IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING, VOL. 5, NO.

4, DECEMBER 2019 935

A Reliable Channel Estimation Scheme Using

Scattered Pilot Pattern for IEEE 802.22-Based
Mobile Communication System
Ruiting Ouyang , Takeshi Matsumura , Member, IEEE, Keiichi Mizutani , Member, IEEE,
and Hiroshi Harada, Member, IEEE

Abstract—Current vehicular communication systems are breakdowns of individual vehicles, the approach of the vehi-
disadvantageous in dealing with a full-range futuristic intelli- cle from a blind spot, and wide-area traffic conditions will be
gent transportation system (ITS) with regard to coverage and exchanged through the V2V (Vehicle-to-Vehicle) and/or V2I
cost. On the contrary, TV white-spaces have the potential to
provide a wide-coverage, large-capacity, and cost-effective con- (Vehicle-to-Infrastructure) communication.
nectivity for ITS because of the excellent propagation properties, There are two major standards for the vehicular communi-
and there are increasing demands for applying IEEE 802.22 to cation network around the world [3]: IEEE 802.11p [4] and
mobile communication system capable of a wide coverage of ARIB STD-T109 [5]. IEEE 802.11p is based on the WLAN
10–30 km. Meanwhile, since IEEE 802.22 was standardized as (Wireless Local Area Network) system and well-known
a fixed communication system, its scattered pilot pattern hinders
accurate channel estimation in a highly mobile environment due as DSRC/WAVE (Dedicated Short Range Communication/
to the Doppler shift and long-delay multipath fading. In this Wireless Access in Vehicular Environment). This system oper-
paper, we propose a reliable channel estimation scheme enabling ates at 5.9 GHz with a bandwidth of 75 MHz and 70 MHz
an IEEE 802.22-based mobile communication system in rural in the U.S. and Europe, respectively, and at 5.8 GHz with
areas without altering physical layer parameters. The proposed a bandwidth of 80 MHz in Japan. The V2V and V2I commu-
scheme effectively uses pilot symbols in adjacent subcarriers to
extend interpolation range and shorten interpolation interval and nications utilizing DSRC/WAVE offers a real-time information
offers high-accurate and stable channel estimation while main- exchange among communication devices, while its coverage is
taining flexibility in downstream/upstream ratio. As a simulation literally too short. To expand the coverage, a vehicular ad-hoc
result, our proposed scheme sufficiently achieves a required bit mode is supported. However, all vehicles need to be in the
error rate of 2 × 10−4 in a highly mobile long-delay multipath communication range of each other [6], and thus, the network
fading environment even in the presence of high carrier frequency
offset. is occasionally or frequently disconnected in rural areas or
during hours with low traffic volume such as late night even
Index Terms—Channel estimation, cognitive radio, in urban areas. As shown in Fig. 1(a), deployment of the RSUs
IEEE 802.22, TV white-spaces (TVWS), wireless regional
area network (WRAN). (Road Side Units) along the road side is one of the solutions to
complement the vehicular ad-hoc networks [7], while frequent
hand-overs pose destabilization of the network, and plenty of
I. I NTRODUCTION RSUs lead to high infrastructure cost. Besides DSRC/WAVE,
LONG with the increasingly complicated traffic envi- a 10 MHz bandwidth at 760 MHz is assigned to the ITS based
A ronment in the recent decades, a future-oriented ITS
(Intelligent Transportation System) which can solve traffic
on ARIB STD-T109 in Japan [5]. ARIB STD-T109 offers
a wider coverage of up to 900 m, while the allocated band-
problems such as traffic congestions and accidents has become width is only 9 MHz and, obviously, insufficient to cope with
an indispensable research topic [2]. In the futuristic trans- the annually increasing capacity demands and the diversifi-
portation system, the full-range of information such as sudden cation of the ITS services. Thus, it is necessary to develop
a vehicular communication system with wide coverage and
Manuscript received February 14, 2019; revised June 6, 2019; accepted sufficient operation bandwidth.
July 12, 2019. Date of publication July 23, 2019; date of current ver-
sion December 12, 2019. A part of this work was funded by ImPACT TVWS (TV White-Spaces) have the potential to provide
Program of Council for Science, Technology and Innovation (Cabinet Office, a wide-coverage, large-capacity, and cost-effective connectiv-
Government of Japan). This paper was presented in part at WPMC, Chiang ity for the ITS and to form advanced heterogeneous vehicular
Rai, Thailand, Nov. 2018 [1]. The associate editor coordinating the review
of this paper and approving it for publication was G. Ding. (Corresponding networks because of its superior propagation properties and
author: Ruiting Ouyang.) high spectrum availability [8]–[10]. IEEE 802.22 was stan-
R. Ouyang, K. Mizutani, and H. Harada are with the Graduate dardized as a fixed point-to-multipoint WRAN (Wireless
School of Informatics, Kyoto University, Kyoto 606-8501, Japan (e-mail:
[email protected]). Regional Area Network) system using TVWS [11], capable
T. Matsumura is with the Graduate School of Informatics, Kyoto University, of a wide communication range of 10–30 km and a large-
Kyoto 606-8501, Japan, and also with the Wireless Systems Laboratory, scale multiple access with up to 512 CPEs (Customer Premise
National Institute of Information and Communications Technology, Yokosuka
239-0847, Japan. Equipments) for one BS (Base Station). Also, recently reported
Digital Object Identifier 10.1109/TCCN.2019.2930594 compact and lightweight IEEE 802.22-based radio equipment
2332-7731 c 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on March 06,2025 at 12:56:35 UTC from IEEE Xplore. Restrictions apply.
 936 IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING, VOL. 5, NO. 4, DECEMBER 2019

TABLE I
S YSTEM PARAMETERS OF IEEE 802.22 [11]

parameters regarding the channel estimation are described. In

Section III, we apply some conventional channel estimation
schemes to IEEE 802.22-based mobile communication and
clarify their defects. In Section IV, we propose a reliable chan-
nel estimation and phase compensation scheme applicable to
the DS subframe and evaluate its performance by simulation.
In Section V, we apply a combined scheme with the proposed
channel estimation scheme and 2-branch diversity to the US
Fig. 1. Deployment scenarios for wide area mobile communication system. subframe and evaluate its performance. Finally, the paper is
concluded in Section VI.

