Estimation of UFMC Time-Varying Fading Channel Using Adaptive Filter

Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

2018 International Conference on Promising Electronic Technologies

Estimation of UFMC Time-Varying Fading Channel Using


Adaptive Filter
Ali Jamoos Moataz Hussein
Department of Electronic and Communication Engineering Information and Communication Technology Center
Al-Quds University Al-Quds Open University
Jerusalem, Palestine Jerusalem, Palestine
ali.jamoos@staff.alquds.edu 99999 moataz@qou.edu

Abstract – Universal filtered multicarrier (UFMC) OFDM especially when it combines with offset quadrature
modulation technique was suggested as a strong candidate system amplitude modulation (OQAM). It uses digital filtering to solve
for future 5G mobile communication systems. Indeed, it combines sensitivity interference between adjacent subcarriers, less
the advantages of orthogonal frequency division multiplexing sensitive to CFO, more clearer carriers, more resistance to
(OFDM) and filter bank multicarrier (FBMC) modulation estimation error due to time/frequency shift by removing or
techniques while avoids their drawbacks. UFMC seems to be more elimination of sidelobes.
effective for future 5G applications and scenarios like a machine
Due to all mentioned limitations, and the vision of 5G to serve
to machine (M2M), vehicle to8vehicle (V2V), device to device
(D2D) and the Internet of Things (IoT). This paper address the wide different smart mobiles, extreme Mobile broadband,
estimation and equalization of UFMC time-varying fading massive M2M, IoT, V2V, D2D [7-9]. The key design of
channels using adaptive filters based on comb pilot symbols candidate system modulation technique is flexibility, versatility,
arrangement. To exploit the fading channel statistics, the fading scalability, and efficiency to serve wide range of applications
process evolution is modeled by an autoregressive (AR) model and
access [8, 9]. Supporting M2M is big challenging feature of
tracked by a Kalman filter. The AR model parameters are
obtained by solving the so-called Yule-Walker (YWE) equations 5G , it will be serve around 50 billion machines that expected
based on the Bessel autocorrelation function of the fading channel to be connected by the year 2020 [10]. Hence, for 5G, BER
with known Doppler rate. The result of MATLAB simulation show performance for proposed schemes better than current system
the effectiveness of the proposed Kalman filter based channel are under study [11, 12]. Some candidate's schemes like FBMC
estimator as compared with the conventional ones like recursive
[13], and Generalized Frequency Division Multiplexing
least square (RLS) and least8 mean square8 (LMS) channel
estimator. (GFDM) [14] and UFMC [15]. As an alternative to OFDM and
FBMC, UFMC has various advantages over OFDM and FBMC.
Keywords— UFMC, fading channel, channel estimation, pilot [16, 17] Shows that UFMC has better spectral efficiency than
symbols, Kalman filter, Autoregressive model. FBMC. Similarly, UFMC is less implementation complexity
than FBMC and reverse compatibility to older system. It is also
I. INTRODUCTION implemented based on IFFT/FFT pairs and use prototype filter
called Chebyshev which can decrease spectral leakage
The current multiple access in a 4G mobile generation is
OFDM which is widely employed in Power Line drawbacks of OFDM and FBMC systems. As we review
communication, , UWB, ADSL, Digital Audio broadcasting current multiple access for LTE and their drawbacks to support
(DAB), digital video broadcasting (DVB), and WLAN a new application for future 5G. Another new candidate, which
standards, Wireless metropolitan network (WMAN) and can be used in future 5G is UFMC. UFMC is a generalization
WiMAX [1,2]. The advantages of OFDM is easy transceiver of the advantage features of OFDM and FBMC [18]. In addition,
implementation where it is performed using pair of (IFFT, FFT) it is avoid the drawbacks of the other two systems. It is
at transmitter and receiver respectively. The implementation of operation based on filtering consecutive subband (a group of
pair is very easy by digital processing [3, 4]. subcarriers) in modulation and demodulation. Some points of
However, it had many drawbacks and restrictions like cyclic interests of UFMC are: It is a simple transceiver called UF-
prefix(CP), high overhead, higher sensitivity to carrier OFDM, with filtering length equal to 1 (L=1). The non-
frequency offset (CFO) and phase variance of the received orthogonal carriers are not important, which means better
signal related to the reference at receiver [5,6], high Peak
spectral efficiency and low latency. Supporting 5G application
average power ratio (PARP) which reduces the power of
transmitter and relatively low spectral power density (PSD). like the internet of thing (IoT), M2M and D2D [19] which
FBMC is short for filter bank multicarrier modulation release strict synchronous transition between transmitter and
technique. It one of possible system candidate to be employed receiver. V2V required a signal that is less sensitivity to CFO
in future 5G. The implementation of FBMC is the same OFDM to achieve reliable communication between nodes mobile,
system based on IFFT/FFT pairs operation, and it used a bank Support M2M that require new physical interface and new
of digital filters like: prototype filter instead of guard interval media access control layer to ensure communication between
with a cyclic prefix to raise spectral efficiency compared to machines to any like machine, human, vehicle. Another

