Cosine

Modulated Filterbank Technique for Cognitive UWB Rajarshi Mahapatra Applied Research Group Satyam Computer Services Ltd. Bangalore, India E-mail: rajarshi_mahapatraWsatyam.com

Abstract-Ultra wideband (UWB) communication systems can achieve high data rate by utilizing high spectrum bandwidth. This signal typically overlaps with many existing and potentially futuristic services. In the recent years, cgnitive radio (CR) has been an emerging approach for a more efficient usage of the precious radio spectrum resources. Therefore, an UWB device will successful if it has the intelligence of the cognitive radio in terms of spectrum usage. In this paper, we study different aspects of cosine modulated filterbank (CMFB) techniques over other techniques, like OFDM and multitaper method, the popular candidate schemes for spectral sensing. We intend to justify that cognitive system based on CMFB technique is more resource conscious than the traditional OFDM technique. Our study indicates that the filterbank technique is more powerful than the other techniques, when used spectral analysis. Hence, the CMFB can be an ideal candidate for cognitive UWB systems. Keywords- UWB, Fiterbank, OFDM

I. INTRODUCTION Ultra-Wideband (UWB) technology is one of the promising solutions for next -generation wireless communications to support high data rate in short distance. UWB systems spread the transmitted signal power over an extremely large frequency band, and the power spectral density (PSD) of the signal is flat across the frequency band. Due to the wide bandwidth of the transmitted signal, UWB radio will be sharing the environment with other radio systems, some possibly creating UWB multiple access interference, and others creating narrowband interference in the UWB radio bands.

The FCC allocates 3.1-10.6 GHz spectrum band for UWB communication and have set the conditions that UWB devices occupy more than 500 MHz of bandwidth. These spectrum allocations of UWB devices include IEEE 802.1 la devices (5 GHz band) and WiMax devices (3 and 5 GHz bands) and Japan astronomy band around 4 GHz. They have also regulated the limit of interference from UWB radiators to other radio systems in accordance with the specified spectrum mask. Recently, multi band orthogonal frequency division multiplexing (MB-OFDM) is proposed for UWB communication by considering all the above facts [1, 2]. In this multiband OFDM system very simple equalization is made possible by converting the broadband frequency selective channel into a set of parallel flat-fading subchannels. This is achieved using the inverse fast Fourier transform (IFFT)

1-4244-1005-3/07/$25.00 02007 IEEE.

processing and by inserting a time domain guard interval, in the form of a cyclic prefix (CP), to the OFDM symbols at the transmitter. By dimensioning the CP longer than the maximum delay spread of the radio channel, interference from the previous OFDM symbol, referred to as inter-symbolinterference (ISI) will only affect the guard interval. At the receiver, the guard interval is discarded to elegantly avoid ISI prior to transforming the signal back to frequency domain using the fast Fourier transform (FFT). These simple techniques make OFDM popular. Its inherent spectrum sensing nature, always tracks neighboring spectrum and make use of it for communication purpose. This feature is incorporated in MB- OFDM, which makes the technique cognitive and one can say MB-OFDM is a technique for cognitive UWB radio.

The term "cognitive radio" was first coined by Mitola [3]. It is an approach for a more efficient usage of the precious radio spectrum resources. A cognitive radio use the un-used spectrum by learning the environment, and by managing and adapting to various dimensions of time, frequency, space, power, and coding. A CR enable secondary users are allowed to transmit and receive data over portions of spectra when primary users are inactive. In other words, CR can avoid a particular portion of spectrum, which is already allocated for some other purpose, with the help of its inherent capabilities and switch over to another portion of spectrum. This inherent property of CR is very useful for implementation of cognitive UWB system. Since, other frequency domains are also present in UWB spectrum band; it is an essential requirement for an UWB device to have cognitive properties. MB-OFDM system is considered as one of the candidates for cognitive UWB. But OFDM technique has two main disadvantages, which includes spectrum leakage and long cycle prefix. A larger number of zero tones are required to create a notch with significant depth. An improved technique has been used to enhance the bandwidth efficiency by using zero padding instead of null tones. On the other hand, OFDM system needs cycle prefix (CP) to mitigate fading, which increases the overhead and decreases the effective data rate [1] [2]. While enabling a very efficient and simple way to combat multipath dffect, the CP is pure redundancy, which decreases the spectral efficiency of the system. As a consequence, there has recently been a growing interest towards alternative multicarrier schemes, which could provide the same robustness without requiring a CP, thus offering improved spectral efficiency [4] [5]. In [4], a channel equalization technique has