II. IEEE 802.22 S YSTEM

demonstrates the practicality with high transmission power
IEEE 802.22 was standardized in June 2011 as the
and sufficient throughput [12]. If IEEE 802.22 can be applied
first WRAN standard supporting unlicensed operation in
to mobile communication as shown in Fig. 1(b) [13], [14],
the TVWS from 54 to 862 MHz with cognitive radio
vehicles widely scattered in rural areas acquire the ability to
technologies [11]. Table I shows system parameters of IEEE
connect a network directly and stably without RSUs, leading
802.22. This system supports fixed point-to-multipoint com-
to highly efficient V2I communication systems in a cost-
munication between one BS and up to 512 CPEs. Channel
effective manner. However, the scattered pilot pattern of IEEE
bandwidths of 6, 7, and 8 MHz are supported to allow a global
802.22 in both time and frequency domains obstructs accurate
use in various countries. The PHY parameters are designed to
channel estimation in a highly mobile long-delay multipath
enable a transmission distance of up to 30 km, and its coverage
fading environment unlike the comb pilot pattern of IEEE
can be expanded to 100 km with an appropriate schedul-
802.11p. In addition, IEEE 802.22 supports flexible DS/US
ing at BS. To support such wide coverage, −94.5 dBm and
(Downstream/Upstream) ratio, and there are some cases in
−91.3 dBm are required as the minimum receiver sensitivity
which only one pilot symbol is allocated in the time domain for
to achieve a BER of less than 2 × 10−4 at BS and CPE,
the US subframe because the US subframes has no preambles.
respectively, for QPSK with a code rate of 1/2 in the case of
In this paper, we propose a reliable channel estimation
a 6 MHz channel bandwidth.
scheme enabling an IEEE 802.22-based wide area mobile com-
munication system. The proposed channel estimation scheme
effectively utilizes the pilot symbols in adjacent subcarriers to A. Frame Structure
extend interpolation range and shorten interpolation interval, In the normal mode, one superframe consists of
leading to sufficient communication quality. The main contri- 16 frames [11]. The first frame consists of a superframe
butions of our work are: 1) adhering to the PHY (Physical preamble, a frame preamble, an SCH (Superframe Control
layer) parameters defined in the standard and enabling easy- Header), an FCH (Frame Control Header), and the data pay-
implementation in any IEEE 802.22-based systems, 2) provid- load. The following 15 frames consist of a frame preamble, an
ing accurate, stable, and flexible channel estimation regardless FCH, and the data payload. The superframe preamble used for
of DS/US ratio even in a highly mobile long-delay multipath coarse frequency synchronization and signal detection is com-
fading environment in the presence of CFO (Carrier Frequency posed of four repetitions of STS (Short Training Sequence)
Offset). As a simulation result, we achieve a required BER (Bit and CP (Cyclic Prefix) in one OFDM (Orthogonal Frequency
Error Rate) of 2 ×10−4 for QPSK and 16QAM in a long- Division Multiplexing) symbol duration. Here, CP is equiv-
delay multipath environment at a velocity of 80 km/h and alent to STS. The frame preamble used for fine frequency
demonstrate the feasibility of the IEEE 802.22-based wide synchronization and channel estimation is composed of two
area mobile communication system. repetitions of LTS (Long Training Sequence) and CP in also
The rest of this paper is organized as follows. In Section II, one OFDM symbol duration. Here, CP is equivalent to the
general specifications of IEEE 802.22 and important system second half of LTS.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on March 06,2025 at 12:56:35 UTC from IEEE Xplore. Restrictions apply.
 OUYANG et al.: RELIABLE CHANNEL ESTIMATION SCHEME USING SCATTERED PILOT PATTERN 937

TABLE II TABLE III

C OMBINATIONS OF THE N UMBER OF OFDM S YMBOLS FOR DS AND US IEEE 802.22 P ROFILE A C HANNEL M ODEL [13]
S UBFRAMES W ITH 6 MHz BANDWIDTH AND 1/4 CP L ENGTH

variations occur in both time and frequency domains due to

the multipath fading and large Doppler shift.

D. Carrier Frequency Offset and Synchronization

In addition to the channel variations caused by the multipath
fading and Doppler shift, CFO (Carrier Frequency Offset)
caused by frequency mismatch between BS and CPE induces
a phase rotation and ICI (Intercarrier Interference) and impairs
channel estimation accuracy. In IEEE 802.22, a carrier
frequency accuracy within ±2 ppm is required at BS, and CPE
should be synchronized with the BS carrier frequency with an
accuracy of less than 2 % of the subcarrier spacing under the
static conditions. Specifically, in the case of a 6 MHz chan-
nel bandwidth, the subcarrier spacing is 3.347 kHz, and thus,
a maximal residual CFO of about 67 Hz must be factored in
as an additional phase rotation.
Fig. 2. Pilot pattern of IEEE 802.22 [11].
Here, we examine how much residual CFO is fluctu-
ated in a highly mobile multipath fading environment. As
above-mentioned, five repetitions of STS (including CP) in
In the frequency domain, 60 subchannels are assigned for
the superframe preamble are used for coarse frequency syn-
each frame. In the time domain, each frame is composed
chronization. Thus, the transmitted sample in the superframe
of 26–41 OFDM symbols according to the CP length and
preamble is expressed by
channel bandwidth and divided into DS and US subframes.
With a 6 MHz bandwidth and a CP length of 1/4, the num- ss [n] = ss [n + pN ], (1)
ber of OFDM symbols per frame is 26 in each superframe,
and there are five combinations of the number of OFDM where ss [n] is the n-th transmitted sample, N = 512 is the STS
symbols for DS and US subframes as shown in Table II to length, and 0 ≤ p ≤ 4. In an ideal AWGN (Additive White
support various asymmetric throughput requirements between Gaussian Noise) channel (i.e., no multipath fading), received
DS and US. Note that because of the limitation of the samples of the superframe preamble rs [n] can be expressed by
BCC (Binary Convolutional Code) block size, the number of rs [n] = ss [n]e j2πnfCFO tsp + z [n], (2)
payload symbols in the US subframe must be even.
where fCFO indicates CFO between transmitter and receiver,
tsp indicates a time interval between samples, and z[n] is
B. Pilot Pattern for Channel Estimation
AWGN. In a multipath fading environment, rs [n] can be
Fig. 2 shows the pilot repetition pattern of IEEE 802.22 [11]. expressed by
Pilot symbols are sparsely allocated in every seven symbols
and every seven subcarriers, and each symbol and subcarrier rs [n] = h[n]ss [n]e j2πnfCFO tsp + z [n], (3)
has only one pilot symbol. This pilot repetition pattern is con- where h[n] indicates channel variations caused by long-delay
sistent regardless of US or DS subframe but changed at BS multipath fading. Since the superframe preamble has the peri-
receivers because the pilot symbols are inserted before the odicity in (1), CFO can be coarsely estimated by calculating
interleaving process for US. the autocorrelation of the adjacent STS sequences as follows:
1
C. Channel Model fˆCFO_S = ·
2πNtsp
Table III shows the IEEE 802.22 Profile A channel model ⎧ ⎫
⎨3 N ⎬
consisting of six paths with a maximum delay of 21 μs, assum- arg rs∗ [n + pN ]rs [n + (p + 1)N ] , (4)
ing a wide coverage of 30 km in rural areas [15]. In this model, ⎩ ⎭
p=0 n=1
a frequency-selective fading occurs in wideband communica-
tion systems such as IEEE 802.22 because the delay dispersion where fˆCFO_S is a coarsely estimated frequency offset through
is sufficiently longer than the wavelength. In addition, when STS. Here, arg{·} and {·}∗ indicate phase and complex con-
applying IEEE 802.22 to mobile communication, the channel jugate of a complex number, respectively. Then, CFO is

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on March 06,2025 at 12:56:35 UTC from IEEE Xplore. Restrictions apply.
 938 IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING, VOL. 5, NO. 4, DECEMBER 2019

Fig. 3. Residual CFO distribution after frequency synchronization in AWGN

channel at rest and in IEEE 802.22 Profile A channel at 80 km/h.
Fig. 4. Transmitter and receiver architecture.