978-1-5386-5697-6/18/$31.00 ©2018 IEEE 43


DOI 10.1109/ICPET.2018.00014
advantage is new signal design to allow different users to II. UFMC
communicate at the same time in different subbands. Support
In [35], the idea of UFMC system is presented in fig. (1.a).
the fragment spectrum and this allows sending small data The operation of UFMC is filtering group of subband instead of
packets. No CP, instead of that setting time of filter used as filtering single carrier or whole band. To produce UFMC signal,
guard period. UFMC system is a generalization of combination every subbands carrying number of carriers convert to parallel
advantages of current system OFDM and FBMC systems and and pass to IFFT block. After that, applying filtering operation
avoided their limitations and restrictions. It is operation based to decrease OOB. The output of each filter was summed to form
in a filter of a consecutive subset of the complete band UFMC Signal. Moreover, since the number of a complex
modulation and demodulation instead of filtering the whole symbol in each subband is less than the IFFT points multiply
band or single carrier. with a number of subbands, the rest of block padded with
Fading process estimation and equalization procedures are number of zeros.
critical in a wireless communication system for symbol Mathematically, the information input to the UFMC system are
recovering bits at the receiver side. Transmitted signal is complex of QAM symbol ( ) is divided into subbands called
sensitive to many effects during traveling to the receiver. B, k have the values of (1 … ), Where “N” refers to IFFT
Few UFMC studies have handle the issue of the channel constellation points. Each of subbands having numbers of
estimation and equalization to the current date [20-22] subcarriers k and of complex symbols. All subbands are
compared to many studies were addressed in OFDM in the padded with zeros carriers to make each block contain
literature [23-26] and FBMC [27-30]. Channel estimation symbols. Then converting frequency symbols =
techniques in the UFMC and FBMC/OQAM are different in [ ] to time domain = [ ] by using
conventional techniques used in OFDM because the channel IFFT block. The result of each IFFT converted to serial. Each
frequency response values are real in OFDM and complex in padded zero symbol is filtered with chebyshev window
FBMC and UFMC. Second reason is subbands filtering determined side lobe attenuation level and filter length .
operation for to decrease out of band emission (OOB) may Dolph-Chebyshev window have low sidelobe levels different
produce dissimilar filter gain on different subcarriers for one from fixed window like Hanning windows. Forgiven sidelobe
UFMC symbol [31]. In [20] the author proposed estimation of attenuation and filter length, Chebyshev window yield narrow
UFMC channel in an uplink multiuser in timing and frequency main lobe compared to fixed window functions. The
offset and shows that the conventional pilot-aided channel characteristics of the filter are shifted in frequency according to
estimation used in OFDM is efficient in UFMC. In [21] author subcarriers for each subband. The output signal for each
introduces channel estimation based on combo type pilot subband from each filter is added, and then transform to the
pattern based on various algorithms like LSLI, DFT, and frequency domain. The closed form of Chebyshev window can
RMMSE. In [22] the authors introduce estimation using PRN be express as follows:
to be as guard interval and TS at time domain to minimize total
of transmitted pilots in the frequency domain and the effect will
be improvement in spectral efficiency. [ ]
W , = (1)
All proposed channel estimation techniques are categorized ( )
as blind techniques and training sequence/pilot based
techniques. The advantage of the blind technique is no need to Where denotes for IFFT points, k=0, 1, 2….M-1.
insert pilots, but the drawbacks are complexity and requiring a β = cosh [ cosh (10 ) ] , where α = (2, 3 ,4).The
large amount of data and this means loss of efficiency. While parameters adjust the side lobe levels by the expression Side-
the other technique is lower computational complexity. lobe level in dB=−20 . For example, if, we have = 3 then
In our research, we addressed pilot based technique. Many the side-lobes level below the main-lobe with 60dB, where the
approaches were developed when using the Kalman filter for main lobe peak normalize to zero. The chebyshev side lobes are
channel estimation [27] [32-34]. In [34] authors introduce the called ripple in the stop band, comstellationsince they are equal
combination of Kalman filter with AR process to track channel in heights. The number of subbands B will be different
variations achieving better results of BER performance than according to according to application requirements. As
conventional adaptive estimators like LMS and RLS for MC- example, if the UFMC system employed in fragmented
CDMA. While [32-33] describes the Kalman estimator with an spectrum then number of subbands equal to the accessible
AR model to estimation fading channel in OFDM. Recently, spectrum portion in the fragmented spectrum.
dual Kalman filters are proposed in [27] to estimate In other words, we have input data X that will be divided to B
FBMC/OQAM fading channels process. The fading model subband with k subcarriers. Each subband passing through
performed by an AR, its coefficients are gained by (YWE). IFFT block symbolized by matrix V. The result of IFFT block
The structure of this paper is formed as follows. We will will be serialized and passing to filter represent with matrix F.
introduce UFMC transceiver in section II. The estimation of We can express the baseband UFMC signal [ ] by:
fading process using AR model tracked by Kalman estimator in
section III. Simulation and analyses results will be presented in /
section IV. Finally, The conclusion will be present in section V. [ ]= ∑ ∑ ∑ (1)