Figure 1. Block schematic of CMFB-MCM transceiver

been studied in filter bank multicarrier modulation by considering exponentially modulated filter bank (EMFB). They have also compared the result with OFDM with certain simulated parameters. In [5], the authors have proposed a cosine-modulated filterbank has blind detection capability and have compared the simulation results for CMFB against OFDM-based system. Block schematic of CMFB-MCM transceiver The above efforts have motivated us to consider the filterbank technique for UWB systems. In this paper, we study various aspects of filter bank techniques and subsequently, investigate its advantages over OFDM technique. Furthermore, the inherent spectrum sensing properties of CMFB transforms the UWB system into cognitive one. From our study, it has been found that the filterbank technique has several advantages over OFDM. This indicates that the filterbank technique can be a strong candidate for cognitive UWB systems in the precious radio environment. The rest of the paper is organized as follows. Section II describes the basic principle of CMFB-based system. Section III presents the comparisons between CMFB and OFDM system. The cognitive nature of CFMB systems are described in Sec. IV. Finally, Sec. V concludes the work.

transmit signal. The received signal at the channel output is passed though an analysis CMFB demodulator which under perfect channel condition, i.e., when no distortion is induced by the channel, recovers the data symbols, perfectly. These are then passed through a decoder and parallel-to-serial converter (P/S), to generate the received bit stream of data. A detailed mathematical analysis of the proposed system is discussed in [5].

Figure 2 presents a block diagram for an M-band CMFB. Subsequently, a typical set of magnitude responses of the CMFB are shown in Fig. 3. As evident from the figure, the subcarrier bands are overlapped, and the separations between the bands are controlled through the judicial design of the CMFB filters. In [5], the authors have shown that each subchannel in CMFB carries a sequence of PAM symbols and in a CMFB-MCM system, each PAM sequence occupies a bandwidth of p/M, where M is the number of band in CMFB system. The bandwidth occupied by the CMFB signal is equivalent to that of a QAM signal with almost zero excess bandwidth. In receiver, the CMFB-MCM system uses a simplified version of blind equalizer developed in [5] to equalize the channels. The equalization in each subchannel of the CMFB-MCM receiver is achieved by using a two-tap equalizer. The simplicity of the equalizer is the key to the success of the blind equalization algorithnir and makes CMFB technique useful for wireless communication.

II.

OVERVIEW OF COSINE-MODULATED FILTER BANK TECHNIQUE CMFB was introduced as a multi-carrier modulation (MCM) technique for wideband data transmission over wireless channels, for data transmission over digital subscriber lines at the beginning under the common terminology of discrete wavelet multitone (DWMT). Due to unavailability of proper equalization, DWMT can not be used for wireless communication. Later on, it was also found to be useful for wireless communication with imp roved equalizer under the name of CMFB. In CMFB, data symbols are pulse-amplitude modulated (PAM), which uses single sideband (SSB) modulation for transmission. By using SSB, CMFB is considered a bandwidth-efficient system similar to quadrature amplitude modulated (QAM) systems. It is also noted that a CMFB signal is a combination of a number of vestigial sideband (VSB) signals packed together within the smallest possible bandwidth. The bandwidth occupied by the CMFB signal is equivalent to that of a QAM signal with almost zero excess bandwidth [5]. A schematic representation of a CMFB -MCM transceiver is shown in Fig. 1. The incoming stream of data bit first goes through a serial-to-parallel converter (S/P) and encoder block which generates a set of symbols at the modulator input. The CMFB modulator maps the data symbols to different carrier bands and adds all the modulated symbols to generate the

Figure 2. CM4FB structure for CM\FB-MCM transceiver.

l,

.i~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~o

Figure 3. Response of CMFB subband filter COMPARISON OF THE PROPOSED CMFB WITH MBOFDM FOR UWB COMMUNICATION In MB-OFDM system, each band of 528 MHz is considered for UWB communication [1]. As discussed in [1], the system blocks the astronomy band by sending the dummy tone. As mentioned in the literature, the MB-OFDM emissions can be reduced by desired 23 dB by expense of 120 MHz within the III.