compensated by fˆCFO_S as follows: TABLE IV

S IMULATION PARAMETERS FOR CFO E STIMATION
ˆ
rL [n] = rL [n]e −j2πn f CFO_S tsp , (5)
where rL [n] and rL [n] indicate received samples and
compensated samples of the frame preamble, respectively.
Subsequently, LTS in the frame preamble is used for fine CFO
estimation by calculating the autocorrelation of the adjacent
LTS sequences as follows:
L

1
fˆCFO_L = arg rL ∗ [l ]rL [l + L] , (6)
2πLtsp
l=1
where l is the sample index in LTS, and L = 1,024 is the LTS
length. Finally, the estimated CFO fˆCFO is given by [16]
fˆCFO = fˆCFO_S + fˆCFO_L . (7)
The accuracy of the estimated CFO was evaluated by computer
simulation by assuming that both BS and CPE have the largest
but opposite-sign frequency deviations. (i.e., the maximum
transmitting binary data are firstly randomized by a scram-
CFO between BS and CPE comes to ±4 ppm.) Simulation
bler. Then, the binary data segmented into MCS (Modulation
parameters are summarized in Table IV. Fig. 3 shows his-
togram plots of the residual CFO (i.e., fCFO − fˆCFO ) after the
and Coding Scheme)-dependent blocks are channel-encoded
and modulated. After the modulation, the data are mapped
synchronization process in the AWGN channel at rest (0 km/h)
to designated subcarriers and converted from the frequency
and in the IEEE 802.22 Profile A channel (i.e., multipath
domain to the time domain (i.e., parallel-to-serial conversion)
fading environment) at a velocity of 80 km/h, respectively.
by the IFFT (Inverse Fast Fourier Transform). Finally, CP is
Here, the bin width of both histograms is 1 Hz. As shown
prefixed as a guard interval. At the receiver, frequency of the
in Fig. 3(a), the residual CFO is trivial within ±7 Hz in the
local oscillator is firstly locked by the synchronization process.
AWGN channel (without multipath fading). Meanwhile, high
Then, the processes of the transmitter are performed in reverse,
velocity and multipath fading degrade the accuracy of the
while the channel estimation and equalization are performed
estimated CFO, and the residual CFO occasionally exceeds
before the subcarrier demapping. Here, the residual CFO is
±67 Hz as shown in Fig. 3(b). However, the standard devi-
compensated by the channel estimation process.
ation of the residual CFO is about 17.4 Hz, and more than
99.7 % are within ±67 Hz according to the 3-sigma rule even
in a multipath fading environment at a velocity of 80 km/h. III. C ONVENTIONAL C HANNEL E STIMATION S CHEMES
From this result, a residual CFO of 67 Hz can be applied as IEEE 802.22 is originally standardized for the fixed commu-
the worst case for evaluation in the following sections. nication, in which channels ideally keep invariant in the time
domain due to no Doppler shift. Oppositely, for the mobile
E. Transmitter and Receiver Architecture communication, channel fluctuates in the time domain even in
Figs. 4(a) and 4(b) show the transmitter and receiver block the same subcarrier due to the Doppler shift. In this section,
diagrams of IEEE 802.22, respectively. At the transmitter, we overview some conventional channel estimation schemes

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on March 06,2025 at 12:56:35 UTC from IEEE Xplore. Restrictions apply.
 OUYANG et al.: RELIABLE CHANNEL ESTIMATION SCHEME USING SCATTERED PILOT PATTERN 939

for V2V and other wireless communication systems. Then, we

apply conventional channel estimation schemes to the IEEE
802.22-based mobile communication system and clarify their
defects by simulation.

A. Conventional Channel Estimation Schemes

Channel estimation is an indispensable process at receivers
in any wireless communication system for characterizing
channels and removing channel variations from the received
signals. In this sub-section, some conventional channel esti-
mation schemes for IEEE 802.11p, IEEE 802.16, and LTE
systems are briefly introduced and compared.
IEEE 802.11p is based on IEEE 802.11a, and a comb-type
pilot pattern is adopted [4]. To support mobile communication
(i.e., V2V and V2I communication) with IEEE 802.11p, some
channel estimation techniques are proposed. The preamble-
based CDP (Constructed Data Pilots) channel estimation
scheme offers high performance in time-variant V2V channels
by using the high correlation characteristics between adjacent
two symbols in the time domain [17]. However, the chan-
nel estimation accuracy is significantly degraded in a strong
multipath fading environment. Another preamble-based chan-
nel estimation technique, which updates channel variations
estimated at the preamble by generating a replica of recovered
OFDM data symbol [18]–[20], can improve the channel esti-
mation accuracy even with a scattered pilot pattern. However,
the feedback circuit for the replica generation requires re- Fig. 5. Interpolation and extrapolation channel estimation scheme.
encoding, re-interleaving, and serial-to-parallel conversion
processes and significantly increases the complexity of the
receiver. where fd,max and τrms denote the maximum Doppler
In the downlink of LTE and IEEE 802.16e, a conven- frequency and rms delay spread, respectively. Here, τrms is
tional linear interpolation and extrapolation scheme is widely about 2.77 μs for the IEEE 802.22 Profile A channel model in
applied [21], [22]. Firstly, channel variations of the received Table III, and fd,max is about 35.06 Hz for a 80 km/h velocity
pilot symbols are estimated using the predefined transmitted at 473 MHz. Thus, Tc,50% comes to about 12.1 ms, and corre-
pilot symbols. Then, channel variations of the data symbols sponding 32 symbols can be considered coherent for a 1/4 CP
between pilot symbols in the time domain are estimated by length and a 373 μs symbol duration. Meanwhile, BW c,50%
interpolation and extrapolation. Finally, channel variations of comes to about 72.3 kHz, and corresponding 21 subcarriers
the remaining data symbols within the estimated data symbols with a subcarrier spacing of 3.347 kHz can be considered
(including the pilot symbols) are interpolated and extrapolated coherent. This means that the channel variations in the time
in the frequency domain. This interpolation and extrapolation domain are relatively modest compared to the frequency
scheme offers low computational complexity and relatively domain. Thus, interpolation and extrapolation are preferably
accurate channel estimation in conventional cellular systems. applied in the time domain by referring the channel variations
In the following sub-section, we introduce a polar-linear inter- of the pilot symbols and the frame preamble.
polation and extrapolation scheme applicable to the IEEE Figs. 5(a) and 5(b) depict interpolation and extrapolation
802.22-based mobile communication system. channel estimation schemes applicable to the DS and US sub-
frames of the IEEE 802.22 system with 14 OFDM symbols,
B. Polar-Linear Interpolation and Extrapolation Channel respectively. (i.e., Case 4 and Case 2 in Table II are applied
Estimation Scheme for IEEE 802.22-Based Mobile System to DS and US subframes, respectively, and two pilot symbols
exist in each subcarrier.) Here, the superframe preamble is not
In a highly mobile long-delay multipath fading environment,
shown in Fig. 5(a) because it is not used for the channel esti-
frequency-selective fading and the Doppler shift impair the
mation. Therefore, Symbol 0 indicates the frame preamble in
similarity of channel variations between the pilot symbols. The
the DS subframe. The process detail is described as below:
coherence time (Tc ) and the coherence bandwidth (BWc ) with
1) Firstly, for the DS subframe, channel variations between
a 50 % correlation are given by [23]
the pilot symbols in the frame preamble are linearly inter-
0.423 polated in the frequency domain. Here, a typical polar-linear
Tc,50% ≈ , (8)
fd,max interpolation shown in Fig. 6 [24] is applied and amplitude
1 variations and phase rotations are separately interpolated and
BW c,50% ≈ , (9)
5τrms extrapolated. By representing channel variation of the pilot