44
N-Point Prototype
IDFT Filter

+
AWGN
P/S

[ ]

N-Point Prototype
IDFT Filter Varying
110110 Modulation DAC
S/P + Channel
P/S

N-Point
Prototype
IDFT Filter
+
P/S

Figure 1.a: Block Diagram of UFMC Transmitter


˄
˄

Time
Domain ADC
Processing
+
˄ ˄ S/P
Kalman
Frequency Filtering
Domain Based 2N-
Processing Channel
P/S Point
110110 Demodulation per estimation
Subcarriers and FFT
equalization

˄
Zero
Padding

Figure 1.b: Block diagram of UFMC Receiver

Where N denotes for a number of FFT constellation points, lying between [0,2π]. To utilize the statistical characteristics of
while L denotes for the length of Chebyshev filter and fading signal, it is described the by PSD and zero order Bessel
denotes for Chebyshev Filter. ACF. The PSD of fading signal represented by well-known U
UFMC has more elasticity for filtering every subband by its shaped band limited jakes spectrum with maximum Doppler
spectrum. It’s can allow the scheme to adapt service types by frequency [36].
regulating the subband and filter coefficients only. UFMC , for f |f |
provide access for two different applications in the same time )
like M2M and D2D in two dissimilar subbands with disparate S(f) = ( ) (4)
necessity and frame structure without creating inter-symbol 0, Otherwise
band interference (ISBI) due to reduction OOB [22]. Where = , with v light speed and , is the wavelength of
wave carriers.
III. Kalman Filtering Based Channel Estimation
The Autocorrelation function (ACF) can be express by:
As we mentioned in section II, the delivered signal from the
transmitter is transformed S/P and passed through FFT = (2 | |) (5)
transformer block. The output of FFT serialize again then Where denotes zero ordered Bessel function, denote
symbol demapper retrieve original data. symbol period, and point to Doppler rate. In [37] use
The received vector signal at the receiver after passing general Autoregressive (AR) model to generate correlated
channel impulse response (CIR) can be represented by a Rayleigh processes with fading process that shape the spectrum
transmitted signal [ ] with the channel in addition to of the uncorrelated Gaussian variates. Fading process can be
additive white Gaussian noise as in (2): designed by order of AR in [37]:
[n] = [ ]ℎ [n] + [ ] (2) ℎ[ ] = − ∑ ℎ[ − ] + ( ) (6)
By Substitution value of x in (2) we get: Where ( ) refers to complex white Gaussian noise process
/ with zero mean and driving process covariance σ . a
[n] = ℎ [n] ∑ ∑ ∑ [ ] +
denotes for the AR parameters. We can express the PSD of
(3) AR model by:
Where ℎ denotes for the complex-valued of fading process
over sub-band carriers for the symbols and denotes ( )= ( )
(7)

for AWGN process. Where denotes the normalized frequency. ℎ ( ) denotes
We assume the noise processes to be mutually (i.i.d) distributed fading process that estimated by the Kalman filter. The
