 64, no. 1, pp. 248–262, [121] S. Xu, Y. Qian, and R. Q. Hu, “Data-driven edge intelligence for robust
Jan. 2015. network anomaly detection,” IEEE Trans. Netw. Sci. Eng., vol. 7, no. 3,
[99] J. Li, C. Natalino, D. P. Van, L. Wosinska, and J. Chen, “Resource pp. 1481–1492, Jul.–Sep. 2020.
management in fog-enhanced radio access network to support real-time [122] S. Xu, D. Fang, and H. Sharif, “Efficient network anomaly detection
vehicular services,” in Proc. IEEE 1st Int. Conf. Fog Edge Comput. for edge gateway defense in 5G,” in Proc. IEEE Global Commun. Conf.
(ICFEC), 2017, pp. 68–74. (GLOBECOM) Workshop, Waikoloa, HI, USA, Dec. 2019, pp. 1–5.
[100] J. Wang, B. He, J. Wang, and T. Li, “Intelligent VNFs selection based [123] S. Xu, Y. Qian, and R. Q. Hu, “A semi-supervised learning approach
on traffic identification in vehicular cloud networks,” IEEE Trans. Veh. for network anomaly detection in fog computing,” in Proc. IEEE Int.
Technol., vol. 68, no. 5, pp. 4140–4147, May 2019. Conf. Commun. (ICC), Shanghai, China, 2019, pp. 1–6.
[101] R. W. L. Coutinho and A. Boukerche, “Guidelines for the design of [124] S. Gyawali and Y. Qian, “Misbehavior detection using machine learn-
vehicular cloud infrastructures for connected autonomous vehicles,” ing in vehicular communication networks,” in Proc. IEEE Int. Conf.
IEEE Wireless Commun., vol. 26, no. 4, pp. 6–11, Aug. 2019. Commun. (ICC), Shanghai, China, 2019, pp. 1–6.
[102] A. Alamer, S. Basudan, and X. Lin, “A privacy-preserving incentive [125] S. Gyawali, Y. Qian, and R. Q. Hu, “Machine learning and reputation
framework for the vehicular cloud,” in Proc. IEEE Int. Conf. Internet based misbehavior detection in vehicular communication networks,”
Things (iThings) IEEE Green Comput. Commun. (GreenCom) IEEE IEEE Trans. Veh. Technol., vol. 69, no. 8, pp. 8871–8885, Aug. 2020.
Cyber Phys. Soc. Comput. (CPSCom) IEEE Smart Data (SmartData), [126] S. Gurugopinath, P. C. Sofotasios, Y. Al-Hammadi, and S. Muhaidat,
Halifax, NS, Canada, 2018, pp. 435–441. “Cache-aided non-orthogonal multiple access for 5G-enabled vehicular
[103] C. Lai, K. Zhang, N. Cheng, H. Li, and X. Shen, “SIRC: A networks,” IEEE Trans. Veh. Technol., vol. 68, no. 9, pp. 8359–8371,
secure incentive scheme for reliable cooperative downloading in high- Sep. 2019.
way VANETs,” IEEE Trans. Intell. Transp. Syst., vol. 18, no. 6, [127] S. Garg, K. Kaur, G. Kaddoum, S. H. Ahmed, and D. N. K. Jayakody,
pp. 1559–1574, Jun. 2017. “SDN-based secure and privacy-preserving scheme for vehicular
networks: A 5G perspective,” IEEE Trans. Veh. Technol., vol. 68, no. 9,
[104] J. Ni, X. Lin, K. Zhang, and X. Shen, “Privacy-preserving real-time
pp. 8421–8434, Sep. 2019.
navigation system using vehicular crowdsourcing,” in Proc. IEEE 84th
[128] H. Xu, M. Dong, K. Ota, J. Wu, and J. Li, “Toward software defined
Veh. Technol. Conf. (VTC-Fall), Montreal, QC, Canada, 2016, pp. 1–5.
dynamic defense as a service for 5G-enabled vehicular networks,”
[105] M. Agiwal, A. Roy, and N. Saxena, “Next generation 5G wireless
in Proc. Int. Conf. Internet Things (iThings) IEEE Green Comput.
networks: A comprehensive survey,” IEEE Commun. Surveys Tuts.,
Commun. (GreenCom) IEEE Cyber Phys. Soc. Comput. (CPSCom)
vol. 18, no. 3, pp. 1617—1655, 3rd Quart., 2016.
IEEE Smart Data (SmartData), Atlanta, GA, USA, 2019, pp. 880–887.
[106] X. Duan, Y. Liu, and X. Wang, “SDN enabled 5G-VANET: Adaptive