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on March 06,2025 at 12:56:35 UTC from IEEE Xplore. Restrictions apply.
 940 IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING, VOL. 5, NO. 4, DECEMBER 2019

symbol in the α-th symbol of the β-th subcarrier as hα,β ,

the absolute amplitude variation of the symbol between pilot
symbols in the frame preamble |h0,β | can be estimated by
|h0,2k −1 | + |h0,2k +1 |
|h0,2k | = , for 0 < 2k < 840,
2
(10)
|h0,2k | + |h0,2k +2 |
|h0,2k +1 | = , for 840 < 2k < 1680,
2
(11)
|h0,842 | − |h0,839 |
|h0,840 | = + |h0,839 |, (12)
4 Fig. 6. Polar-linear interpolation and extrapolation scheme.
|h0,842 | − |h0,839 |
h0,841 = h0,842 − . (13)
4
Here, the symbol index α = 0 represents the frame pream- where w represents the index of data symbols within the
ble. Also, the phase rotation of the symbol between pilot extrapolation range.
symbols in the frame preamble θ0,β (= arg{h0,β } ) can be
estimated by
C. Performance Evaluation
θ0,2k −1 + θ0,2k +1
θ0,2k = , for 0 < 2k < 840, (14) We evaluated the performance of the conventional chan-
2
θ0,2k + θ0,2k +2 nel estimation schemes with the simulation of BER char-
θ0,2k +1 = , for 840 < 2k < 1680, acteristics in the case of 14 OFDM symbols shown in
2
(15) Figs. 5(a) and 5(b). Here, the preamble-based channel estima-
θ0,842 − θ0,839 tion scheme [17], [18] was applied to the DS subframe only
θ0,840 = + θ0,839 , (16) due to no frame preamble in the US subframe, and the typical
4
θ0,842 − θ0,839 polar-linear interpolation and extrapolation channel estimation
θ0,841 = θ0,842 − . (17) scheme was applied to DS and US subframes. Table V shows
4
simulation parameters, and Fig. 7 shows BER characteristics
Thus, the estimated channel variation of the symbol between
as a function of Eb /N0 in a long-delay multipath fading envi-
pilot symbols in the frame preamble h0,β is calculated by
ronment with a CPE velocity of 80 km/h and a residual CFO of
h0,β = h0,β cosθ0,β + j h0,β sinθ0,β . (18) 67Hz. Here, Scheme 1, 2, and 3 indicate the CDP channel esti-
mation scheme in [17], the preamble-based channel estimation
Note that this process is not performed for the US subframe
scheme in [18], and the typical polar-linear interpolation and
due to no frame preamble.
extrapolation channel estimation scheme, respectively. Here,
2) Secondly, for channel variations of data symbols between
a BER of 2 × 10−4 is used as a metric in accordance with
the frame preamble and the first pilot symbol and between the
the target BER for the receiver sensitivity as explained in
first and second pilot symbols, the polar-linear interpolation
Section II. From the results, Scheme 1 and 2 show much worse
scheme is applied in the time domain as follows:
performance than Scheme 3 since the preamble-based chan-
hp0 ,β − h0,β nel estimation schemes severely accumulate estimation error
h̃u,β = · u + h0,β , (19)
p0 with longer frame length. Besides, Scheme 1 and 2 require
hp1 ,β − hp0 ,β a frame preamble and cannot be applied to the US subframe.
h̃v ,β = · (v − p0 ) + hp0 ,β , (20) Thus, preamble-based schemes are not suitable for the IEEE
p1 − p0
802.22-based system.
θp0 ,β − θ0,β Even though Scheme 3 achieves the best performance,
θu,β = · u + θ0,β , (21)
p0 a required BER of 2 × 10−4 was not achieved for neither DS
θp1 ,β − θp0 ,β nor US even in the case that Eb /N0 = 30 dB. Fig. 8 shows
θv ,β = · (v − p0 ) + θp0 ,β , (22)
p1 − p0 individual BER of each OFDM symbols for QPSK at a veloc-
where u and v denote the index of the data symbols within ity of 80 km/h with a residual CFO of 67 Hz in the case
the interpolation interval, and p0 and p1 denote the index of that Eb /N0 = 30 dB. Obviously, almost all symbols exceeded
the pilot symbols in the time domain (e.g., in the case of the a required BER of 2 × 10−4 , and the BER performance on
first subcarrier of the DS subframe, p0 = 1 and p1 = 8). the far side of the extrapolation range significantly became
After that, the rest of the data symbols are extrapolated in worse. Thus, it is essential to reduce the number of error bits
the time domain as follows: in both interpolation and extrapolation ranges to achieve the
required BER. Furthermore, the typical polar-linear interpola-
hp1 ,β − hp0 ,β tion and extrapolation scheme needs at least two pilot symbols
h̃w ,β = · (w − p1 ) + hp1 ,β , (23)
p1 − p0 in the time domain. Therefore, the US subframe requires at
θp1 ,β − θp0 ,β least 14 OFDM symbols, and the flexibility in the DS/US ratio
θw ,β = · (w − p1 ) + θp1 ,β , (24)
p1 − p0 is considerably limited.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on March 06,2025 at 12:56:35 UTC from IEEE Xplore. Restrictions apply.
 OUYANG et al.: RELIABLE CHANNEL ESTIMATION SCHEME USING SCATTERED PILOT PATTERN 941

TABLE V
S IMULATION PARAMETERS FOR BER P ERFORMANCE E VALUATION

Fig. 8. Individual BER in each OFDM symbol with typical polar-linear

interpolation and extrapolation channel estimation scheme.