zero-mean and variance . The fading process ℎ [ ] can be estimation of fading process ℎ ( ) over the along the
Perform as a complex Gaussian process ℎ ( )= carrier. The carrier subscript is dropped for simplicity and
|ℎ ( )| ( )
with zero mean. With uniformly phase ( ) clarity, then the state vector can be express as follows:

45
= ℎ ℎ … ℎ (8) Channel equalizing is estimated using multiplying received
signal with normalize version of the complex conjugate of the
channel estimate as:
∗( )
The expression (8) can be expressed as state space form as in ( ) = ( ) (20)
( )
follows:
= + (9)
Where: IV. Simulation Results
− − ⋯ −
In this part, we will estimate high varying fading channel
1 0 ⋯ 0 process between conventional filters like LMS, RLS and our
= Is the state transition model
⋮ ⋮ ⋱ ⋮ proposed estimator Kalman filter combined with the
0 ⋯ 1 0 autoregressive model with comb pilot arrangement 1:8. In
matrix and g express as in follows:
addition, we will compare power spectral density between three
= [1 0 … 0] modulation system OFDM, FBMC and UFMC.
In addition, by substation of equation (9) in equation (2), it
Fig. (2) shows the BER performance of UFMC using different
follows:
mapping techniques. It is clear that the best QAM modulation
= + (10) when we use the Bits per subcarrier (M=4).
Where:
= [ 0 ⋯ 0] (11)
We will use a Kalman filter to estimate the fading process /
for the state vector by assumed a set of observations
,⋯, are known. another process needed so-called
innovation process which refer to a difference between
observed value and optimal value which can be obtained as
follows:
= − / (12)
The variance of the innovation process as follows:
∗]
= [ = / + (13)
Where:
/ represent a priori error covariance matrix that be get
repetitively as follows:
/ = / + (14)
The Kalman gain is enumerated by the next equation:
Fig.2: BER vs SNR for UFMC QAM Modulation

= / (15)

The estimation of a posteriori of fading process given by:


/ = / + (16)

And the state vector given by:


ℎ =ℎ / = / (17)
The update of / is get repetitively as follows:
/ = / − / (18)
We assign the initial value for the state vector / to zero
vector i.e. / = 0 and the initial value for the error
covariance matrix / to identity matrix i.e. / =
According to fig. (1.b), the received signal passed to 2N-
FFT block to tranform it into frequency domain, this block (2N-
FFT ) is very necessary for recovering symbols at N subcarriers
from “ + − 1" received samples. The result of FFT block
transformer is given by:

( )=

∑ [ ] ⁄
, = 0,1. . .2 − 1 (19) Fig.3: PSD vs Normalize frequency for UFMC, 10 Subband, 20 Subcarriers
for each

46
Fig. 4: PSD vs Normalize frequency Power for OFDM, 200 Subcarriers

Figure 8: BER vs SNR for UFMC System, = .

