, pp.22-27 http://dx.doi.org/10.14257/astl.2013 Fractional Delay Filter Based Wideband Self- Interference Cancellation Hao Liu The National Communication Lab. The University of Electronic Science and Technology of China, Chengdu, China liuhao@uestc.edu.cn Abstract. This paper introduce Fractional Delay Filter (FD) into Self- Interference Cancellation based Full-Duplex communication system. The paper explains that an accurate delay of baseband signal in cancellation channel is very important for the performance of self-interference cancellation. It emphasise that a delay other than a phase alignment can extend the cancellation to wideband with reasonable performance. Simulations proved the feasibility of application of FD in self-interference based Full-Duplex system. Keywords: Full-Duplex; Fractional Delay Filter; Self-interference Cancellation 1 Introduction Most communication systems use Time Division Duplexing (TDD) or Frequency Division Duplexing (FDD); they require two separate channels to achieve full duplex, which separate the uplink and downlink channels in the time domain and frequency domain, respectively. In recent years, Technologies based on cancellation that can transmit and receive data at the same time and same frequency have been investigated and demonstrated. These technologies broke the basic percept that a radio cannot transmit and receive signal on the same frequency at the same time, it can potentially double the channel capacity in the ideal case. The basic idea to achieve Full-Duplex is to use an additional transmitting channel. The additional transmitting channel generate signal called cancellation signal from the transmitting signal, and add it with the receiving signal at the receiving antenna or RF circuits. The receiving signal consists of interesting signal from other nodes and the feedback of transmitting signal from itself. Ideally, the cancellation signal is a perfect inverse of the feedback. When it adds with the receiving signal, it cancels the feedback and only interesting signal goes to analog to digital converter. Experiments show that 70dB~80dB of self-interference suppression is needed in short-range communication to achieve full-duplex communication. Some encouraging research results can be found in [1], [2]. In [1], Two transmit antenna and one receive antenna is used, the two transmit antenna are placed at distance d and d 2 away from the receive antenna, the is the wavelength. As ISSN: 2287-1233 ASTL Copyright 2013 SERSC
the two transmit antenna is offset by 2, the signals from them cancel each other. This method is also considered as beam forming, it creates null position at where the receive antenna is. In [2], Researchers use Analog Cancellation and Digital Cancellation together for larger attenuation of the self-interference signal. The most interesting part is the compensation of phase in the additional transmit channel. With this phase compensation, the cancellation signal is the inverse of the feedback signal. Experiments prove that cancellation based Full-Duplex is feasible in short-range communication. When power spectrum of transmit signal is narrow band, Methods in [1], [2] have good enough self-interference cancellation performance in short range communication. However, when signal is wide band, their performance is very limited. For example, when signal bandwidth is less than 0.625MHz, the prototype in [2] has a enough performance. The performance of method in [1] is sensitive to antenna position and carrier frequency, and it has reasonable performance only for narrow band signal too. With narrow band signal limit, they cannot be used in WI- FI/Bluetooth, as these standards have wideband signal. As shown in [3], selfinterference cancellation method in [2] can be extended to wideband by applying signal phase alignment per subcarrier. However, this phase alignment per subcarrier method is feasible only in OFDM system. This paper introduce fractional filter into self-interference cancellation based Full- Duplex system to extend it to wideband signal, and it can be applied in any system without restrict of multi-carrier system. The rest of this paper is organized as follows. In Section 2, we analyzed the system model of the self-interference cancellation. Results indicate that a very small delay of the transmit signal in additional cancellation channel is a perfect way to generate cancellation signal. In section 3, we describe the fractional filter based selfinterference cancellation method. As an ideal fractional delay filter is not realizable, we used a simple truncated filter as an approximation, and proved its feasibility to extend the self-interference cancellation to wideband for any type of signal. 2 System Model of Self Interference Cancellation Fig. 1 shows the system model of a self-interference cancellation node. The transmit baseband signal X t is split into two flows. One of them is converted to analog, goes to TX, and then radiate from TX antenna. Another one is processed by block f, and then converted to analog, and finally adds with signal from Rx antenna through the air channel. Here, we only consider the path delay through air channel from TX to, and ignore all other non-ideal characteristics of the such as nonlinear, IQ impairment, etc. that is to say the circuit is perfect linear and memory less, and also we consider the air channel has only one direct path. Then signals in Fig. 1 can be expressed as formula (1)~(3) 23 Copyright 2013 SERSC