For the IEEE 802.22 Profile A channel model, BWc,90%

comes to 7.2 kHz that is approximately twice the subcarrier
spacing. This means that the channel variations of adjacent
subcarriers are substantially coherent, and the interpolation
range can be extended by effectively utilizing the pilot symbols
in adjacent subcarriers. In other words, a polar-linear interpo-
lation process can be applied to some data symbols originally
placed in the extrapolation range by utilizing the channel vari-
ations of the pilot symbols in adjacent subcarriers. Besides,
the interpolation interval can be shortened with the assistance
of the pilot symbols in adjacent subcarriers for reducing the
number of errors.
In addition, since the estimation accuracy of the symbols on
the far side of the extrapolation range is considerably inferior
to that of the inner symbols as shown in Fig. 8, we introduce
an extra phase compensation technique to reduce the number
of error bits of the farthest and second farthest symbols. In
the case of higher residual CFO (e.g., 67 Hz at 473 MHz),
the phase rotation between the OFDM symbols are dominated
by residual CFO rather than multipath fading. Since CFO lin-
Fig. 7. BER characteristics with preamble-based channel estimation scheme early rotates the phase in the time domain unlike the multipath
and typical polar-linear interpolation and extrapolation channel estimation
scheme.
fading, the additional phase compensation technique simply
estimates the phase rotation of the farther symbol by adding
the phase rotation caused by residual CFO to the estimated
IV. P ROPOSED C HANNEL E STIMATION AND P HASE phase rotation of the symbol placed in front.
C OMPENSATION S CHEME FOR DS 2) Algorithm: Fig. 9 shows a diagram of the proposed pilot
We propose a reliable and flexible channel estimation all-copied channel estimation scheme with an extra phase com-
scheme offering high estimation accuracy with the assistance pensation technique for the DS subframe with 14 OFDM sym-
of pilot symbols in adjacent subcarriers. Specifically, by effec- bols. Firstly, the channel variations between pilot symbols in
tively utilizing channel variations of the pilot symbols in the frame preamble are polar-linearly interpolated (Fig. 9(a)).
adjacent subcarriers, the proposed scheme extends the inter- Then, channel variations are copied from pilot symbols in adja-
polation range and shortens the interpolation interval. In this cent subcarriers (Fig. 9(b)). Hereinafter, data symbols with
section, we first describe a pilot all-copied channel estima- the channel variations copied from pilot symbols in adjacent
tion scheme utilizing all pilot symbols in adjacent subcarriers. subcarriers are referred as to sample symbols.
Then, we also propose a simplified pilot partly-copied chan- To reduce the extrapolation range and shorten the interpola-
nel estimation scheme enabling the reduction of the number tion interval to the utmost extent, all pilot symbols in adjacent
of utilized pilot symbols. subcarriers are utilized. For example, in the case of the extrap-
olation range of Subcarrier 4, the channel variation of the pilot
symbol (Symbol 14) in Subcarrier 3 is copied to Subcarrier 4
A. Pilot All-Copied Channel Estimation Scheme
to reduce the extrapolation range (extrapolation range is com-
1) Principle: The coherence bandwidth (BWc ) with a 90 % pletely eliminated in this case), then the channel variation of
correlation is expressed as follows [23]: the pilot symbol (Symbol 12) in Subcarrier 5 is also copied to
1 Subcarrier 4 to shorten the interpolation interval between the
BW c,90% ≈ . (25)
50τrms pilot symbol (Symbol 9) and the sample symbol (Symbol 14).

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on March 06,2025 at 12:56:35 UTC from IEEE Xplore. Restrictions apply.
 942 IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING, VOL. 5, NO. 4, DECEMBER 2019

Finally, phase rotation of the farthest and second farthest

data symbols in the extrapolation range are compensated to
further improve the channel estimation accuracy (Fig. 9(d)).
The phase rotations of these data symbols are obtained by
simply adding the phase rotation calculated from the residual
CFO to the estimated phase rotation of the data symbols placed
in front. When the second farthest data symbol is not a pilot
symbol or sample symbol, phase rotations of the farthest and
the second farthest data symbols of the β-th subcarrier are
calculated by
θG−1,β = θG−2,β + θCFO , (26)
θG,β = θG−2,β + 2θCFO , (27)
where G is the number of OFDM symbols except for the
superframe preamble and the frame preamble in the DS sub-
frame (i.e., G = 14 in Fig. 9), θα,β is the estimated phase
rotation of the α-th symbol of the β-th subcarrier after the
polar-linear interpolation and extrapolation, and θCFO is the
estimated phase rotation between adjacent symbols caused by
CFO. Note that θCFO is calculated from the difference of the
average phase rotation of the pilot symbols between the frame
preamble and the first OFDM symbol as follows:
Q−1  
q=0 θ 1,7q+1 − θ 0,7q+1
θCFO = , (28)
Q
where Q = 240 is the number of pilot symbols in each OFDM
symbol, 7q+1 is the index of the pilot symbol in the first
OFDM symbol as shown in Fig. 9, θ1,7q+1 is the phase rota-
tion of the corresponding pilot symbols, and θ0,7q+1 is the
estimated phase rotation of the symbols in the frame preamble
after the interpolation process.
When the second farthest data symbol is a pilot or sam-
ple symbol, the phase rotation of the farthest data symbols is
obtained by
θG,β = θG−1,β + θCFO . (29)
3) Performance Evaluation: We applied the pilot all-copied
channel estimation scheme to the DS subframe with 14 OFDM
symbols (i.e., Case 4 in Table II) and evaluated BER charac-
teristics with the parameters shown in Table V. Fig. 10 shows
BER characteristics as a function of Eb /N0 . As a result, with
a residual CFO of 67 Hz, a BER of 2×10−4 was achieved for
QPSK and 16QAM, and an error floor for 64QAM was sig-
nificantly improved to less than 1 × 10−3 compared to Fig. 7.
Fig. 9. Diagram of pilot all-copied channel estimation scheme for DS
subframe.
These results indicate that the pilot all-copied channel esti-
mation scheme sufficiently enhances the tolerance against the
channel variations in a highly mobile long-delay multipath
fading environment.
In addition, the copying process is also applied to all inter-
polation ranges to prevent an interpolation interval from being
long (Fig. 9(b)). For example, in the case of Subcarrier 5, B. Pilot Partly-Copied Channel Estimation Scheme
the channel variations of both pilot symbols in adjacent sub- 1) Algorithm: To simplify the complexity of the algorithm
carriers (i.e., Symbol 9 in Subcarrier 4 and Symbol 10 in and to mitigate the burdens of the receiver, we also propose
Subcarrier 6) are copied to Subcarrier 5 as sample symbols. a pilot partly-copied channel estimation scheme that reduces
After the copying process, channel variations of all the remain- the number of sample symbols. The principle is same as the
ing data symbols are polar-linearly interpolated or extrapolated pilot all-copied channel estimation scheme, and the extra phase
(Fig. 9(c)). compensation technique is also applied.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on March 06,2025 at 12:56:35 UTC from IEEE Xplore. Restrictions apply.
 OUYANG et al.: RELIABLE CHANNEL ESTIMATION SCHEME USING SCATTERED PILOT PATTERN 943