BER performance for Doppler rate = 0.0259 in Fig. (8),


The BER performance for the Kalman estimator is better than
the other estimators. Also, we notice improvement of BER for
RLS estimator for the same scenario (Doppler rate smaller than
the Doppler rate in Fig. (7).

Fig. 5: PSD vs Normalize frequency Power for FBMC, 200 Subcarriers,


Overlapped symbols K=4
Fig. (3), Fig. (4) and Fig. (5) illustrate respectively the PSD
performance for the same number of subcarriers (200
Subcarriers) for UFMC, OFDM, and FBMC respectively. The
PSD for candidate multiple access UFMC is higher than the rest
Candidate system.
Given Fig. (6) that shows the envelope of an estimated fading
channel generated by Rayleigh fading model with Doppler rate,
= 0.0185.
According to Fig. (7), it illustrates the BER performance for
Kalman estimator in addition to LMS, RLS for higher Doppler
rate = 0.0556, v = 150km/h . It is notable that the
performance of the Kalman estimator has lower BER.

Figure 9: BER vs SNR for UFMC System, = .


The Simulation results in Fig. (9). BER performance for low
Doppler rate = 0.0185, Kalman estimator is still the best
among the others. Also, we notice BER performance
improvement for RLS estimator for lower Doppler rate.

Fig 6: Channel Generation according to Rayleigh Fading Generator, = VI.Conclusions


. , µ=0.03, λ=0.4
Current multiple access used in LTE is not a candidate for
future mobile generation dues to their limitations and
restrictions. After comparative between various candidate
systems, the UFMC system has a strong candidate for later
mobile generation 5G. In this research, we address the
estimation and equalization of the UFMC fading channel by
Kalman filter combined with the autoregressive process. The
simulation results show that the Kalman filter overcomes
performance conventional filter like LMS, RLS, and shows the
performance of UFMC cope the performance of current system
Figure 7: BER vs SNR for UFMC System, = . used in current 4G.

