Improved F-OFDM scheme based on DFT-FB

SHAO Kai, LI Yan, MEN Lijun

(1.Chongqing Key Laboratory of Mobile Communication, Chongqing University of Posts and Communications, Chongqing 400065,P.R. China; 2.Office of Chongqing Municipal Committee of Political, Chongqing 401147, P.R. China)

ImprovedF-OFDMschemebasedonDFT-FB

SHAO Kai1, LI Yan1, MEN Lijun2

(1.Chongqing Key Laboratory of Mobile Communication, Chongqing University of Posts and Communications, Chongqing 400065,P.R. China; 2.Office of Chongqing Municipal Committee of Political, Chongqing 401147, P.R. China)

Due to using long subband filters to reduce the out-of-band emission(OOBE), the complexity of filtered-OFDM(F-OFDM)is very high. In order to solve this problem, based on the thoughts of filter bank processing, i.e. using polyphase network(PPN) to reduce the system complexity and filtering in each subcarrier to lower OOBE, this paper proposes subcarrier filtering technology of filter bank to improve F-OFDM. Compared with F-OFDM, the proposed scheme can obtain the same bandwidth efficiency and less implementation complexity. The simulation results also show that the proposed scheme can not only optimize the flexibility of system, but also improve the transmission performance and anti-channel fading ability in AWGN(additive white Gaussian noise) channel and fading channel.

cyclic prefixed-orthogonal frequency division multiplexing(CP-OFDM); DFT-FB; F-OFDM

1 Introduction

The fifth-generation (5G) cellular communication expect to serve the mobile users gigabit connection, the capacity increase will be up to three orders of magnitude compared with current long-term evolution (LTE) systems. There is a widespread agreement that such an ambitious goal will be realized in a combination of asynchronous heterogeneous network scenarios, such as enhanced mobile broadband (eMBB), massive machine type communications (mMTC), ultra-reliable and low latency communication machine type communications (uMTC). This demands the physical modulation system spectrum containment waveform, scenario-dependent with several challengeable elements, e.g. the superior tailored services capability and higher bandwidth efficiency.

The well-known cyclic prefixed orthogonal frequency division multiplexing (CP-OFDM) will be invalid in 5G co-existence scenarios, because its good performance is guaranteed only when synchronism and inter-subcarrier orthogonality is strict maintained. Specifically, CP-OFDM has one major limitation, i.e. poor sidelobe confinement suffered from the use of rectangular impulse, which results high out of band emission (OOBE) to neighbor spectral components. OFDM sidelobe control has gained a lot of attention，filtered OFDM[1-3]filtering the complete multicarrier band, is one direct reasonable method to eliminate the OOBE of OFDM symbols. Appling a filtering functionality on the whole band requires a wider passband filter, which is convenient to implement due to shorter filter length. However, filtered OFDM only focus on the issue of spectrum containment, which is difficult to provide scenario-dependent service. UFMC[4]as one variant of filtered OFDM, applies filtering to subsets of the complete band instead of the whole band. Separated filtering on subband could be used to accommodate various data connection with flexible configure of carrier numbers. F-OFDM[5]improved filtering performance of UFMC by allowing the filter length to exceed the CP length in order to guarantee better frequency-and time-localization[6-7]. However, UFMC or F-OFDM both are based on the construct of OFDM, constant subcarrier space seriously restricts the flexibility of scenario-dependent tailored service, and subband-based filtering has ignored inter-carrier interference (ICI) caused by frequency jitter or offsets.

This paper proposes a subcarrier-based filtering method to improve F-OFDM scheme. Discrete Fourier Transform-Filter bank (DFT-FB) will be adopted to improve F-OFDM, simplify named the improved scheme as Multitask-Filter Bank (M-FB).