As a result, we obtained almost same BER performance as

the pilot all-copied channel estimation scheme, and a required
BER of 2 × 10−4 was achieved for QPSK regardless of the
number of OFDM symbols. In addition, when the number of
OFDM symbols is larger than 8, the required BER of 2×10−4
was achieved for 16QAM.
For the comprehensive evaluation, we also simulated BER
performance by taking into account the frequency synchro-
nization described in Section II-E. Fig. 13 shows BER char-
acteristics of the DS subframe as a function of Eb /N0 with
various numbers of OFDM symbols. Even with a maximal
CFO of 1,892 Hz, a required BER of 2 × 10−4 was also
achieved for QPSK and 16QAM regardless of the number
of OFDM symbols by applying the frequency synchroniza-
tion. From these results, we confirmed that the proposed pilot
Fig. 10. BER characteristics of DS subframe using pilot all-copied channel partly-copied channel estimation scheme has the reliability and
estimation scheme.
flexibility to support various numbers of OFDM symbols in the
DS subframe in a highly mobile long-delay multipath fading
environment.
Figs. 11 shows a diagram and a flowchart of the pilot
partly-copied channel estimation scheme applicable to the DS C. Complexity Analysis
subframe with 14 OFDM symbols. Firstly, channel variations
between pilot symbols in the frame preamble are linearly inter- Approximate computational complexity of two proposed
polated (Fig. 11(a)-(1)). Then, the copying process of channel channel estimation schemes for the DS subframe are com-
variations is applied to the subcarriers where three or more pared in Table VI. Here, the number of copying operations
data symbols are placed in the extrapolation range for expand- and subtraction and division operations in one DS subframe
ing the interpolation range if possible (Fig. 11(a)-(2)). Here, are considered. Note that i represents the index of symbol in
to reduce the number of sample symbols but to maximize each repetition of pilot pattern, and ∗ represents the mini-
the interpolation range, the proposed scheme utilizes only mum integer larger or equal to ∗. Since the number of pilot
the farther pilot symbol when both adjacent subcarriers have symbols in each OFDM symbol is 240, the number of pilot
available pilot symbols. In the case of Subcarrier 4, chan- symbols in each frame is expressed by 240·G. Therefore, the
nel variation of the pilot symbol (Symbol 14) in Subcarrier 3 number of copying operations in the pilot all-copied channel
is copied to Subcarrier 4, although Subcarrier 5 has a pilot estimation scheme is calculated as 480·G since the channel
symbol in Symbol 12. variation of each pilot symbol is copied to two adjacent sub-
In addition, the copying process is applied to all interpola- carriers. Meanwhile, the number of copying operations in pilot
tion intervals with five or more data symbols to prevent the partly-copied scheme is reduced to 240·(G–1) since channel
long interpolation interval (Fig. 11(a)-(2)). Here, to reduce the variations of pilot symbols are partly copied to adjacent sub-
number of sample symbols in the interpolation range but to carriers as shown in Fig. 11(a). After the copying process,
prevent a relatively long interpolation interval, only the pilot the number of sample symbolsand pilot symbols in every
symbol closer to the center of the interpolation interval is 7 subcarriers is calculated as 3 · 6i=0  G−i
7  by applying the
utilized when both adjacent subcarriers have available pilot pilot all-copied channel estimation scheme. In each subcar-
symbols. In the case of Subcarrier 5, channel variation of rier, the number of division and subtraction operations equals
the pilot symbol (Symbol 9) in Subcarrier 4 is copied to to twice of the sum of sample symbols and pilot symbols, and
Subcarrier 5, although Subcarrier 6 has a pilot symbol in thus, the number
 of subtraction and division operations comes
Symbol 10. When two pilot symbols in adjacent subcarri- to 1440 · 6i=0  G−i 7 . Similarly, for the pilot partly-copied
ers are symmetrically close to the center of the interpolation channel estimation scheme, the number of sample symbols and
interval, the pilot symbol farther from the frame preamble is pilot symbols in every 7 subcarriers is calculated as 2G–1, and
selected. After the copying process, channel variations of all thus, the number of subtraction and division operations comes
the remaining data symbols are polar-linearly interpolated and to 480·(2G–1). In the case of G = 14, the total number of
extrapolated (Fig. 11(a)-(3)). Finally, CFO-derived phase rota- copying operations and subtraction and division operations in
tions of the farthest and second farthest data symbols in the the pilot partly-copied channel estimation scheme is reduced
extrapolation range are compensated by applying (26)–(29) to by approximately 40 % compared with the pilot all-copied
further improve channel estimation accuracy (Fig. 11(a)-(4)). channel estimation scheme.
2) Performance Evaluation: We applied the pilot partly-
copied channel estimation scheme to the DS subframe and V. P ROPOSED C HANNEL E STIMATION S CHEME W ITH
evaluated BER characteristics. Fig. 12 shows BER character- 2-B RANCH D IVERSITY FOR US
istics as a function of Eb /N0 by applying the parameters in In this section, we evaluate BER characteristics of the
Table V with various numbers of OFDM symbols in Table II. US subframe by applying the pilot partly-copied channel

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on March 06,2025 at 12:56:35 UTC from IEEE Xplore. Restrictions apply.
 944 IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING, VOL. 5, NO. 4, DECEMBER 2019

Fig. 11. Diagram of pilot partly-copied channel estimation scheme for DS subframe.

estimation scheme and confirm that the absence of the frame adjacent subcarriers. In addition, the proposed channel esti-
preamble impairs performance in a highly mobile long- mation scheme can flexibly support shorter US subframe with
delay multipath fading environment. Then, we introduce the less than 14 OFDM symbols. Fig. 14 shows a diagram and
2-branch MRC (Maximum Ratio Combining) diversity to a flowchart of the proposed channel estimation scheme for
improve the channel estimation accuracy. the US subframe in the case of 14 OFDM symbols. Firstly,
the copying process of channel variations is applied to the
A. Pilot Partly-Copied Channel Estimation Scheme subcarriers where three or more data symbols are placed in
For the US subframe, channel estimation accuracy can be the extrapolation range for expanding the interpolation range
also improved with the assistance of the pilot symbols in if possible. Here, to obtain a considerably long interpolation

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on March 06,2025 at 12:56:35 UTC from IEEE Xplore. Restrictions apply.
 OUYANG et al.: RELIABLE CHANNEL ESTIMATION SCHEME USING SCATTERED PILOT PATTERN 945

Fig. 12. BER characteristics of DS subframe using pilot partly-copied

channel estimation scheme with a residual CFO of 67 Hz.

Fig. 13. BER characteristics of DS subframe using pilot partly-copied chan-

nel estimation scheme by applying frequency synchronization with a CFO of
1,892 Hz.

TABLE VI
C OMPUTATIONAL C OMPLEXITY OF P ROPOSED C HANNEL E STIMATION
S CHEMES FOR DS S UBFRAME

range, the proposed scheme utilizes the farther pilot symbols

when both adjacent subcarriers have available pilot symbols.
In addition, the copying process is also applied to all inter-
polation intervals with five or more data symbols to prevent Fig. 14. Diagram of pilot partly-copied channel estimation scheme for US
a long interpolation interval. Here, the pilot symbol closer subframe.
to the center of the interpolation interval is utilized when
both adjacent subcarriers have available pilot symbols. In the
case of Subcarrier 3, channel variation of the pilot symbol symbol with a larger index is selected. After the copying pro-
(Symbol 3) in Subcarrier 4 is copied as a sample symbol, cess, channel variations of all the remaining data symbols are
although Subcarrier 2 has a pilot symbol in Symbol 7. When polar-linearly interpolated or extrapolated.
two pilot symbols in adjacent subcarriers are symmetrically Fig. 15 shows BER characteristics of the US subframe as
close to the center of the interpolation interval, the pilot a function of Eb /N0 with various numbers of OFDM symbols

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on March 06,2025 at 12:56:35 UTC from IEEE Xplore. Restrictions apply.
 946 IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING, VOL. 5, NO. 4, DECEMBER 2019

Fig. 15. BER characteristics of US subframe using pilot partly-copied

channel estimation scheme. Fig. 17. BER characteristics of US subframe with 2-branch diversity by
applying typical polar-linear interpolation and extrapolation channel estima-
tion scheme.