47
REFERENCES Vehicular Technology Conference (VTC Fall), 2015 IEEE 82nd, pp. 1-5.
IEEE, 2015.
[1] Li, Ye Geoffrey, and Gordon L. Stuber, eds. orthogonal frequency
[22] Zhang, Lei, Chang He, Juquan Mao, Ayesha Ijaz, and Pei Xiao. "Channel
division multiplexing for wireless communications. Springer Science &
estimation and optimal pilot signals for universal filtered multi-carrier
Business Media, 2006.
[2] K. Baum, Kevin, Brian Classon, and Phillire Sartori. Principles of (UFMC) systems." In Personal, Indoor, and Mobile Radio
broadband OFDM cellular system design. Wiley, 2008. Communications (PIMRC), 2017 IEEE 28th Annual International
[3] Farhang-Boroujeny, Behrouz. "OFDM versus filter bank multicarrier." Symposium on, pp. 1-6. IEEE, 2017.
IEEE signal processing magazine 28, no. 3 (2011): 92-112. [23] Wang, Rong, Jingye Cai, Xiang Yu, and Sirui Duan. "Compressive
[4] H Zhang, Haijian, Didier Le Ruyet, Daniel Roviras, Yahia Medjahdi, and channel estimation for universal filtered multi-carrier system in high-
Hong Sun. "Spectral efficiency comparison of OFDM/FBMC for uplink speed scenarios." IET Communications 11, no. 15 (2017): 2274-2281.
cognitive radio networks." EURASIP Journal on Advances in Signal [24] Ozdemir, Mehmet Kemal, and Huseyin Arslan. "Channel estimation for
Processing 2010 (2010): 4. wireless OFDM systems." IEEE Communications Surveys & Tutorials 9,
[5] G.Bochechka. “Methods of channel estimation no. 2 (2007): 18-48.
based on built in pilot-signals in OFDM systems”. Telecommunications [25] Coleri, Sinem, Mustafa Ergen, Anuj Puri, and Ahmad Bahai. "Channel
and Transport, 2009, No3, pp. 38-42. estimation techniques based on pilot arrangement in OFDM systems."
[6] G. Bochechka, Estimation of the beginning of the OFDM block and
IEEE Transactions on broadcasting 48, no. 3 (2002): 223-229.
frequency shift in system IEEE 802.11a.T-Comm Telecommunications
and Transport, 2009, №5, pp.34-37. [26] Manton, Jonathan H. "Optimal training sequences and pilot tones for
[7] F.Hu, “Opportunities in 5G Networks a research and development OFDM systems." IEEE Communications Letters 5, no. 4 (2001): 151-153.
perspective”, CRC Press,2016 [27] Lin, Jia-Chin. "Least-squares channel estimation for mobile OFDM
[8] Cisco,https://www.cisco.com/c/en/us/solutions/collateral/service- communication on time-varying frequency-selective fading channels."
provider/visual-networking-index-vni/mobile-white-paper-c11- variations 4, no. 6 (2008): 12.
520862.html [28] Aldababseh, Mahmoud, and Ali Jamoos. "Estimation of FBMC/OQAM
[9] Tullberg, Hugo, Heinz Droste, Mikael Fallgren, Peter Fertl, David fading channels using dual Kalman filters." The Scientific World Journal
Gozalvez-Serrano, Eiman Mohyeldin, Olav Queseth, and Yngve Seien. 2014 (2014).
"METIS research and standardization: A path towards a 5G system." [29] Kofidis, Eleftherios, Dimitrios Katselis, Athanasios Rontogiannis, and
In Globecom Workshops (GC Wkshps), 2014, pp. 577-582. IEEE, 2014. Sergios Theodoridis. "Preamble-based channel estimation in
[10] ‘‘More than 50 billion connected devices,’’ Ericsson, Stockholm, Sweden, OFDM/OQAM systems: A review." Signal Processing 93, no. 7 (2013):
White Paper 284 23-3149, Feb. 2011. [Online]. Available: 2038-2054.
[11] http://www.akos- [30] Kofidis, Eleftherios, and Dimitrios Katselis. "Improved interference
rs.si/files/Telekomunikacije/Digitalna_agenda/Internetni_protokol_Ipv6/ approximation method for preamble-based channel estimation in
More-than-50-billion-connected-devices.pdf FBMC/OQAM." In Signal Processing Conference, 2011 19th European,
[12] Wunder, Gerhard, Peter Jung, Martin Kasparick, Thorsten Wild, Frank pp. 1603-1607. IEEE, 2011.
