Università degli studi Guglielmo Marconi

5G enabling technologies

Romeo Giuliano
[email protected]
Topics
 Massive MIMO Communications
 Millimeter-Wave Mobile Communications
 Non-Orthogonal Multiple Access (NOMA) for Future Radio Access
 New Multicarrier Modulations for 5G
 Generalized Frequency Division Multiplexing: A Flexible
 Multi-Carrier Waveform for 5G
 Spectrally Efficient Frequency Division Multiplexing for 5G
 Full-Duplex Wireless Communications for 5G
 Device-to-Device Communications over 5G Systems
 Conclusions

2
Enabling Technologies
 To provide specifics to support the
application requirements, 5G enabling Increase of
technologies are necessary for an ultra number of
broadband network. connected
 Envisaged technologies: devices
 MIMO, new radio waveforms
Enormous
 Advanced cancellation interference New services
techniques (e.g. ICIC, CoMP) traffic growth
 Techniques to improve the throughput
(e.g. carrier aggregation, dual
connectivity, simultaneous Tx-Rx)



Small cells, ultra dense network, cloud
RAN, Device-to-Device
Specific radio interface/architecture (from
5G
LTE-A) such as massive Machine Type
Communications (mMTC) and public
safety communications

3
Multiple Input Multiple Output (MIMO)
 Area throughput [bit/s/km2] = Bandwidth [Hz] Cell density [cells/km2]
Spectral efficiency [bit/s/Hz/cell]
 In previous network generations have greatly resulted from cell densification (see urban
environment) and allocation of more bandwidth
 MIMO improves the Spectral Efficiency of future networks

 Single Input Single Output capacity is according to Shannon formula (not

efficient)
 In the cellular system, Shannon capacity is multiplied by NBS.
 MIMO strategies:
 Single point-to-point MIMO
 Multi-User MIMO

SU-MIMO MU-MIMO 4
MIMO: basic principle
 Nt tx antennas, quasi-static channel (i.e. Tb ≪ Tcoh), Nr rx antennas
 H is the Nr × Nt channel matrix with whose entries hij are complex
channel gains (transfer functions) from the j-th transmit to the i-th
receive antenna.
 The received signal vector: r = Hs + n = x + n contains the signals
received by Nr antenna elements, where s is the transmit signal
vector and n is the noise vector.
 Consider a singular value decomposition of the channel: H = WLU†,
where L is a diagonal matrix containing singular values, and W and
U† are unitary matrices composed of the left and right singular
vectors, respectively.
 The received signal is: r = Hs + n = WLU†s + n
 Multiplication of the transmit data vector by matrix U and
the received signal vector by W† diagonalizes the channel:
W†r = W†WLU†Us + W†n; r’ = Ls +n’
 RH (rank of matrix H) parallel channels (eigenmodes of the
channel)  the capacity of parallel channels just adds up.
(.)† = ((.)T)*
5
Massive MIMO Communications
 Spatial multiplexing
 High-rate signal is split into multiple lower-rate streams and each stream is transmitted from a
different transmit antenna in the same frequency channel
 Diversity
 Selecting the best signal among the received ones
 Beamforming
 Single stream with a given phase, depending on the position of the reference user

 Massive MIMO is when the number of antennas are much greater than the sum of
the antennas of the receiving users: M ≫ K
 For Multi-User MIMO, the channel response for two users (i.e. h1, h2) becomes asymptotical
orthogonal: h1 h2 /M → 0 (BS can completely separate users’ signals)
 TDD to save pilot signals
 Challenges: pilot contamination in a multi-cell scenario, users have similar paths (low
scattering), pilot standardization

6
Millimeter-Wave Mobile Communications (mmWave)
 Complementary to the low frequency band  coexistence of low frequency
and high frequency communication (hybrid network)
 The conventional LTE is responsible for seamless coverage due to its longer range
 The mmWave communications are served as multiple hotspots scattered in the cell.
 Advantages:
 Potential availability of large amount of spectrum
 Large continuous spectrum chunks.
 Frequency and Channel propagation issues:
 Higher pathloss, high rain attenuation
 Reduced diffraction, higher attenuation from outdoor-to-indoor (e.g. walls)
 Smaller antenna size, use of beamforming
 Candidate frequencies: 6 GHz, 15 GHz (group 6-20 GHz), 28 GHz, 38GHz (group
20-50 GHz), 60 GHz, and E-band (71–76GHz, 81–86GHz) (group 50-100 GHz).