Fig. 18. BER characteristics of US subframe with 2-branch diversity by

applying pilot partly-copied channel estimation scheme.
Fig. 16. 2-branch MRC diversity.

improvement. As shown in Fig. 16, BS receives two signals

by applying the proposed scheme with the parameters in transmitted from one CPE through independent transmission
Table V. As a result, with a residual CFO of 67 Hz, the channels. As shown in Fig. 16(b), two received signals are
BER characteristics were obviously improved in comparison combined at the BS receiver, and the combined received signal
with the typical channel estimation scheme in Fig. 7. However, y’ is expressed by
a required BER of 2 × 10−4 was not achieved for any modu-
lation scheme. Due to the absence of the frame preamble and  y1 h̃1∗ + y2 h̃2∗
the different bit interleaving process from the DS subframe, y = 2 2
, (30)
relatively long extrapolation ranges still exist in some subcar- h̃1 + h̃2
riers even after the copying process as shown in Fig. 14(a)
where y1 and y2 are received signals of each branch, and
and degrade the channel estimation accuracy. Besides, due to
h̃1 and h̃2 are estimated channel variations attributed to each
the absence of the frame preamble, it is impossible to obtain
transmission channel.
the residual CFO accurately and to apply the extra phase
Fig. 17 shows BER characteristics of the US subframe as
compensation technique.
a function of Eb /N0 by applying the 2-branch MRC diversity
Meanwhile, the BER curves in Fig. 15 show similar char-
and the typical polar-linear channel estimation scheme with
acteristics regardless of the number of OFDM symbols. This
the parameters in Table V. As a result, a required BER of
indicates that the proposed channel estimation scheme has the
2×10−4 was achieved only for QPSK. In addition, the typical
potential to break through the limitation of the number of
polar-linear channel estimation scheme requires at least two
OFDM symbols even without the frame preamble.
pilot symbols in each subcarrier and limits the flexibility in
the number of OFDM symbols.
B. Diversity Scheme On the contrary, a combined scheme with the 2-branch
Since the design constraint of BS is more relaxed than CPE, MRC diversity and the proposed pilot partly-copied channel
we introduce the 2-branch MRC diversity for performance estimation scheme improves the channel estimation accuracy

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on March 06,2025 at 12:56:35 UTC from IEEE Xplore. Restrictions apply.
 OUYANG et al.: RELIABLE CHANNEL ESTIMATION SCHEME USING SCATTERED PILOT PATTERN 947

even in the case that the number of OFDM symbols is less [8] Y. Han, E. Ekici, H. Kremo, and O. Altintas, “Resource allocation
than 14. Fig. 18 shows BER characteristics of the US sub- algorithms supporting coexistence of cognitive vehicular and IEEE
802.22 networks,” IEEE Trans. Wireless Commun., vol. 16, no. 2,
frame by applying the parameters in Table V. Although the pp. 1066–1079, Feb. 2017.
BER performance depends on the number of OFDM symbols [9] H. Zhou, N. Cheng, Q. Yu, X. S. Shen, F. Bai, and D. Shan, “Toward
because of the difference in the position of sample sym- multi-radio vehicular data piping for dynamic DSRC/TVWS spectrum
sharing,” IEEE J. Sel. Areas Commun., vol. 34, no. 10, pp. 2575–2588,
bols, a required BER of 2 × 10−4 was achieved for QPSK Oct. 2016.
and 16QAM with various numbers of OFDM symbols from [10] T. Matsumura et al., “TV white-space vehicle communication system
8 to 16. In addition, when the number of OFDM symbols in Singapore based on LTE technology,” in Proc. IEICE Tech. Rep.
SmartCom, vol. 115, no. 273, Oct. 2015, pp. 37–40.
is 12 and 16, a required BER of 2 × 10−4 was achieved [11] IEEE Computer Society, Part 22: Cognitive Wireless RAN Medium
even for 64QAM. From these results, the combined scheme Access Control (MAC) and Physical Layer (PHY) Specifications:
demonstrated the capability of reliabile and flexible chan- Policies and Procedures for Operation in the TV Bands, IEEE Standard
802.22TM -2011, Jul. 2012.
nel estimation for the US subframe even in a highly mobile [12] T. Matsumura, H. Ueno, K. Mizutani, and H. Harada, “Compact IEEE
long-delay multipath fading environment. 802.22-based radio equipment enabling easy installation for regional
area network system using TV white-spaces,” in Proc. IEEE LANMAN,
2018, pp. 1–6.
[13] T. Jiang, Z. Wang, L. Zhang, D. Qu, and Y.-C. Liang, “Efficient spec-
VI. C ONCLUSION trum utilization on TV band for cognitive radio based high speed
In this study, we proposed a reliable and flexible chan- vehicle network,” IEEE Trans. Wireless Commun., vol. 13, no. 10,
pp. 5319–5329, Oct. 2014.
nel estimation scheme applicable to the IEEE 802.22-based [14] J. Chen, B. Liu, H. Zhou, L. Gui, N. Liu, and Y. Wu, “Providing vehicu-
wide area mobile communication system. In comparison with lar infotainment service using VHF/UHF TV bands via spatial spectrum
the conventional channel estimation schemes, the proposed reuse,” IEEE Trans. Broadcast., vol. 61, no. 2, pp. 279–289, Jun. 2015.
[15] E. Sofer and G. Chouinard, WRAN Channel Modeling, document 802.22-
scheme improved channel estimation accuracy with the effec- 05/0055r7, IEEE, Piscataway, NJ, USA, Aug. 2005.
tive assistance of the pilot symbols in adjacent subcarri- [16] E. Sourour, H. El-Ghoroury, and D. McNeill, “Frequency offset estima-
ers and reduced the number of errors in the extrapolation tion and correction in the IEEE 802.11a WLAN,” in Proc. IEEE VTC,
vol. 7, 2004, pp. 4923–4927.
range with the extra phase compensation technique even in [17] Z. Zhao, X. Cheng, M. Wen, B. Jiao, and C.-X. Wang, “Channel esti-
a highly mobile long-delay multipath fading environment. The mation schemes for IEEE 802.11p standard,” IEEE Intell. Transp. Syst.
proposed scheme also offered the flexibility with regard to the Mag., vol. 5, no. 4, pp. 38–49, Oct. 2013.
[18] R. Funada, H. Harada, Y. Kamio, S. Shinoda, and M. Fujise, “A high-
number of OFDM symbols in the US and DS subframes. As mobility packet transmission scheme based on conventional standardized
a simulation result for the DS subframe, a required BER of OFDM formats,” in Proc. IEEE VTC, 2002, pp. 204–208.
2 × 10−4 was achieved for QPSK and 16QAM with a velocity [19] R. Funada, H. Harada, Y. Kamio, S. Shinoda, and M. Fujise, “A new
amplitude and phase compensation scheme under fast fading environ-
of 80 km/h and a CFO of 1,892 Hz, assuming a harsh mobile ment for OFDM packet transmission systems,” in Proc. IEEE VTC,
environment with a long-delay multipath fading. For the US vol. 4, 2001, pp. 2093–2097.
subframe, the combined scheme of the 2-branch diversity and [20] T. Zemen, L. Bernado, N. Czink, and A. F. Molisch, “Iterative time-
variant channel estimation for 802.11p using generalized discrete prolate
the proposed channel estimation scheme achieved a BER of spheroidal sequences,” IEEE Trans. Veh. Technol., vol. 61, no. 3,
2 × 10−4 for QPSK and 16QAM with a velocity of 80 km/h pp. 1222–1233, Mar. 2012.
and a residual CFO of 67 Hz regardless of the number of [21] F. Weng, C. Yin, and T. Luo, “Channel estimation for the downlink of
3GPP-LTE systems,” in Proc. IEEE ICNIDC, 2010, pp. 1042–1046.
OFDM symbols, despite the absence of the preambles and dif- [22] S. Galih, R. Karlina, F. Nugroho, A. Irawan, T. Adiono, and
ferent pilot patterns caused by the interleaving process. These A. Kurniawan, “High mobility data pilot based channel estimation for
results demonstrated the feasibility of the IEEE 802.22-based downlink OFDMA system based on IEEE 802.16e standard,” in Proc.
IEEE ICCAS-SICE, 2009, pp. 2333–2338.
wide area mobile communication system in rural areas. [23] A. Zajic, Mobile-to-Mobile Wireless Channels. Norwood, MA, USA:
Artech House, 2013.
[24] M. Jiang, S. Huang, and W. Wen, “Adaptive polar-linear interpolation
R EFERENCES aided channel estimation for wireless communication systems,” IEEE
Trans. Wireless Commun., vol. 11, no. 3, pp. 920–926, Mar. 2012.
[1] R. Ouyang, T. Matsumura, K. Mizutani, and H. Harada, “A robust
channel estimation for IEEE 802.22 enabling wide area vehicular
communication,” in Proc. WPMC, 2018, pp. 52–57.
[2] N. Lu, N. Cheng, N. Zhang, X. Shen, and J. W. Mark, “Connected
vehicles: Solutions and challenges,” IEEE Internet Things J., vol. 1,
no. 4, pp. 289–299, Aug. 2014.
[3] J. Heinovski, F. Klingler, F. Dressler, and C. Sommer, “Performance
comparison of IEEE 802.11p and ARIB STD-T109,” in Proc. IEEE
VNC, Dec. 2016, pp. 1–8.
[4] IEEE Computer Society, IEEE Standard for Information Technology—
Local and Metropolitan Area Networks—Specific Requirements—
Part 11: Wireless LAN Medium Access Control (MAC) and Physical Ruiting Ouyang received the B.E. degree in com-
Layer (PHY) Specifications Amendment 6: Wireless Access in Vehicular munication science and engineering from Fudan
Environments, IEEE Standard 802.11pTM -2010, Jul. 2010. University, China, in 2016, and the M.I. degree
[5] 700 MHz Band Intelligent Transport Systems, Version 1.2, ARIB in communications and computer engineering from
Standard STD-T109, Dec. 2013. Kyoto University, Japan, in 2018, where she is cur-
[6] M. L. Sichitiu and M. Kihl, “Inter-vehicle communication systems: A rently pursuing the Ph.D. degree with the Graduate
survey,” IEEE Commun. Surveys Tuts., vol. 10, no. 2, 2nd Quart., 2008. School of Informatics. She currently researches the
[7] S. Al-Sultan, M. M. Al-Doori, A. H. Al-Bayatti, and H. Zedan, “A topic of the development of IEEE 802.22-based
comprehensive survey on vehicular ad hoc network,” J. Netw. Comput. mobile communication system.
Appl., vol. 37, pp. 380–392, Jan. 2014.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on March 06,2025 at 12:56:35 UTC from IEEE Xplore. Restrictions apply.
 948 IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING, VOL. 5, NO. 4, DECEMBER 2019