[129] I. Ahmad, S. Shahabuddin, T. Kumar, J. Okwuibe, A. Gurtov, and
vehicle clustering and beamformed transmission for aggregated traffic,”
M. Ylianttila, “Security for 5G and beyond,” IEEE Commun. Surveys
IEEE Commun. Mag., vol. 55, no. 7, pp. 120–127, Jul. 2017.
Tuts., vol. 21, no. 4, pp. 3682–3722, 4th Quart., 2019.
[107] N. Zhang, S. Zhang, P. Yang, O. Alhussein, W. Zhuang, and X. S. Shen,
“Software defined space-air-ground integrated vehicular networks:
Challenges and solutions,” IEEE Commun. Mag., vol. 55, no. 7, Sohan Gyawali (Member, IEEE) received the
pp. 101–109, Jul. 2017. Ph.D. degree in computer engineering from the
[108] X. Ge, Z. Li, and S. Li, “5G software defined vehicular networks,” University of Nebraska–Lincoln in 2020. He is cur-
IEEE Commun. Mag., vol. 55, no. 7, pp. 87–93, Jul. 2017. rently an Assistant Professor with the Department
[109] E. Coronado, S. N. Khan, and R. Riggio, “5G-EmPOWER: A software- of Computer Science, University of Texas Permian
defined networking platform for 5G radio access networks,” IEEE Basin. His research interests include misbehavior
Trans. Netw. Service Manag., vol. 16, no. 2, pp. 715–728, Jun. 2019. detection, cybersecurity, cellular based vehicular
[110] H. Li, M. Dong, and K. Ota, “Control plane optimization in software- communications, machine learning and intelligent
defined vehicular ad hoc networks,” IEEE Trans. Veh. Technol., vol. 65, transportation systems. He has served as a peer
no. 10, pp. 7895–7904, Oct. 2016. reviewer and a technical program committee mem-
[111] S. A. A. Shah, E. Ahmed, M. Imran, and S. Zeadally, “5G for vehicular ber for several top-tier journals and conferences. He
communications,” IEEE Commun. Mag., vol. 56, no. 1, pp. 111–117, is a member of the IEEE Young Professionals and IEEE communications
Jan. 2018. society.
[112] X. Huang, R. Yu, J. Kang, Y. He, and Y. Zhang, “Exploring mobile
edge computing for 5G-enabled software defined vehicular networks,” Shengjie Xu (Member, IEEE) received the M.S.
IEEE Wireless Commun., vol. 24, no. 6, pp. 55–63, Dec. 2017. degree in telecommunications from the University of
[113] G. Luo et al., “Cooperative vehicular content distribution in edge com- Pittsburgh, and the Ph.D. degree in computer engi-
puting assisted 5G-VANET,” China Commun., vol. 15, no. 7, pp. 1–17, neering from the University of Nebraska–Lincoln.
Jul. 2018. He is currently a Tenure-Track Assistant Professor
[114] C. Campolo, A. Molinaro, A. Iera, and F. Menichella, “5G network with the Beacom College of Computer and Cyber
slicing for vehicle-to-everything services,” IEEE Wireless Commun., Sciences, Dakota State University. His research
vol. 24, no. 6, pp. 38–45, Dec. 2017. interests include AI security, adversarial machine
[115] S. Zhang, W. Quan, J. Li, W. Shi, P. Yang, and X. Shen, “Air-ground learning, and intelligent networking system. He cur-
integrated vehicular network slicing with content pushing and caching,” rently serves as an editorial board member for
IEEE J. Sel. Areas Commun., vol. 36, no. 9, pp. 2114–2127, Sep. 2018. two international journals. He has served as a peer
[116] H. Zhou, W. Xu, Y. Bi, J. Chen, Q. Yu, and X. S. Shen, “Toward reviewer and a technical program committee member for several top-tier jour-
5G spectrum sharing for immersive-experience-driven vehicular com- nals and conferences. He is a member of IEEE Communication Society,