3260-3267 MHz radio astronomy bands. This is done using 29 null tones. This is due to the spectrum leakage of OFDM system. This leakage can be further reduced by using zero padding. MB-OFDM systems use cycle prefix to combat the effect of multipath fading and delay spread. The prefix length is about 60 ns to multipath tolerance for the same. This is a redundant data, which contributes in the final data rate. Thus, due to spectrum leakage, the UWB system sacrifices some bandwidth to block the transmission at certain pre-used/preallocated frequency bands. Using cycle prefix further consumes some amount of the bandwidth, thereby reducing bandwidth availability for data communication significantly. On the other hand, CMFB scheme does not use cycle prefix, which is an essential component in OFDM technique. Subsequently, i uses simple blind equalization which has no counterpart in OFDM. Furthermore, he CMFB receiver is inherently a coherent detector, whereas OFDM uses coherent and non-coherent detector. Therefore, any comparison of CMFB with OFDM has to be based on an OFDM system with a coherent detector; because non-coherent detection incurs a loss of 3 dB in signal-to-noise ratio. OFDM systems with coherent detectors usually use pilot symbols for channel estimation and tracking. These extra components of OFDM increase the overhead of MB-OFDM UWB system and decrease the actual data rate. These overhead are not completely overcome in CMFB-based system but partly.

Thus, the CMFB technique has several advantages over OFDM for cognitive UWB communication. The CMFB is found to be more bandwidth efficient, since cyclic prefix used in OFDM are absent here. Moreover, pilot symbols that are commonly used in OFDM, for channel estimation/equalization, are also absent in CMFB as it has blind equalization capability [5].

Figure 4. Spectrum of OFDM signal.

The spectrum of CMFB and OFDM are shown in Figs. 3 and 4. It is evident from the figures that individual subband spectrum of CMFB is effected by adjacent bands only, whereas in OFDM, it is effected by the entire subcarrier spectrum. More interesting results are found in [4], which provide comparisons of BER and FER performances between filterbank and OFDM techniques. In this literature, among the various combination, the authois have compared between 2-PAM for filterbank and QPSK for OFDM. It has been found that due to the absence of time domain guard interval and reduced frequency domain

guard bands, higher spectral efficiency is achieved in filterbank multicarrier. This excess transmission capacity can be used to transmit more redundant data (lower coding rate) while maintaining similar information data rate compared to OFDM. This is favorable to the filterbank technique in the FER/BER performance comparison as somewhat less energy in filterbank is sufficient to result in similar error probability as compared to OFDM. In another study [9], it has been shown that the filterbank technique is less sensitive to frequency offset than OFDM. However, the computation complexity of filterbank techniques is higher then the OFDM-based system. In the case of CMFB-MCM, we note that since data symbols are PAM, in order to have the same number of bits per block in both OFDM and CMFB-MCM systems, the block Jngth M in CMFBMCM should be twice the block length in OFDM. Accordingly, the signaling rate in CMFB should be twice that in OFDM system. In addition to this, both modulator and demodulator in CMFB-MCM are more complex than their OFDM counterparts. Apart from these, filterbank technique has cognitive spectrum sensing capabilities like OFDM that supports its candidature as a physical layer technique for cognitive UWB communications. The cognitive properties of the filterbank technique are discussed in next section. IV. COGNITIVE RADIO PROPERTIES OF CMFB SYSTEM The demand of radio -frequency (RF) spectrum is increasing to support the user needs in wireless communication. Since the RF spectrum is limited, future wireless communication should be spectrally efficient. On the other hand, in a study, FCC found that 70% of allocated spectrum is unused. To utilize the unused spectrum and to support user demands, cognitive radio (CR) technique has been proposed. CR is an inclusive of software-defined radio, has been proposed as the means to promote the efficient use of the spectrum by exploiting the existence of spectrum holes [10]. The idea behind cognitive radio is that cognitive users will actively search the spectrum for available frequency bands, dynamically adjusting their transmissions so as to avoid interference with other users. Therefore, a cognitive radio system must be equipped with a spectrum pooling mechanism that continuously senses the radio activity in the environment and decides which parts of the spectrum are available and thus may be used by secondary users. This task should be performed with a very high probability of correct detection over all active frequency bands in order to minimize interference on primary users. To achieve this, Weiss et al [11] proposed a distributed spectrum pooling protocol where all the nodes (the base as well as mobile stations) participate in a channel sounding process. In essence, each node in the cognitive radio system is equipped with a spectrum analyzer for sensing the radio activity over the band of interest. This spectrum pooling enables the further utilization of already licensed frequency bands, which is the main aim of several regulatory authorities worldwide. The spectrum pooling mechanism enhances the spectral efficiency by overlaying a new mobile radio system on an existing one without requiring any changes to the actual licensed system. In

[2], where an OFDM based cognitive radio is considered, the same fast Fourier transformer (FFT) that is used for demodulation of the "payload" signals is also used for spectral analysis. Haykin [10], on the other hand, has noted the potential problems of spectral estimation using FFT and instead has proposed the Thomson's multitaper method (MTM) [12] as a better candidate.

