The Vault

A Resource Block Based Filtered OFDM Scheme and Performance Comparison
Research Paper / Feb 2014

Abstract— Dynamic spectrum sharing is a key technique to meet the high data rate demands of future wireless systems evolving towards heterogeneous networks. To make the best use of available frequency resources without creating interference to neighbor nodes, spectrally contained and scalable waveforms are needed. While OFDM has become the most popular multicarrier modulation (MCM) scheme, it suffers from high side lobes. Filtering is a popular technique used to reduce the out-of-band emission of OFDM. However, to utilize non-contiguous spectrum fragments, filters need to be dynamically designed for each fragment, making the use of filtered OFDM challenging. On the other hand, OFDM-OQAM, a filter bank based multicarrier (FBMC) modulation, has emerged as an alternative to OFDM to offer better spectral containment and more flexibility for dynamic spectrum allocations. In this paper, we first present a novel technique, called resource block (RB) filtered OFDM (RB- F-OFDM), which divides the available spectrum fragments into chunks of contiguous subcarriers, referred to RBs, and generates and filters the signal transmitted on each RB individually. This approach has the advantage of being modular and scalable since the same transmit and receive modules are used for all RBs. Then, we compare the performance of OFDM, filtered OFDM, RB F-OFDM, and OFDM-OQAM. The simulations are performed under various adjacent channel interference (ACI) conditions. Simulation results show that OFDM-OQAM offers the best performance under channels with moderate delay spread and ACI; however, it has the undesired high latency and its signal is not backwards compatible to legacy OFDM receivers. The performance of RB F-OFDM is similar to filtered OFDM while adding support for non-contiguous spectrum, making it a viable solution for dynamic spectrum sharing systems. Index Terms—Multicarrier, OFDM, filtered OFDM, FBMC, OFDM-OQAM, resource block filtered OFDM I. INTRODUCTION In today’s wireless communication networks, spectrum usage is rigidly allocated and licensed by government regulations, limiting access to specific uses and licensees. Various radio access technologies and baseband waveforms are designed according to such regulations. Networks that have evolved under these regulations and standardization processes operate independent of one another, e.g. WLAN [1] vs. cellular networks such as LTE [2]. It is recognized that such tight spectrum allocation control and autonomous network operation does not lend itself to sharing available spectrum and results in underutilization of available spectrum [19]. Therefore, it is envisioned that future wireless communication networks will enable improved spectrum utilization by use of dynamic spectrum allocation [18], [19]. For example, a cognitive radio network permits such access by enabling nodes to identify and use spectrum that may be dynamically shared among multiple nodes. When spectrum utilization is increased and assignment of adjacent channels is common, the adequacy of current radio access technologies, or more specifically, the baseband waveforms, designed for legacy systems must be questioned for the future networks. Multicarrier modulation (MCM) techniques enable transmission of a set of data over multiple narrow band subcarriers simultaneously. Orthogonal Frequency Division Multiplexing (OFDM) [10] has become the most widely used baseband MCM waveform in modern wireless communication systems. However, OFDM suffers from large side lobes that contribute to undesired out-of-band emission and large peak- to-average power ratio (PAPR). Excessive out-of-band emissions partially contributed from large side lobes at the baseband can lead to strong adjacent channel interference (ACI), especially in small cell systems such as those proposed for use in TV white space and heterogeneous networks [3]. Filter bank multicarrier (FBMC) modulation is a family of MCM techniques proposed as an alternative to OFDM to overcome the aforementioned drawbacks. OFDM-Offset QAM (OFDM-OQAM) is a popular FBMC scheme [4]-[9]. In OFDM-OQAM, subcarriers of the signal overlap to achieve a high spectral efficiency. Different from OFDM, the real and imaginary parts of the QAM symbols are processed separately with 2×symbol rate. A prototype filter needs to be carefully designed to minimize or zero out ISI and ICI while keeping the side lobes small. Even though OFDM-OQAM has much better spectral containment, it introduces undesired inherent latency of a few symbol durations, and its transmit signal is not backwards compatible to legacy OFDM receivers. One common technique used to reduce the out-of-band emission of OFDM is to use filtering. However, to utilize non- contiguous spectrum fragments, filters need to be dynamically designed for each fragment, making the use of filtered OFDM challenging [20]. To overcome this challenge, we first propose a new technique where the available spectrum fragments are divided into chunks of contiguous subcarriers, referred to as resource blocks (RB), and the signal transmitted on each RB is generated and filtered individually. This approach has the advantage of being modular and scalable since the same transmit (or receive) module is used for all RBs. Moreover, its transmitted signal is backwards compatible to legacy OFDM receivers and its receiver could also demodulate legacy OFDM signals. Then, the raw BER performance of the aforementioned MCM schemes is compared under various ACI and channel delay spread conditions. A Resource Block Based Filtered OFDM Scheme and Performance Comparison Jialing Li, Kenneth Kearney, Erdem Bala, Rui Yang InterDigital Communications, LLC, Melville, NY The rest of the paper is organized as follows: In Section II, a brief overview of OFDM, filtered OFDM, OFDM-OQAM is given. In Section III, the proposed technique, RB filtered OFDM (RB F-OFDM), is presented. In Section IV, the simulation setup is described and simulation results are presented. Finally, Section V concludes the paper. II. FBMC SYSTEMS The transmitted signal, x[n] in a general FBMC scheme, can be expressed as ݔሾ݊ሿ ൌ ෍ ෍ ݏ௞,௟݃௞ሾ݊ െ ݈ܮ௦ሿ ெିଵ ௞ୀ଴௟ (1) where M is the number of subcarriers, s௞,௟ is the lth symbol in the kth subcarrier, Ls is the number of samples per transmit symbol spacing, and ݃௞ሾ݊ሿ is the synthesis filter for the kth subcarrier. At the receiver, the estimated lth symbol ̂ݏ௞,௟ in the kth subcarrier is ̂ݏ௞,௟ ൌ ሺݕሾ݊ሿ כ ௞݂ሾ݊ሿሻ௡ୀ௟௅ೞ (2) where y[n] is the received signal and ௞݂ሾ݊ሿ is the analysis filter for the kth subcarrier. OFDM is a well known MCM technique where the synthesis and analysis filters are implemented by IFFT and FFT [9]. To suppress the high side lobes of an OFDM signal, one of the commonly used techniques is filtering, resulting in filtered OFDM. In OFDM-OQAM, complex valued input symbols are separated into the real and imaginary parts and transmitted with ܮ௦ ൌ ܮ/2 whereܮ is the number of sample per symbol duration. A π/2 phase shift is introduced between any pair of adjacent OFDM-OQAM symbols in both frequency and time to ensure orthogonality [4], [11]. However, OFDM-OQAM has the undesired large inherent latency and its signal is not backwards compatible to legacy OFDM receivers. III. RESOURCE BLOCK FILTERED OFDM In many cases, the spectrum may be fragmented and the available spectrum may be non-contiguous as shown in Figure 1. In such cases, a filter needs to be designed for each available chunk of spectrum, resulting in a big challenge for implementation of filtered OFDM especially when the spectrum availability changes dynamically. Figure 1 Utilization of fragmented spectrum To overcome this disadvantage, we propose a new MCM scheme to utilize non-contiguous spectral holes. In this method, called resource block filtered OFDM (RB F-OFDM), the available spectrum is divided into chunks of contiguous subcarriers, called a resource