7
mmWave
 Use of mmWave for a unified access and backhaul (UAB) network
 Radio Access for small cells and hotspots
 Back-Hauling (narrow beam)
 The UAB network utilizes a C/U split configuration
 The control-plane (C-plane) is managed by macro base stations (MBs) through low
frequency bands
 The user plane (U-plane) uses a mmWave base stations (mBs) through high frequency
bands.
 Backhaul link from the mBs to the MBs
 Use of mmWave, to save fiber cost deployments
 Use of in-band backhaul (in LTE-A use of out-band Tx for relays), due to spectrum
availability
 Request to have a unified air interface  reuse of radio access parameters
of lower frequencies, adoption of configurable parameters for flexibility
8
mmWave
 Radio access and backhaul share a continuous bandwidth  Need of interference
mitigation between radio access and backhaul
 All mBs share a same resource partition ratio so as not to interfere with each other
 The ratios can also be dynamically adapted over time.
 Dynamic Resource Allocation
 Backhaul band (BBH) and Radio Access Band (BRA), B = BBH + BRA
 Backhaul Throughput of the n-th mBS = TBHn = RBH BBH, and radio access Throughput = TRAn
= RRAn BRA = RRAn (B – BBH), with RBHn and RRAn are the corresponding data rates
 For any mB, its throughput Tn = min[TBHn, TRAn], triangular function with BBH
 Strategies:
 Max-Min: Maximization of the minimum throughput
 Max-Sum: Maximization of the sum throughput
 Quasi-PF: Maximization of the satisfactory factor

9
Non-Orthogonal Multiple Access (NOMA) for Future Radio
Access
 Code Division Multiple Access (CDMA)
 An user is assigned a unique code sequence (spreading code) to enable multiple access 
every user increases the interference level (background noise), power control is mandatory
 Code sequences have to be chosen very carefully: good cross-correlation to reduce the
multiple access interference (MAI), good auto-correlation to reduce the inter symbol
interference (ISI).
 Large Area Synchronised CDMA
 Research efforts for spreading sequences, which benefit from zero correlation values for the
delay-induced code offset is in the so-called zero correlation zone (ZCZ) or interference free
window (IFW)
 The resultant LAS codes exhibit an IFW, where both the off-peak aperiodic auto-correlation
values and the aperiodic cross-correlation values become zero. Therefore this system has
zero ISI and zero MAI, as long as the time-offset of the codes is within the IFW.
 Disadvantages: limited number of IFW codes; auto-correlation and cross-correlation function
higher outside the IFW

10
Non-Orthogonal Multiple Access (NOMA)
 Multi-Carrier CDMA (MC-CDMA), a combination of OFDM and CDMA
 Spreads each user symbol in the frequency domain
 Advantages: decreasing the symbol rate  simpler synchronization
 In uplink, non-orthogonal users cause multi-user interference (MUI): need of
MultiUser detection  computationally complex
 Proposal of partition the available subcarriers into groups and distribute users
among the groups to reduce the MUD complexity: Group Orthogonal MC-
CDMA (GO-MC-CDMA)
 The users that are assigned subcarriers of the same group are separated using
spreading codes  no interference from other groups
 Each group has a smaller number of users, making the optimum MUD computationally
feasible within each group