munications,” IEEE Wireless Commun., vol. 24, no. 6, pp. 30–37, IEEE Computer Society, and ACM. He is also a member of IEEE Technical
Dec. 2017. Committees, including TCCIS, TCGCC, and TCBD.
Yi Qian (Fellow, IEEE) received the Ph.D. degree Rose Qingyang Hu (Fellow, IEEE) received the
in electrical engineering from Clemson University, Ph.D. degree from the University of Kansas. She
Clemson, SC, USA. He is currently a Professor is a Professor with the Electrical and Computer
with the Department of Electrical and Computer Engineering Department and an Associate Dean
Engineering, University of Nebraska–Lincoln. His for Research with the College of Engineering,
research interests include communications and Utah State University, where she also directs the
systems, and information and communication Communications Network Innovation Lab. She has
network security. He was previously the Chair of published extensively in leading IEEE journals and
the IEEE Technical Committee for Communications conferences and also holds numerous patents in
and Information Security. He was the Technical her research areas. Her current research interests
Program Chair for IEEE International Conference on include next-generation wireless system design and
Communications in 2018. He serves on the Editorial Boards of several inter- optimization, Internet of Things, mobile edge computing, and artificial intel-
national journals and magazines, including as the Editor-in-Chief for IEEE ligence in wireless networks. Besides a decade academia experience, she
W IRELESS C OMMUNICATIONS. He was a Distinguished Lecturer for IEEE has more than 10 years of Research and Development experience with
Vehicular Technology Society. He is currently a Distinguished Lecturer for Nortel, Blackberry, and Intel, as a Technical Manager, a Senior Wireless
IEEE Communications Society. System Architect, and a Senior Research Scientist, actively participating
in industrial 3G/4G technology development, standardization. He is cur-
rently serving on the editorial boards of the IEEE T RANSACTIONS ON
W IRELESS C OMMUNICATIONS, the IEEE T RANSACTIONS ON V EHICULAR
T ECHNOLOGY, the IEEE Communications Magazine, and the IEEE
W IRELESS C OMMUNICATIONS. She also served as the TPC Co-Chair for
the IEEE ICC 2018. She is currently an IEEE Vehicular Technology Society
Distinguished Lecturer.

Challenges and Solutions For Cellular Based V2X Communications

Uploaded by

Copyright:

Available Formats

Challenges and Solutions For Cellular Based V2X Communications

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Challenges and Solutions For Cellular Based V2X Communications

Uploaded by

Copyright:

Available Formats

222 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 23, NO.

1, FIRST QUARTER 2021

Challenges and Solutions for Cellular

Fig. 1. Outline of this article.

V2X communications due to network backbone infrastructure

B. Cellular Network Based V2X Communications

information related to V2X and provides UE authorization sta-

Fig. 5. Operating scenarios for LTE V2X, described by [27].

Fig. 8. Side link establishment for synchronization, described by [34].

If vehicle densities are different for vehicles moving in

Fig. 11. 5G based vehicular communication model.

E. SDN Wi-Fi networks. Authors have done extensive simulations

2) Solutions: vehicular application. In addition, there should be a proper

R EFERENCES [26] R. Molina-Masegosa and J. Gozalvez, “LTE-V for sidelink 5G V2X

You might also like