2 F-OFDM

F-OFDM is one scenario-dependent transceiver configuration, where the whole used bandwidth is split into several subband to serve different UE (user equipment). Various connections would be assigned to appropriate spectrum resource, and then a specific subband filter is applied to shape the waveform of OFDM signal for the purpose of spectrum containment.
Fig.1 dictates an F-OFDM block diagram. Assuming there are totalDUEs, each UE covers various subcarrier numbers. Here UE1occupiesMconsecutive subcarriers and generates F-OFDM symbols through filteringLconsecutive OFDM symbols.Nis the IFFT(inverse fast Fourier transform) size, whereN>Mdue to the cyclic prefix. mathematically, each transmitter obtains:

*f(n),

With

(2)

and

sl(n)

(3)

Wheres(n) is the OFDM symbol,f(n) is the pulse shaping filter whose main-lobe bandwidth is equal to the assigned subcarriers,Ngis the CP length,dl,mdenotes the transmitter data symbols on the subcarriermof thelthsymbol, and the assigned subcarrier range is {1,2…,M}.

So, the signal at the receiver can be written as

*hi(n)+c(n)

(4)

(5)

3 M-FB scheme

Fig.1 Block diagram of F-OFDM transceiver

Fig.2 Block diagram of M-FB transceiver

Assuming the same scenario with
Fig.1, UE1occupiesMconsecutive subcarriers and generates output symbols through filteringLconsecutive symbols. Finally, the transmitter in output is

(6)

With

(7)

*hi(n)+c(n)

(8)

the signal in the output is

(9)

(10)

To sum up, according to different UE requirements, the parameters can be flexible set, such as choose different transmission rate through configuring interpolation factor and decimation factor, design different filter to achieve different band attenuation.

Consequently, the proposed M-FB construction could generate symbols as similar as F-OFDM done. However, the proposed scheme can provide much more flexibility and advantages as follows. 1) DFT-FB has no requirement of synchronization and orthogonal[8]; 2) Subcarrier filtering has the capability to eliminate the ICI; 3) There is possible to configureKivalue and subcarrier filter for various subband, which produce different subcarrier waveform or subcarrier space for different connection demands[9]. Next part we will give performance analysis at a special case of DFT-FB, whereki>M, meaning that no overlapping between neighbor subcarriers. Usually, we called this special example as filtered multitone (FMT) modulation[10].

4 Performance analysis

4.1 Bandwidth efficiency

(11)

Bandwidth efficiency is defined asη=symbolrate/bandwidth, obviously, bandwidth efficiency of F-OFDM is

(12)

which is usually less than 1 because of the presence of CP.

Fig.3 Time-frequency phase-space lattice representation of an F-OFDM system and M-FB.

Fig.3b presents the time-frequency phase-space lattice of the improved system. Due to the usage of bandlimited prototype filter, the frequency-spacing will be expending toF=(1+α)/T, whereαis the roll-off factor. Thus, the density of data symbol in eachTFunit area is

(13)

The bandwidth efficiency is

(14)

which is also less than 1 although absence of CP. Comparing Eq.(12) and Eq.(14), when 1+α=T/TFFT, two system are possible to obtain same bandwidth efficiency. For example, assumingTCPis one quarter ofT,T/TFFT=1.25,αshould be set as 0.25 for reaching same bandwidth efficiency.

To conclude, the proposed system is capable to obtain spectrum efficiency as similar as F-OFDM even at the worst case. Much better performance could be gotten by enforcing overlap between neighbor subcarriers. Mazo limit[11-12]presents the possibility of overlapped carriers without the deterioration of transmission performance.

4.2 System complexity

From the spectrum containment aspect of view, both systems improved filtering capability to guarantee better frequency and time localization. The most worth considering factor is the implement complexity. Assuming there are 64 subcarriers in one subband and total 1 000 QPSK symbol for transmission,
Fig.4 shows the power spectrum density (PSD) of F-OFDM and the proposed system respectively. Here Hanning window with 128 time-windowing mask based on soft-truncation technical has been used in F-OFDM and length of 256 SRRC filter with 0.2 roll-off factor has been used for proposed system. Tab.1 shows normalization system complexity, where set OFDM complexity to 1. Using split-radix algorithm[13]to calculate two system complexity in every subband under this circumstance. According to[14-15], we only care real multiplication as addition can be ignore when compared to multiplication. it’s obvious that the M-FB exhibit much less complexity than F-OFDM from Tab.1.