11
Non-Orthogonal Multiple Access (NOMA)
 Low Density Signature CDMA (LDS-CDMA)
 Basic idea: to switch off a large number of chips for each symbol so the
signature matrix will be a sparse matrix  each user will spread its data over
a small number of chips
 Reduced processing gain AND reduced number of users at each received
chip K≪N  reduced receiver complexity
 Further reduction in MUD using grouped-based technique
 LDS-CDMA performance are near single-user performance even under a
load of 200%
 Drawback: in wideband channels, its performance degrades due to multipath
fading (ISI, i.e. more users will interfere in one chip)
 Proposal of LDS-OFDM, which introduces sparse code multiple access
(SCMA) for high order modulations
12
New Multicarrier Modulations for 5G
 Spread use of OFDM for 4G (not selected for 3G for computational
complexity). Complexity reduced with rectangular pulse, cyclic prefix and
advances in amplifiers  adoption of FFT
 In a multi-user context, need of time and frequency alignment to maintain the
orthogonality
 Specific 5G requirements are asynchronous access,
frequency offset compensation, coexistence issue,
per-user channel equalization
 Problem of the spectral containment
 Switching off subcarriers  not a good solution

13
New Multicarrier Modulations for 5G: pulse shaping
 Another solution: adopting a basic pulse whose spectral secondary lobes can
be neglected.
 Lower spectral lobes  longer time pulse  reduction in spectral efficiency
 In OFDM add of cyclic prefix to the basic pulse, MCP+M  (MCP+M)/r
 Two strategies:
 Filter bank (abandon the time limitation) [next slides]
 Pulse shaping (maintaining the time limitation of the
pulse)
 Pulse shaping: rectangular pulse is replaced by
 Hanning pulse, Raised Cosine
 Reduction in the spectral efficiency
 Computational complexity similar to the classic OFDM
transceiver.

14
New Multicarrier Modulations: Filter Bank Multi Carrier
 Filter Bank Multi Carrier (FBMC): a low-pass prototype filter with K M coefficients is shifted
on the frequency axis by integer multiples of the subcarrier spacing, leading to a set of M
subchannels.
 K emitted multicarrier symbols overlap
 The overlapping factor K determines the filter attenuation characteristics and, thus, the spectral
separation between adjacent users
 The cyclic-prefix is not necessary  equalization to mitigate the channel distortion  no delay in the
cyclic prefix
 Receiver scheme is more complex  feasible only for low overlapping factors (K=2, 3)
 Data mapping cannot be applied to FBMC, because of the filter overlapping
 The signals transmitted on the subcarriers k + 1 and k – 1 generate interference in the
subcarrier k. Two approaches:
 Offset-QAM modulation (i.e. FBMC-OQAM). The principle relies on the impulse responses of
interference filters that cross the zero axis at multiples of the symbol period M, but with a shift of M/2
between the real part and the imaginary part. Then, full rate is obtained if the symbol rate is doubled and
real and imaginary data samples alternate in the time domain and in the frequency domain;
 Pulse amplitude modulation (i.e. FBMC-PAM). The phase shift p/2 is applied to the subcarriers of
adjacent sub-channels and the transmitter input is a real sequence of data samples

15
New Multicarrier Modulations: Filter Bank Multi Carrier
 Transmission of purely imaginary symbol on a subcarrier and a purely real
symbol on the adjacent interfering subcarrier: iM/2 is the misalignment
 Further transformation for FBMC-PAM (by Malvar): good spectral
confinement even for K=2 (low delay)
FBMC-OQAM FBMC-PAM

16
New Multicarrier Modulations: Filter Bank Multi Carrier
 FBMC-OQAM
 Higher spectral containment (more significant for larger K)
 Greater robustness to the imperfections of the timing and Carrier Frequency Offset (CFO)
estimation procedures (more significant for larger K)
 Increase of the computational complexity
 Inefficiency in short-burst transmissions
 The FBMC-PAM
 Higher spectral containment and synchronization requirement
 Smaller complexity than FBMC-OQAM
 Marginal inefficiency in short-burst transmissions
  in many use cases, FBMC-PAM represents a good compromise
  only in extreme scenarios (spectrum fragmentation), FBMC-OQAM with
K=3 (or also with K = 4) may become necessary.
17
Generalized Frequency Division Multiplexing: A Flexible
Multi-Carrier Waveform for 5G
 In 5G need of:
 Loose synchronization for IoT
 Low latency for Tactile Internet
 Huge throughput (or bitpipe) applications
 High coverage and dynamic spectrum allocation for
regional/rural areas
 Adoption of MIMO and multi-antennas
 Not a specific waveform useful for all scenarios but…
 … a flexible waveform
 … reconfigurable for any applications
 … feasible to customize PHY such that it can be seen as a virtual service for
upper layers.