Vol.44(Networking X t Tx XRF t Tx Ant. f Tx t Y t - Rx ADC + A X RF Rx Ant. t Fig. 1. System model of self-interference cancellation j ct (1) X RF t X t e The baseband signal j ct C RF t f X t e AX RF t jc t AX t e jc jct AX t e e (3) X t consist of components with different frequency. To analyze the system s spectrum response, now we consider it as a single frequency j st s signal, i.e. X t e. Substitute X t with it in equation (3), AX RF t jst s jc jct Ae e e js jsts jc jct Ae e e e (4) Ideally, processed by f, the t should equal to AX RF t, that results in perfect cancellation of the feedback signal. In [2], [4], the process f is to phase align on the baseband signal. In this case, the t is as j jct t AX te e jsts j jct A e e e (5) When, compared with equation (4), we can easily found that there is a j e S difference between equation (4) and (5) in the baseband signal. This difference adds an extra phase offset on receive signal, and it changes with the frequency of baseband signal. When the frequency of baseband is small, this difference is small too. However, if the frequency of baseband signal is big, this difference is considerably big. Then the cancellation is poor. That s why the prototype in [2] has a good self-interference cancellation performance in a narrow (2) Copyright 2013 SERSC 24
band only. And it also explains that why the phase alignment per carrier in [3] can extend to wideband. As the propagation delay is less than the sample interval of baseband signal, a conventional digital delay line is not practical. To solve this problem, we propose to use a fractional delay filter in the cancellation channel to delay the signal for fraction of sample interval. X t Tx XRF t Tx Ant. FD Cancellation t OSC j e - Y t Rx ADC A X RF t + Rx Ant. Fig. 2. Fractional Filter based self interference cancellation system The fractional filter based self-interference cancellation system is depicted in Fig. FD now, the fractional filter can 2. The function f is a fractional delay filter delay signal for any. Output of the local oscillator is phase shifted by it goes to Cancellation of cancellation channel. Then the C t is with t j c t A X t e RF j e (6) Compare equation (6) with equation (3), the AX RF t before is exactly same t. Ideally, The feedback from transmit channel can be perfectly canceled in Rx even if the baseband signal X t is a wideband signal. 3 Self Interference Cancellation Performance of Fractional Delay Filter Fractional delay means, with given uniform sampling data, a delay that is a noninteger multiple of the sample interval. Digital fractional delay (FD) filters provide a useful building block that can be used for fine-tuning the sampling instants. [5] We simulated the performance of FD based self-interference cancellation. In the simulation, we assume the frequency of carrier is 2.4GHz, the sample rate of -9 baseband signal is 100MHz, the delay from TX antenna to Rx antenna is 310 s, which corresponds to fractional delay 0.3 of the sample interval. The results of the simulation are as TABLE I. it is obvious that performance improved as filter order increases, for the reason that the approximation error decreases. When order of FD increases, the effective bandwidth of the FD is also 25 Copyright 2013 SERSC
Vol.44(Networking extended. When bandwidth of baseband signal is less than 500KHz, the performance of Phase Align method is better than FD filter based method. But when bandwidth is larger than 500KHz, the performance of Phase Align method decrease faster as bandwidth increase. Table 1. Self-Inteference Cancellation Performance (db) FD based Method Frquency of Phase Align (Orders) Baseband signal Method 11 21 31 61 100KHz -48.98-51.65-51.89-51.98-54.49 200KHz -46.89-48.35-48.49-48.63-48.47 500KHz -41.26-41.73-51.93-42.9-40.51 1MHz -35.82-36.39-37.17-41.8-34.49 2MHz -26.65-32.46-36.19-45.13-28.47 5MHz -35.51-46.75-35.54-52.23-20.52 10MHz -28.65-41.28-44.48-49.09-14.51 20MHz -28.73-34.58-38.04-43.85-8.525 4 Conclusion and Future Work The work above proved that the Fractional Delay Filter can be used in selfinterference cancellation based Full-Duplex system, and it can extend effective bandwidth for any kind of signal without the multi-carrier assumption. Acknowledgment This work is supported in part by National Grand Special Science and Technology Project of China under Grant No. 2010ZX03006-002-02, the Important National Science and Technology Specific Projects of China under Grant 2013ZX03001024, and the Fundamental Research Funds for the Central Universities (China) under Grant ZYGX2010X002. References 1. J. Il Choi, M. Jain, K. Srinivasan, P. Levis, and S. Katti, Achieving single channel, full duplex wireless communication, in Proceedings of the sixteenth annual international conference on Mobile computing and networking - MobiCom 10, 2010, p. 1. 2. M. Duarte and A. Sabharwal, Full-duplex wireless communications using off-the-shelf radios: Feasibility and first results, in 2010 Conference Record of the Forty Fourth Asilomar Conference on Signals, Systems and Computers, 2010, pp. 1558 1562. 3. A. Sahai, G. Patel, and A. Sabharwal, Pushing the limits of Full-duplex: Design and Realtime Implementation, p. 12, Jul. 2011. Copyright 2013 SERSC 26
4. M. Duarte, C. Dick, and A. Sabharwal, Experiment-Driven Characterization of Full-Duplex Wireless Systems, IEEE Transactions on Wireless Communications, vol. 11, no. 12, pp. 4296 4307, Dec. 2012. 5. V. Valimaki and T. Laakso, Principles of fractional delay filters, in 2000 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No.00CH37100), 2000, vol. 6, no. June, pp. 3870 3873. 27 Copyright 2013 SERSC