Fig.4 PSD of F-OFDM and M-FB

MCMOFDMF-OFDMM-FBComplexity120.891.33

4.3 Bit error rate

From the transmission performance aspect of view, [16] indicates that subcarrier filtering obtains the resistance of multipath fading.
Fig.5 compares the performances of the M-FB and F-OFDM in one subband under the AWGN channel and Fading channel conditions. Here 16 subcarriers are considered and total 1 000 QPSK symbol for transmission, here Hanning window with 128 time-windowing mask based on soft-truncation technical has been used in F-OFDM and length of 256 SRRC filter with 0.2 roll-off factor has been used for proposed system. At the case of M-FB, the BER performance is superior to F-OFDM regardless of in AWGN channel or fading channel. Due multipath fading, M-FB still exhibits better property even in Fading channel.

Fig.5 BER performance of F-OFDM and M-FB in AWGN channel and Fading channel of F-OFDM and M-FB

4.4 Flexibility of M-FB

From the flexibility of scenario-dependent tailored service aspect of view, those adjustable factors determine the flexibility of M-FB. Here all adjustable factors are including roll-off-factor of prototype filter(r), overlapping factor (Ov) between neighbor subcarriers, interpolation value and subcarrier bandwidth.
Fig.6 implement 4 subbands in frequency domain with individual parameter configuration, where overlapping factor is 0, 1/4, 1/2, simulation subband width set as 16, 32, 64, and prototype filter roll-off-factor is 0.25, 0.5,0.75. Obviously, those alterable parameters can be easily set to meet different scenario requirements.

Fig.6 Flexibility of M-FB

5 Conclusion and future work

This paper proposes DFT-FB technology to improve F-OFDM. Filtering on subcarriers is adopted in proposed M-FB system instead of subband filtering. Several merits such as implementation complexity and flexibility are easily obtained. The simulation result also shows that proposed scheme exhibit better transmission performance and stronger capability to resist channel fading than F-OFDM due to the characteristic of filter bank. For future network spectrum efficiency is one kill consideration of modulation system. At the worst case of DFT-FB, non-overlapping situation can guarantee similar spectrum efficiency as F-OFDM, and better performance could be reached by introducing overlap between neighbor subcarriers. However, intended overlapping results extra ICI, which might impact the transmission performance. This paper has demonstrated several overlapping subcarriers examples in section 4.4 for the purpose of flexible configuration. Though Mazo limits support the possibility from the aspect of theoretic analysis, the levels of overlapping and matched receiving technical are two most important considerations for future work should be focused on.

[1] CHUNG C D. Spectral precoding for rectangularly pulsed OFDM[J]. IEEE Transactions on Communications, 2008, 56(9):1498-1510.

[2] BAZZI J, WEITKEMPER P, KUSUME K, et al. Design and Performance Tradeoffs of Alternative Multi-Carrier Waveforms for 5G[C]// IEEE GLOBECOM Workshops.[S.l.]:IEEE, 2015:1-6.

[3] WEITKEMPER P, BAZZI J, KUSUME K, et al. Adaptive filtered OFDM with regular resource grid[C]// ICC-2016 IEEE International Conference on Communications Workshops.[S.l.]:IEEE, 2016:1-6.

[4] FRANK S, THORSTEN Wild. Waveform contenders for 5G — OFDM vs. FBMC vs. UFMC[C]//International Symposium on Communications, Control and Signal Processing.[S.l.]: Conference Publications, 2014:457-460.

[5] TONG W, MA J, HUAWEI P Z. Enabling technologies for 5G air-interface with emphasis on spectral efficiency in the presence of very large number of links[C]// Asia-Pacific Conference on Communications.[S.l.]:IEEE, 2015.

[6] ZHANG X, JIA M, CHEN L, et al. Filtered-OFDM-Enabler for Flexible Waveform in the 5th Generation Cellular Networks[C]//Global Communications Conference (GLOBECOM), 2015 IEEE. San Diego, CA:IEEE, 2015:1-6.

[7] ABDOLI J, JIA M, MA J. Filtered OFDM: A new waveform for future wireless systems[C]// IEEE, International Workshop on Signal Processing Advances in Wireless Communications. IEEE, 2015.[S.l.]: IEEE, 2015:66-70.

[8] VAIDYANATHAN P P. Multirate systems and filter banks[M]// Multirate systems and filter banks. 1 edition.[S.l.]: Prentice Hall,1992: 3385-3388.