Schaich, Yejian Chen, Stephan Ten Brink et al. "5GNOW: non- [31] Chen, Chih-Wei, and Fumiaki Maehara. "An enhanced MMSE
orthogonal, asynchronous waveforms for future mobile subchannel decision feedback equalizer with ICI suppression for
applications." IEEE Communications Magazine 52, no. 2 (2014): 97-105. FBMC/OQAM systems." In Computing, Networking and
[13] Schaich, Frank, and Thorsten Wild. "Waveform contenders for 5G— Communications (ICNC), 2017 International Conference on, pp. 1041-
OFDM vs. FBMC vs. UFMC." In Communications, Control and Signal 1045. IEEE, 2017
Processing (ISCCSP), 2014 6th International Symposium on, pp. 457-460. [32] Zhang, Lei, Ayesha Ijaz, Pei Xiao, Muhammad Ali Imran, and Rahim
IEEE, 2014. Tafazolli. "MU-UFMC system performance analysis and optimal filter
[14] Farhang-Boroujeny, Behrouz. "OFDM versus filter bank multicarrier." length and zero padding length design." arXiv preprint
IEEE signal processing magazine 28, no. 3 (2011): 92-112. arXiv:1603.09169 (2016).
[15] G. Fettweis, Gerhard, Marco Krondorf, and Steffen Bittner. "GFDM- [33] W. Chen and R. Zhang, “Estimation of time and frequency selective
generalized frequency division multiplexing." In Vehicular Technology channels in OFDM systems: a Kalman filter structure,” in Proc. IEEE
Conference, 2009. VTC Spring 2009. IEEE 69th, pp. 1-4. IEEE, 2009. GLOBECOM'04, pp. 800-803, Nov. 2004.
[34] Jamoos, Ali, Ahmad Abdo, and Hanna Abdel Nour. "Estimation of
[16] Vakilian, Vida, Thorsten Wild, Frank Schaich, Stephan ten Brink, and OFDM time-varying fading channels based on two-cross-coupled Kalman
Jean-Francois Frigon. "Universal-filtered multi-carrier technique for filters." In Novel Algorithms and Techniques in Telecommunications,
wireless systems beyond LTE." In Globecom Workshops (GC Wkshps), Automation and Industrial Electronics, pp. 287-292. Springer, Dordrecht,
2013 IEEE, pp. 223-228. IEEE, 2013.
2008.
[17] F. Schaich, Frank, Thorsten Wild, and Yejian Chen. "Waveform
[35] Kalofonos, Dimitris N., Milica Stojanovic, and John G. Proakis.
contenders for 5G-suitability for short packet and low latency
"Performance of adaptive MC-CDMA detectors in rapidly fading
transmissions." In Vehicular Technology Conference (VTC Spring), 2014
Rayleigh channels." IEEE Transactions on Wireless Communications 2,
IEEE 79th, pp. 1-5. IEEE, 2014.
no. 2 (2003): 229-239.
[18] Gerzaguet, Robin, Nikolaos Bartzoudis, Leonardo Gomes Baltar, [36] Bochechka, Grigory, Valery Tikhvinskiy, Ivan Vorozhishchev, Altay
Vincent Berg, Jean-Baptiste Doré, Dimitri Kténas, Oriol Font-Bach et al. Aitmagambetov, and Bolat Nurgozhin. "Comparative analysis of UFMC
"The 5G candidate waveform race: a comparison of complexity and technology in 5G networks." In Control and Communications (SIBCON),
performance." EURASIP Journal on Wireless Communications and 2017 International Siberian Conference on, pp. 1-6. IEEE, 2017.
Networking 2017, no. 1 (2017): 13. [37] WC Jr, Jakes. "Microwave mobile communications." (1974).
[19] Wang, Xiaojie, Thorsten Wild, Frank Schaich, and A. Fonseca Dos Santos. [38] Baddour, Kareem E., and Norman C. Beaulieu. "Autoregressive modeling
"Universal filtered multi-carrier with leakage-based filter optimization." for fading channel simulation." IEEE Transactions on Wireless
In European Wireless Conference, pp. 1-5. 2014. Communications 4, no. 4 (2005): 1650-1662.
[20] Duan, Sirui, Xiang Yu, and Rong Wang. "Performance Analysis on Filter
Parameters and Sub-bands Distribution of Universal Filtered Multi-
Carrier." Wireless Personal Communications 95, no. 3 (2017): 2359-2375.
[21] Wang, Xiaojie, Thorsten Wild, Frank Schaich, and Stephan ten Brink.
"Pilot-aided channel estimation for universal filtered multi-carrier." In

48

You might also like