Experimental Investigation of Signal Sensing with Overlapped FFT Based Energy Detection

Transcription

1 Wireless Pers Commun (2014) 77: DOI /s Experimental Investigation of Signal Sensing with Overlapped FFT Based Energy Detection Ryo Takai Shoya Uchida Akihiro Sato Mamiko Inamori Yukitoshi Sanada Published online: 30 November 2013 Springer Science+Business Media New York 2013 Abstract Overlapped FFT based energy detection has been proposed as a signal detection scheme in dynamic spectrum access. The overlapped FFT scheme increases the number of FFT frames to reduce the variance of squared noise and improve the detection probability. This paper evaluates the performance of the energy detection with overlapped FFT through experiments. In the experiments, different from the assumption in computer simulation of previous researches, a fixed distortion component caused by a direct current offset is observed. It is shown that the overlapped FFT scheme also works effectively under the existence of the fixed distortion. Numerical results obtained through the experiments show that the overlapped FFT scheme improves the detection probability by up to 0.15 with the noise and the fixed distortion component. The variance of the squared noise also reduces with the overlapped FFT scheme as it is expected in theoretical analysis when the fixed distortion is negligible. Keywords Cognitive radio Overlapped FFT Spectrum sensing R. Takai (B) S. Uchida A. Sato Y. Sanada Department of Electronics and Electrical Engineering, Keio University, Hiyoshi, Kohoku, Yokohama, Kanagawa , Japan rtakai@snd.elec.keio.ac.jp S. Uchida uchida@snd.elec.keio.ac.jp A. Sato asato@snd.elec.keio.ac.jp Y. Sanada sanada@elec.keio.ac.jp M. Inamori Department of Electronics and Electrical Engineering, Tokai University, Kitakaname, Hiratsuka, Kanagawa , Japan inamori@tokai-u.jp

2 554 R.Takaietal. 1 Introduction With the popularization of the Internet and the evolution of the digital signal processing technology in recent years, demands of high-speed and large-capacity wireless communications have been recognized. However, implementation of broadband wireless communication services causes shortage of frequency spectrum. Therefore, cognitive radio (CR) has been actively investigated [1 3]. CR is a communication technology with which a wireless terminal recognizes a status of backbone networks and its surrounding radio environment autonomously. One category of the CR is dynamic spectrum access (DSA). DSA realizes effective usage of frequency resources that is temporarily not in use by a primary user [4]. In the DSA, accurate signal detection is required to avoid interference to a primary user. As one of the signal detection schemes, overlapped FFT based energy detection has been proposed [5,6]. Wang et al. [7] evaluates the characteristics of the overlapped FFT energy detection for the prevention of adjacent channel interference. Tomioka et al. [8] analyzes the noise reduction effect with windowing functions in the overlapped FFT based energy detection. The overlapped FFT scheme reduces the variance of squared noise and improves the detection probability as the overlap ratio increases. Though theoretical analysis and computer simulation of the overlapped FFT scheme have been carried out in [8,9], it has not been investigated through experiments. This paper evaluates the effect of the overlapped FFT scheme through experiments. In the experiments, a fixed distortion component that is caused by a direct current (DC) offset is observed. Therefore, the effect of the overlapped FFT scheme under the influence of the fixed distortion component is measured. This paper is organized as follows. Firstly, spectrum sensing with the overlapped FFT based energy detection is explained in Sect. 2. The effect of the fixed distortion component is also presented in Sect. 2. The probability of detection and the probability of false alarm (PFA) using the overlapped FFT based energy detection is evaluated in Sect. 3. Finally, our conclusions of this paper is presented in Sect System Model 2.1 Overlapped FFT Energy detection with the overlapped FFT is presented in Fig. 1. The input signal of the detector, s(t), is given as s(t) = h(t) x(t) + v(t) (1) where h(t) is the impulse response of the channel between an input of a transmit filter and an output of areceive filter, x(t) is the baseband (BB) signal, v(t) is the additive white Gaussian noise (AWGN), and denotes convolution. Subsequently, the input signal, s(t), is sampled as s[n] =s(nt s ) through an analog-to-digital converter (ADC), where n (n = 0, 1,...)is the index of the samples and T s is the sampling interval. The number of the FFT frames, L, is given as L = I N +1 (2) d

3 Experimental Investigation of Signal Sensing 555 Input Signal (Signal + Noise) Time FFT frame Overlapped FFT frame FFT frame FFT FFT FFT Average Power Frequency Fig. 1 Energy detection with overlapped FFT scheme Compare with Threshold where I is the number of the observation samples, N is the FFT frame size, d is the number of the non-overlapped samples between 2 successive FFT frames, and implies the largest integer less than. The number of the FFT frames increases as the overlap ratio (d ol = 1 d/n) grows. The output of the lth FFT frame in the kth bin, S l [k], is given as S l [k] = N 1 n=0 ( s [N(l 1)(1 d ol ) + n] exp j 2πnk ) N where N is the FFT frame size. Finally, the average energy in the kth bin, S[k] 2, is calculated as Ll=1 S[k] 2 S l [k] 2 = (4) L where l is the index of the FFT frames and L is the number of the FFT frames. 2.2 Analysis on Variance of Squared Noise The following theoretical analysis assumes that the input of the detector only contains the noise. The outputs of FFT frames in each bin follow chi-squared distribution [10]. If the number of the FFT frames, L, increases, the outputs for averaging approximately follow correlated Gaussian distribution due to the law of great numbers [10]. The joint probability density function (PDF) is given as (3)