Takeshi Matsumura (M’08) received the M.E. Hiroshi Harada is a Professor with the Graduate
degree in electronic engineering and the Ph.D. School of Informatics, Kyoto University, and
degree in nano-mechanics engineering from Tohoku an Executive Research Director of Social ICT
University in 1998 and 2010, respectively. From Research Center, National Institute of Information
1998 to 2007, he had been engaged in the and Communications Technology (NICT). In 1995,
research and development of wireless communica- he joined the Communications Research Laboratory,
tions devices in some companies. In 2007, he joined Ministry of Posts and Communications (cur-
the Smart Wireless Laboratory, National Institute rently, NICT). He was a Visiting Professor with
of Information and Communications Technology the University of Electro-Communications, Tokyo,
(NICT) as a Researcher, where he was engaged in Japan, from 2005 to 2014. Since 1995, he has
the white-space communication systems and fifth been researching on software-defined radio, cog-
generation mobile communication systems. From 2016 to 2019, he was nitive radio, dynamic spectrum access network, wireless smart ubiquitous
an Associate Professor with the Graduate School of Informatics, Kyoto network, and broadband wireless access systems on the microwave and
University. Since 2019, he has been a Research Manager with the Wireless millimeter-wave band. He also has joined many standardization committees
Systems Laboratory, Wireless Networks Research Center, NICT and also and forums in the United States as well as in Japan and fulfilled important
a Researcher with the Graduate School of Informatics, Kyoto University. roles for them, especially IEEE 1900 and IEEE 802. He has authored the book
His research interests include white-space communication systems, wide- entitled Simulation and Software Radio for Mobile Communications (Artech
area wireless network systems, and 5G and beyond mobile communication House, 2002). He was a recipient of the Achievement Awards in 2006 and
system. He was a recipient of the Best Paper Award from IEEE LANMAN 2018 and the Achievement Awards of ARIB in 2009 and 2018, on the topic
2017. of research and development on cognitive radio and wireless smart utility
network. He was the Chair of IEEE DySpan Standards Committee and the
Vice Chair of IEEE 802.15.4g, IEEE 802.15.4m, 1900.4, and TIA TR-51.
He was a Board of Director of IEEE Communication Society Standards
Board, SDR Forum, DSA Alliance, and WhiteSpace Alliance. He is the
Keiichi Mizutani received the B.E. degree in Co-Founder of WI-SUN Alliance, and has been served as the Chairman of the
electric, electrical, and system engineering from board and the Chairman of HAN Working Group in the alliance since 2012.
Osaka Prefecture University, Japan, in 2007, and He was the Chair of the IEICE Technical Committee on Software Radio
the M.E. and Ph.D. degrees in electric and elec- and the Chair of Public Broadband Mobile Communication Development
trical engineering from the Tokyo Institute of Committee, ARIB. He is also involved in many other activities related to
Technology, Japan, in 2009 and 2012, respectively. telecommunications. He is a fellow of IEICE in 2009.
He is an Assistant Professor with Kyoto University.
He was an Invited Researcher with Fraunhofer
Heinrich Hertz Institute, Germany, in 2010. From
April 2012 to September 2014, he was a Researcher
with the National Institute of Information and
Communications Technology, where he has been involved in IEEE 802 stan-
dardization activities, namely 802.11af, 802.15.4m, and 802.22b. He currently
researches the topics of physical layer technologies in white space com-
munications, dynamic spectrum access, wireless smart utility networks,
and 4G/5G/beyond 5G systems, including OFDM, OFDMA, MIMO, and
multihop relay network systems. He was a recipient of the Special Technical
Awards from IEICE SR Technical Committee in 2009 and 2017, the Best
Paper Award from IEICE SR Technical Committee in 2010, the Young
Researcher’s Award from IEICE SRW Technical Committee in 2016, and
the Best Paper Award from WPMC2017.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on March 06,2025 at 12:56:35 UTC from IEEE Xplore. Restrictions apply.