Alternatively, filterbank techniques can be used as an efficient tool for spectral analysis. Some literature has been published in this regards. In [8], authors have presented a novel physical layer approach to multicarrier cognitive radios based on filter bank multicarrier modulation. In their approach, they use CMFB as cosine modulated mu ltitone (CMT). In another paper [13], authors shown that for spectrum analysis, filterbanks is more powerful than MTM by offering greatly reduced computational complexity. CONCLUSION In this paper, we study different aspects of filterbank techniques. Several literatures demonstrated that filterbank multicarrier modulation offers much higher spectral efficiency than OFDM when used as a data communication technique. Furthermore, when used spectral analysis, filterbank technique is more powerful than the other techniques. In addition to that filterbanks can adjust the interference temperature - a desirable goal suggested by the FCC for CR systems dynamically, which gives a full control over the spectral leakage. As a result, filterbank multicarrier modulation is a natural candidate for spectrum sensing and data communication in cognitive radio systems and also cognitive UWB system. By considering the above facts, in future we will design a cognitive UWB system based on the CMFB technique. This paper is our first attempt towards this goal. V.

REFERENCES [1] A. Batra, J. Balakrishnan, and A. Dabak, "Multi-band OFDM: A new approach for UWB," in Proc. IEEE Int. Circuits and Systems Symp., vol. 5, Vancouver, BC, Canada, May 2004, pp. 365-368. [2] A. Batra et al., "Design of a Multiband OFDM System for Realistic UWB Channel Environments," IEEE Trans. Microwave Theory and Techniques, vol. 52, no. 9, pp. 2123-2138, Sept. 2004. [3] J. Mitola et al., "Cognitive radio: Making software radios more personal," IEEE Pers. Commun., vol. 6, no. 4, pp. 13-18, Aug. 1999. [4] Tero Ihalainen, Tobias Hidalgo Stitz, Mika Rinne and Markku Renfors, "Channel equalization in filter bank based multicarrier modulation for wireless communications," EURASIP Journal on Advances in Signal Processing, pp. 1-18,2007. [5] B. Farhang-Boroujeny, "Multicarrier modulation with blind detection capability using cosine modulated filter banks," IEEE Trans. Commun., vol. 51,no. 12, pp. 2057-2070, Dec. 2003. [6] S. D. Sandberg and M.A. Tzannes, "Overlapped Discrete Multitone Modulation for High Speed Copper Wire Communications," IEEE J. Select. Areas Commun., vol. 13, no. 9, pp. 1571-1585, Dec. 1995. [7] G. Cherubini, E. Eleftheriou, S. Olcer, J.M. Cioffi, "Filter bank modulation techniques for very high speed digital subscriber lines," IEEE Commun. Mag., vol. 38, no. 5, pp. 98-104, May 2000. [8] P. Amini, R. Kempter, R. R. Chen, L. Lin and B. Farhang-Boroujeny, "Filter Bank Multitone: A Physical Layer Candidate for Cognitive Radios," Software Defined Radio Technical Conference, SDR 2005, November 14-18, 2005. [9] W. Rhee, J. C. Chung and J. Clin ini Jr., "Performance comparison of OFDM and multitone with polyphase filterbank for wireless communication," in Proc. of Veh. Technol. Conf, 18-21 May 1998 Volume: 2, pp. 768-772. [10] S. Haykin, "Cognitive radio: brain-empowered wireless communications," IEEE J Select Areas Commun., vol. 23, no. 3, pp. 201-220, Feb. 2005. [11] T.A. Weiss and F.K. Jondral, "Spectrum pooling: an innovative strategy for the enhancement of spectrum efficiency," IEEE Commun. Mag., Vol. 42, No. 3, pp. S8 - S14, March 2004. [12] Thomson, D.J., "Spectrum estimation and harmonic analysis," Proceedings ofthe IEEE, vol. 70, no. 9, pp. 1055-1096, Sept. 1982. [13] B. Farhang-Boroujeny and R. Kempter, "Multicarrier communication techniques for spectrum sensing and communication in cognitive radios", submitted to IEEE J. Select. Areas Commun.