block (RB) as shown in Figure 1. The signal for each RB is generated and filtered individually and the sum of all signals is added to get the final transmitted signal. Since the same filter can be used for all RBs, spectral holes can be dynamically utilized without posing a big challenge in implementation. { × nfj ke π2 Figure 2 RB F-OFDM transmit module per RB The transmit module for one RB is shown in Figure 2. Specifically, the nth data vector for the kth RB is padded with ሺLଵ െ N) zeros, where N is the number of subcarriers in an RB and ܮଵ ൒ ܰ, and goes through an ܮଵ-point IFFT, where ܮଵ is a power of 2 and ܮଵ ൑ ܮ. After cyclic prefix (CP) insertion and upsampling by ܳ ൌ ܮ/ܮଵ, the signal is filtered with a lowpass filter ݌ሾ݊ሿ and modulated to the kth RB’s frequency fk. The final transmitted signal is then ܠ ൌ ∑ ܠ௞௄ିଵ௞ୀ଴ , as illustrated in Figure 3. At the receiver side, the received signal is demodulated from fk to baseband, filtered, and downsampled before going through an OFDM based receiver as shown in Figure 4 to form the demodulated stream which then is equalized by a one-tap frequency domain equalizer (FDE). The same receive module is used to receive data transmitted on each RB. { Figure 3 RB F-OFDM transmitter block Figure 5 illustrates the advantage of RB F-OFDM over filtered OFDM when non-contiguous spectrum fragments are used for transmission. In this figure, two available spectrum fragments are separated by an unavailable band. When filtered OFDM is used with a filter over the whole band, we can see that the out-of-band emission to the unavailable band between the available fragments is the same as that of OFDM [20]. However, RB F-OFDM can utilize the available fragments with less out-of-band emission to the unavailable band. × nfj ke π2− Figure 4 RB F-OFDM receive module per RB Figure 5 Power spectral density of MCM schemes with fragmented spectrum IV. PERFORMANCE COMPARISON OF THE MCM SCHEMES The performance of the four MCM schemes is compared with and without adjacent channel interference. For the evaluation, a link level simulation tool compliant with the 3GPP Release 8 LTE specifications [2], [12], and [13] are used. Specifically, data bits are first coded and then modulated before being mapped to the specific subcarriers. Then, the transmit signal is generated by using one of the MCM schemes. The generated signal goes through a frequency selective fading channel. At the receiver side, the signal is first processed by the appropriate receiver and then data symbols are demodulated and decoded. The simulation parameters are presented in Table 1. The ACI is generated using the same waveform as the desired signal. The frequency gap between the bands of the desired signal and interference signals is set to 1.5Δf, where Δf is the subcarrier spacing. The power difference between the ACI and the desired signal is a parameter and denoted as ΔPACI. In the simulations, there is either no ACI or with ACI of ΔPACI being 10 or 30 dB since the power difference between received signals from two different types of nodes could be in this range [3]. Table 1 Simulation parameters Parameter Settings Considered MCM schemes OFDM, Filtered OFDM, RB Filtered OFDM, OFDM-OQAM Δf 15 KHz RB size 12 subcarriers Channel models EVA: Extended vehicular A channel with maximum Doppler frequency 5 Hz [16] ΔPACI (dB) No ACI 10 30 Bandwidth (MHz) 10 MHz (5MHz for data, 5MHz for ACI) FFT size 1024 for OFDM, filtered OFDM, and OFDM- OQAM 128 for RB F-OFDM Transmit/Receive filter Filtered OFDM 35-tap RRC RB F-OFDM 53-tap equal ripple OFDM-OQAM Frequency sampling prototype filter with overlapping factor of 4 [11] Equalizer 1-tap FDE Channel estimation Ideal The raw BER performance in EVA channel using 16QAM modulation, without ACI and with ACI of ΔPACI = 10 dB are shown in Figure 6 and Figure 7, respectively. We can observe that in the moderately selective EVA channel, with no ACI, OFDM-OQAM with 1-tap FDE performs close to OFDM at low to medium SNR and exhibits