18
Generalized Frequency Division Multiplexing (GFDM)
 A data block of N = K M symbols transmitted on K subcarriers and M
subsymbols (slots)

19
GFDM: transmitter and receiver

N = 28, K = 4, M = 7

20
GFDM: reduction of out-of-band (OOB) emission
 Guard-symbol GFDM(GS-GFDM). First sub-symbol erased to zero.
 With Cyclic Prefix. Suitable for avoiding the multipath but it is necessary to
put the last subsymbol null
 Windowed-GFDM (W-GFDM)
 Reduction of efficiency
 Linear, raised-cosine (RC) filter

 Precoded WHT: walsh-hadamard (WH) matrix to combine transmitted

symbols in order to provide reliable and low latency communication in
frequency selective channel
 Orthogonal QAM: combination of real and imaginary symbols

21
GFDM: performance

22
GFDM: summary
 Number of subcarriers K and subsymbols M
 Properties of the basic pulse (pulse shaping, circularity, cyclic prefix)
  GFDM is valid for multi-service and multi-user systems

23
Spectrally Efficient Frequency Division Multiplexing (SEFDM)
for 5G
 Aim: optimising spectrum utilisation for higher data rates
 Contrary to OFDM, SEFDM deliberately and counterintuitively violates the
orthogonality rule defined for OFDM by reducing the spacing between the sub-
carriers.
 In SEFDM the sub-carrier spacing is equal to Df = a/T, (T symbol duration), a < 1 level of
bandwidth compression (a=1 for OFDM) SEFDM signal
 Sub-carriers are closer together  non-orthogonal
 self-created ICI

Correlation matrix

24
SEFDM: transceiver and demodulation/detection techniques
Demodulation and detection techniques
 1° stage: Demodulator
 2° stage: first Detector
 3° stage: second Detector

25
SEFDM: summary
 SEFDM increases multiplexing gain with respect to an OFDM
 Use of more sub-carriers in the default OFDM bandwidth
 Sub-carriers suffering from self-created ICI
 More complex signal processing at the receiver
 A linear increase in the number of sub-carriers results in a more than linear increase in
the complexity order of both the demodulation and detection stages
 Necessity of sophisticated detection algorithms
 Subsequently, these algorithms are optimised to make them better-suited for application
in the real world

26
Full-Duplex Wireless Communications for 5G
hI

 A system simultaneously transmitting and receiving signals over the same

frequency band is a fullduplex (FD) system
 Potential of doubling the capacity
 Beneficial to high layers, such as medium access (MAC) layer
 How much should the self-interference (SI) be cancelled?
 Maximum (BS or MS) transmit powers PTx = 24 dBm, the noise floor N=-100 dBm,
Tx-Rx path isolation =15dB SI = PTx – path_isolation – N = 109dB i.e. cancelled SI
 The received signal is known by the receiver. Why is self-interference hard to
subtract?
 The received version is a complicated distorted function of the original clean signal along with
noise
 Three types of sophisticated cancellation techniques: 1. propagation-domain; 2. analog-
circuit-domain; 3. digital-domain

27
Full-Duplex Wireless Communications: SI techniques
 Propagation-domain
 Electromagnetically isolate (in the wireless-propagation-domain) the transmit
chain from the receive chain. Proposed solutions:
 Increasing the Tx and Rx antenna space
 Placing absorptive shielding between them,
 Using different polarization (it transmits only horizontally polarized signals and receives
only vertically polarized signals)
 Using directional transmit and/or receive antennas (i.e., antennas with non-uniform
radiation/sensing patterns)