[9] LIN Y P, VAIDYANATHAN P P. Application of DFT filter banks and cosine modulated filter banks in filtering[C]// Circuits and Systems, 1994. APCCAS '94. 1994 IEEE Asia-Pacific Conference on.[s.l.]: IEEE, 1994: 254-259.

[10] CHERUBINI G, ELEFTHERIOU E, OKER S, et al. Filter bank modulation techniques for very high speed digital subscriber lines[J]. IEEE Communications Magazine, 2000, 38(5):98-104.

[11] MAZO J E. Faster-Than-Nyquist Signaling[J]. Bell System Technical Journal, 1975, 54(8):1451-1462.

[12] SUGIURA S. Frequency-Domain Equalization of Faster-than-Nyquist Signaling[J]. IEEE Wireless Communication Letters, 2013, 2(5):555-558.

[13] BOUGUEZEL S, AHMAD M O, SWAMY M N S. An efficient split-radix FFT algorithm[C]// International Symposium on Circuits and Systems.[S.l.]:IEEE, 2003:IV-65-IV-68.

[14] European project ICT-211887 PHYDYAS. Deliverable D3.1: Transmit/receive processing (single antenna)[EB/OL]. (2008-06-05)[2016-07-22].http://www.ict-phydyas.org.

[15] NOGUET D, GAUTIER M, BERG V. Advances in opportunistic radio technologies for TVWS[J]. Eurasip Journal on Wireless Communications & Networking, 2011, 2011(1):1-12.

[16] SIOHAN P, SICLET C, LACAILLE N. Analysis and design of OFDM/OQAM systems based on filterbank theory[J]. Signal Processing IEEE Transactions on, 2002, 50(5):1170-1183.

Biographies:

SHAO Kai(1977-) ， male ， comes from Yunnan． He is an Associate Professor of Chongqing University of Posts and Telecommunications，China． His main research interests are multicarrier modulation technique for mobile communication，multi-rate digital signal processing and filter bank theory，and related topics． E-mail: shaokai@cqupt.edu.cn.

LI Yan (1991-) ，female，comes from Anhui Province． She is an MSc student at Chongqing University of Posts and Telecommunications．Her research interests are multicarrier modulation techniques for mobile communication and filter bank theory. E-mail: aaa7014525@qq.com.

MEN Lijun(1963-), female, comes from Penglai Shandong province. She is a Consultant-Director of Chongqing Municipal Committee. Her research interests are communication engineering and Communication system operation and maintenance, ect. E-mail:591350175@qq.com.

(编辑：魏琴芳)

基于DFT-FB改进的F-OFDM方案

邵 凯1，李 艳1,门利军2

(1.重庆邮电大学，重庆移动通信重点实验室，重庆 400064; 2. 重庆市委政法委办公室,重庆 401147)

通过对滤波OFDM(filtered-OFDM，F-OFDM)的研究，针对F-OFDM使用较长的子带滤波器降低带外泄漏(out-of-band emission，OOBE)而导致系统实现复杂度较高的问题，基于滤波器组(filter bank)的多相网络(poly phase network，PPN)实现降低复杂度和对每一个子载波进行独立的滤波降低OOBE的思想，提出了使用子载波滤波的滤波器组技术对F-OFDM的改进方案，实现对每个子载波的滤波处理。给出了改进方案的系统结构图并分析了系统原理，通过对改进方案系统的分析及带宽效率和系统实现复杂度的推理计算，得到改进方案可以达到与F-OFDM相同的带宽效率，且在相同的功率谱密度时，改进方案的实现复杂度比F-OFDM系统的低。仿真结果表明，相较于F-OFDM，改进方案不仅可以优化系统处理业务的灵活性，而且在AWGN(additive white Gaussian noise)信道和衰落信道下，在传输性能和抗信道衰落能力方面，也优于F-OFDM系统。

循环前缀正交频分复用(CP-OFDM); DFT滤波器组; 滤波OFDM

2016-08-19

2017-04-09

李 艳 aaa7014525@qq.com

TN925.5DocumentcodeAArticleID1673-825X(2017)05-0618-07

s：The Scientific and Technological Research Program of Chongqing Municipal Education Commission (KJ1400437); The National Science and Technology Major Project (2016ZX03001010)

10.3979/j.issn.1673-825X.2017.05.007