an error floor at high SNR. Loss of orthogonality for OFDM-OQAM under fading conditions results in ISI and ICI which limits the simple 1-tap FDE’s ability to recover the data symbols optimally [14]. More complex equalizers, such as the multi-tap MMSE equalizer, could be investigated to improve the performance as discussed in [14] and [15]. The performance of filtered OFDM is similar to OFDM and OFDM-OQAM at low to medium SNR, while it is slightly worse than OFDM at high SNR. For OFDM, the CP provides the adequate buffer to compensate for the channel but, the equivalent channel including the transmit and receive filters exceeds this CP buffer resulting in ISI and ICI which degrades the performance at higher SNR. The performance of RB F-OFDM is similar to the other schemes at low to medium SNR, while is the worst at high SNR, because it has longer transmit and receive filters. Similar observations hold when 64QAM is used as illustrated in Figure 8 for the case without ACI. When ACI with ΔPACI = 10dB exists, OFDM-OQAM and filtered OFDM perform close to each other, and both outperform OFDM. OFDM-OQAM offers slightly better performance than filtered OFDM due to its better ACI rejection capability. The performance of RB F-OFDM is slightly worse than that of filtered OFDM at high SNR mainly due its longer filter. Similar observations hold when 64QAM is used as illustrated in Figure 9. Figure 6 BER with 16QAM, EVA channel, and no ACI Figure 7 BER with 16QAM, EVA channel, and ACI of ΔPACI = 10 dB Finally, Figure 10 and Figure 11 illustrate the BER comparison of the MCM schemes with ΔPACI = 30 dB and 16 QAM, and 64QAM, respectively. As we can see from these figures, the performance of RB F-OFDM is approximately same as filtered OFDM with 16QAM, and is better than filtered OFDM with 64QAM. The reason for the performance gap with higher modulation order is due to the fact that filtered OFDM uses a shorter filter which has worse spectral containment and weaker ACI rejection capability. If the filter is optimized, we expect filtered OFDM and RB F-OFDM to have approximately similar performance. Among all MCM schemes, OFDM-OQAM is most robust to ACI due to good spectral containment and thus provides the best performance. We can conclude that, in the moderately selective EVA channel, OFDM-OQAM, filtered OFDM and RB F-OFDM mitigate ACI more effectively than OFDM. If ACI power further increases, OFDM-OQAM outperforms the other schemes. However, as pointed out earlier, OFDM-OQAM has the undesired large latency and its signal is not backwards compatible to legacy OFDM receivers. On the other hand, when ACI exists, the performance gap between RB F-OFDM and filtered OFDM is not large and the performance of the RB F-OFDM gets further closer to that of filtered OFDM as the ACI power increases. But, as illustrated in Figure 5, RB F- OFDM has the advantage of being more spectrally agile and can be used to transmit on any unused spectrum fragment. Figure 8 BER with 64QAM, EVA channel, and no ACI Figure 9 BER with 64QAM, EVA channel, and ACI of ΔPACI = 10 dB V. CONCLUSIONS In this paper, we have proposed a new multicarrier modulation scheme called RB F-OFDM. The proposed scheme enables utilization of non-contiguous spectrum, so is more spectrally agile than filtered OFDM. Moreover, the RB F-OFDM transmit signal is backwards compatible to legacy OFDM receivers and the RB F-OFDM receiver could also demodulate legacy OFDM signals. Considering these features, RB F-OFDM presents itself as a viable candidate for future wireless communication networks; specifically cognitive radio systems where efficient spectrum sharing is important. The performance of the proposed scheme has been evaluated and compared with OFDM, filtered OFDM, and OFDM-OQAM under various ACI conditions. Without ACI, OFDM offers the best performance while the performance of filtered OFDM and RB F-OFDM degrade at high SNR compared to OFDM due to fact that the CP length is not enough to remove the ISI completely