28
Full-Duplex Wireless Communications: SI techniques
 Analog-circuit-domain hI
 Aim: SI cancelation in the analog receive-chain circuitry before the ADC
 Idea: to model and predict the distortions caused by SI and compensate
the received signal accordingly in the analog domain. Three schemes:
 Processing of the self-interference signal prior the upconversion (premixer
cancelers)
 Processing of the self-interference signal after the upconversion (post-mixer
cancelers)
 A baseband analog canceler where the canceling signal is generated in
baseband and the cancelation occurs in the analog baseband
Baseband analog canceler
Premixer Post-mixer

29
Full-Duplex Wireless Communications: SI techniques
 Digital-domain
 It works after the ADC quantization
 It applies sophisticated DSP techniques
 Advantages: reduction of circuit complexity and power consumption
 Disadvantages: limitation in the maximum achievable cancelation
 Usually implemented after other SI techniques

30
Full-Duplex Wireless Communications: scenarios
 FD bidirectional communication  FD cooperative communications
 Spectral efficiency is doubled  A source node, a relay node, and a
destination node

31
Full-Duplex Wireless Communications: scenarios
 FD cognitive radio networks  Use of the “listen-before-talk”
 In CRNs, the secondary (unlicensed) strategy by SU. Two problems:
users (SUs) are allowed to share the  Tx time reduction due to sensing
licensed spectrum with the primary  Sensing accuracy impairment due to
(licensed) users (PUs) data transmission
 When SUs have FD capability, they can
perform simultaneous sensing and Tx

32
Device-to-Device Communications over 5G Systems
 Basic idea: Devices being close to each other can activate direct links and bypass
the base station (BS) or access point (AP) by either using cellular communications
resources or using alternative radio technologies such as Wi-Fi.
 Advantages:
 Improved spectral efficiency and radio resource utilization
 Improved link coverage and coverage extension
 Increasing throughput, enabling
high data rate with a low delay due Application scenarios
to the short range
 Cellular data traffic offloading
 Enabling energy savings
 Allowing communications when
the radio infrastructure is damaged
 Content sharing

33
Device-to-Device Communications
 Resource D2D communications management:
 Inband: transmitting on cellular spectrum
 Underlay inband D2D mode: spectrum resources shared between D2D and cellular
communications  mitigation techniques for coexistence
 Overlay inband D2D mode: resources are assigned to D2D communications
 Outband: unlicensed bands such as Wi-Fi
 Challenges:
 Device discovery procedures to detect the presence of other UEs in the neighborhood
 Link setup strategies to properly select the spectrum for D2D communication
 Interference avoidance mechanisms for the coexistence among D2D UEs with cellular
network
 New devices design

34
 D2D Communications: functionalities and architecture
 D2D communications have been introduced in LTE-A in Rel.12. Two main
functionalities:
 ProSe Discovery: the process to identify a UE in proximity of another, using E-UTRAN
 ProSe Communication: the communication between two UEs in proximity through an E-
UTRAN communication path. The communication path can for example be established
directly between the UEs (mode 2) or routed via local evolved-NodeB (eNB) (mode 1).
 ProSe-enabled UE: it is a UE that supports ProSe Discovery and/or ProSe
Communication.
Discovery Communication, mode 1 Communication, mode 2

35
D2D Communications: Uses Cases and Scenarios in Rel. 12
 Restricted/Open ProSe Discovery:
 A ProSe-enabled UE discovers another UE in proximity only if it has previously achieved the permission
(restricted). Example: a friend in a social network
 A ProSe-enabled UE is able to discover neighbor devices without the necessity of a permission (open).
Example: shop/restaurant advertisement
 Network ProSe Discovery: the Mobile Network Operator (MNO) verifies if a UE has the
permission to discover another UE and the proximity.
 Service Continuity Between Infrastructure and E-UTRA ProSe Communication Paths:
the operator should be able to dynamically control the proximity criteria (e.g., range, channel
conditions, achievable QoS) for switching between the two communication paths.
 ProSe-Assisted WLAN Direct Communications: ProSe-enabled UEs with WLAN
capability when they are in Wi-Fi Direct communications range
 ProSe Application Provided by the Third-Party Application Developer:
 APIs provided by the operator to third-party application developers which enable the user to use a wide
variety of new ProSe applications created by third-party developers.
 Need of mechanisms to identify, authenticate and authorize the third-party application to use ProSe
capability features.