when filters are used. The performance of OFDM-OQAM degrades since the one- tap equalizer could not optimally mitigate the ISI and ICI in the moderately or highly frequency selective channels. With ACI, OFDM-OQAM offers the best performance under channels with moderate delay spread and ACI. But it has the undesired large latency and its signal is not backwards compatible to legacy OFDM receivers. In all cases, the performance of the RB F-OFDM scheme is close to that of filtered OFDM and the performance gap between the two further decreases as the ACI power increases. Although filtered OFDM presents itself as an attractive alternative to OFDM-OQAM, multiple and adaptable filters would need to be implemented to utilize spectral holes in a fragmented spectrum when using filtered OFDM. Future work includes the BER and throughput analysis of the MCM schemes with non-contiguous spectrum, and filter optimization. Figure 10 BER with 16QAM, EVA channel, and ACI of ΔPACI = 30 dB Figure 11 BER with 64QAM, EVA channel, and ACI of ΔPACI = 30dB REFERENCES [1] IEEE Std 802.11™-2012, Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications, 2012. [2] 3GPP TS36.211 V8.9.0 (2009-12), “E-UTRA; Physical channels and modulation.” [3] 3GPP TR 36.921 V11.0.0 (2012-09), FDD Home eNode B (HeNB) Radio Frequency (RF) requirements analysis. [4] P. Siohan, C. Siclet and N. Lacaille, “Analysis and design of OFDM/OQAM systems based on filterbank theory,” Signal Processing, IEEE Transactions on, vol.50, no.5, pp.1170-1183, May 2002. [5] T. H. Stitz, Filter Bank Techniques for the Physical Layer in Wireless Communications, Phd dissertation, Tampere University of Technology, 2010. [6] B. Farhang-Boroujeny, “OFDM versus filter bank multicarrier,” IEEE Signal Processing Magazine, pp. 92-112, May 2011. [7] G. Lin, On the design and optimization of OFDM systems, Phd thesis, Norwegian University of Science and Technology, 2006. [8] B. Hirosaki, “An orthogonally multiplexed QAM system using the discrete Fourier transform,” IEEE Trans. Communications, vol. 29, no. 7, pp. 982–989, July 1981. [9] J. Du and S. Signell, “Classic OFDM Systems and Pulse Shaping OFDM/OQAM Systems,” February 2007, Online: http://www.ee.kth.se/~jinfeng/download/NGFDM_report070228.pdf. [10] L. Cimini, “Analysis and Simulation of a Digital Mobile Channel Using Orthogonal Frequency Division Multiplexing,” IEEE Trans. on Communications, July 1985, pp. 665-675. [11] A. Viholainen, M. Bellanger and M. Huchard, “Prototype filter and structure optimization,” Physical Layer for Dynamic Spectrum Access and Cognitive Radio, January 2009. [12] 3GPP TS36.212 V8.8.0 (2009-12), “E-UTRA; Multiplexing and channel coding.” [13] 3GPP TS36.213 V8.8.0 (2009-09), “E-UTRA; Physical Layer Procedures.” [14] J. Louveaux, L. Baltar, D. Waldhauser, M. Renfors, M. Tanda, C. Bader, and E. Kofidis, “Equalization and demodulation in receiver (single antenna),” PHYDYAS report, July 2008. Online: http://www.ict- phydyas.org/delivrables/PHYDYAS-D3.1.pdf. [15] D. S. Waldhauser, L. G. Baltar, and J. A. Nossek, “MMSE subcarrier equalization for filter bank based multicarrier systems,” in Proceedings of IEEE Workshop on Signal Processing Advances in Wireless Communications, July 2008. [16] V. Erceg, et al, "Channel Models for Fixed Wireless Applications," IEEE 802.16.3c-01/29r4, July 2001, Online: http://www.ieee802.org/16/tg3/contrib/802163c-01_29r4.pdf. [17] 3GPP TS 36.101V8.19.0 (2012-09), “E-UTRA; User Equipment (UE) radio transmission and reception.”. [18] Z. Ji and K. Liu, “Dynamic spectrum sharing: A game theoretical overview,” IEEE Communications Magazine, vol. 45, no. 5, pp. 88–94, May 2007. [19] B. Wang and K. J. R. Liu, “Advances in cognitive radio networks: A survey,” IEEE Journal of Selected Topics in Signal Processing, vol.5, no.1, pp.5-23, February 2011. [20] D. Noguet, M. Gautier and V. Berg, “Advances in Opportunistic Radio Technologies for TVWS,” EURASIP Journal on Wireless Communications and Networking, 2011, pp. 1-12.