36
D2D Communications: architecture
 Application servers (ProSe App Server)
 Applications in the UE (ProSe UEs App)
 ProSe Functions
 Seven new interfaces/reference points

EPC

37
D2D Communications: Application Scenarios for 5G
 Local Service
 Social apps, play games, …
 Local data transmission (spectrum efficiency)
 Local advertising service (e.g. discounts and commercial promotions, information about movies)
 Cellular traffic offloading AND in hotspot areas with caching
 Media content from close terminals
 Emergency Communications
 Ad hoc network can be set up based on multi-hop D2D to guarantee wireless communication between users during a natural
disaster
 Improving coverage in blind spot by terrain or buildings
 IoT Enhancement
 Vehicle-to-vehicle (V2V) communication, Vehicle-to-Infrastructure (V2I) or Internet of Vehicles (IoV)
 Low-cost terminals (typically in IoT) can access close special terminals in D2D modality instead of direct connections with BSs.
 Smart home managed by a D2D-based access modality instead of a small cell structure.
 Positioning aid in critical/indoor scenarios
 Pre-deployed terminals can detect the location of terminals to be localized, and support indoor positioning at a low cost in 5G
networks.
 Communication enhancements
 Paired users may directly (in D2D) exchange information about channel status, improving the multi-user MIMO, CoMP,...
 In mmWave, use of D2D to provide LOS for high frequencies

38
Solutions for …

Number increase
Increase of of
Throughput increase New service enabling
connected
numberdevices
of
• Massive MIMO (maggiore connected
• NOMA (applicabile a tipologie • D2D (per servizi di Public Safety,
efficienza spettrale) devices
d’utente molto diverse, con bassa Proximity Services, Vehicle-to-X)
• mmWave (maggiore banda), capacità di processing nel • FBMC (eliminazione del prefisso
Enormous ricevitore, con bassi requisiti di ciclico quindi riduzione della
• Full Duplex (raddoppio della New services
banda) traffic growthsincronizzazione) latenza (e.g. automotive),
• GFDM (flessibilità per supportare • FBMC (spettro molto confinato  aumento complessità)
varie forme d’onda, efficienza uso di parti di spettro non • GFDM (e.g. regional areas)
spettrale cioè maggiore bit rate) assegnabili se si usasse l’OFDM; • Full Duplex (e.g. relay, reti

5G
non serve una sincronizzazione cognitive)
• SEFDM (ottmizzazione dello
estrema tra gli utenti 
spettro)
possibilità di gestire migliaia di
utenti contemporaneamente)
• GFDM (richiesta bassa
sincronizzazione)
• D2D for IoT

39
Conclusions
 Several solution from 5G  Solutions to enhance the amount of
 Flexible air interface
device management
 NOMA (usable for different user types)
 Smart spectrum usage
 FBMC, GFDM
 Multi antenna Tx/Rx  confined spectrum  use of spectrum not
 Versatile network solutions assignable with OFDM;
 Loose synchronization is required 
 Solutions to increase the possibility to simultaneously manage
throughput thousand of devices without the access
control)
 Massive MIMO, mmWave  Solutions to enable new services
 GFDM (flexibility for several  D2D  Automotive, Public Safety, …
waveforms, higher spectral efficiency  Full Duplex  improve the spectrum
then bit rate) usage, relays, cognitive networks, …
 Full duplex (doubling the data rate)  Enhancement in IoT  novel
 SEFDM (spectrum optimization)
applications, …

40
References
 W. Xiang, K. Zheng, X. Shen, “5G Mobile Communications”, Springer, 2017
 Rysavy research, “Mobile Broadband Transformations LTE to 5G”, 5G
Americas, Aug. 2016.
 “5G Technology Evolution Recommendations”, 4G Americas, Oct. 2015
 G. Fettweis, “5G What will it be”, ICC 2013

41