4 556 R.Takaietal. [ 1 p(s 1 [k], S 2 [k],...,s L [k]) = (2π) L/2 exp 1 ] [det R[k]] 1/2 2 S[k]T R[k] 1 S[k] (5) where p(s 1 [k], S 2 [k],...,s L [k]) is the joint PDF of the FFT outputs in the kth bin, S = [S 1 [k] S 2 [k]... S L [k]] T,andR[k] is the correlation matrix of the FFT outputs in the kth bin, which is given as γ 1,1 [k] γ 1,2 [k]... γ 1,L [k] γ 2,1 [k] γ 2,2 [k]... γ 2,L [k] R[k] = (6) γ L,1 [k] γ L,2 [k]... γ L,L [k] where γ i, j is the correlation coefficient between the ith and jth frames. This correlation matrix is a Hermitian matrix and the eigenvalues are real values [11]. The eigenvalues of R[k] are defined as λ[k] = [ λ 1 [k] λ 2 [k]... λ L 1 [k] λ L [k] ] T (7) where λ l is thelth eigenvalue. Finally, the sum of the outputs from the detector is approximated by Gaussian distribution with the following mean and variance, Mean : L λ l [k]σ 2, (8) l=1 Variance : L 2λ l [k] 2 σ 4, (9) l=1 where σ 2 is the variance of the noise, v(t). In computer simulation, the threshold for signal detection is decided with the Gaussian approximation so that P FA satisfies a predetermined value. 2.3 Fixed Distortion Component The power spectrum of the noise measured with the experimental setup described in Sect. 3 isshowninfig.2. The size of FFT is 4,096, the bandwidth of the receive filter is 20 MHz, the sampling interval is 25 ns, the radio frequency (RF) gain and the BB gain of the receiver are 16.0 and 8.0 db, respectively. Figure 2 reveals that there is a large DC offset generated in the direct conversion receiver and the DC offset appears not as a single line of the spectrum. Accordingly, the DC offset interferes to adjacent channels and is observed as a fixed distortion component at the FFT outputs. The input samples composed of the noise and the fixed distortion component in the lth and l + 1th FFT frames are expressed as lth :{(v[d(l 1)]+c[d(l 1)]) (v[d(l 1) + N 1]+c[d(l 1) + N 1])} (10) l + 1th :{(v[dl]+c[dl]) (v[dl + N 1]+c[dl + N 1])} (11) where v[n] is the AWGN and c[n] is the distortion caused by the DC offset in the nth sample, respectively. If d is less than N,(N d) samples of those consecutive two FFT frames are overlapping. The outputs of the lth and (l + 1)th FFT frames in the kth bin are given as

5 Experimental Investigation of Signal Sensing Power (db) Frequency (Hz) Fig. 2 Power spectrum of noise and DC offset (RF gain = 16.0 db, BB gain = 8.0dB) S l [k] = = d(l 1)+N 1 n=d(l 1) ( v[n] exp j 2πnk N ) + C ( Vd(l 1),dl 1 [k]+v dl,d(l 1)+N 1 [k]+c (0 < d < N) V d(l 1),d(l 1)+N 1 [k]+c (d N) (12) and S l+1 [k] = dl+n 1 n=dl ( v[n] exp j ) 2π(n d)k + C N ( Vdl,d(l 1)+N 1 [k] exp( j 2πdk = N )+V d(l 1)+N,dl+N 1[k] exp( j 2πdk N )+C (0<d < N) V d,d+(n 1) [k] exp( j 2πdk N ) + C (d N) (13) where S l [k] and S l+1 [k] are the outputs of the lth and (l + 1)th FFT frames in the kth bin and C is the fixed distortion component. V i, j [k] is the value which is represented as the following equation: V i, j [k] = j n=i ( v[n] exp j 2πnk ). (14) N Finally, the correlation between the lth and (l + 1)th FFT frames in the kth bin, β l,l+1 [k],is calculated as E [ S l [k] S l+1 [k] ] β l,l+1 [k] = E [ S l [k] 2] E [ S l+1 [k] 2] = C 2 + N d N σ 2 n exp ( ) j 2πdk N (0 < d < N) σn 2+ C 2 C 2 (d N) σn 2+ C 2 (15)

6 558 R.Takaietal. Channel (Wired) Transmit Antenna Receive Antenna Signal Generator RF Transceiver Radio Control Board PC Fig. 3 Block diagram of experiment system where σn 2 is the variance of the noise, v[n]. Through the measurement of the correlation between the FFT frames, the magnitude of the fixed distortion component can be determined. Under the existence of the fixed distortion component, the mean and the variance of the squared FFT outputs cannot be calculated in a closed form. Therefore, numerical results are obtained through computer simulation and experiments. 3 Experimental Result 3.1 Experimental Setup Figure 3 illustrates the block diagram of the experimental system. The transmitter and the receiver are connected by a cable in this experiment. As an example of the primary user, the OFDM signal of the IEEE802.11g format is generated by the signal generator [12]. Table 1 shows the specifications of the measurement equipments. The received signal is downconverted and digitized on the Sora radio control board (RCB) [13]. The samples are passed to the memories on the personal computer (PC). The PC picks up the stored samples and carries out FFT as well as energy detection continuously. The experimental conditions are listed in Table 2. The symbols of the transmit signal are modulated with QPSK on each subcarrier. The number of subcarriers and data subcarriers are 64 and 52, respectively. The tap coefficients of the transmit filter are calculated from the Fourier transform of the 11 g spectrum mask and multiplied with a Hamming window. The filter length is 150T s and the bandwidth is 20 MHz. The center frequency of the channel is GHz. As channel models an AWGN channel and an Indoor Residential-A model are assumed [14]. The delay profile of the Indoor Residential-A model is shown in Table 3.The bandwidth of the received signal is 40 MHz. As for the receiver, the observation period for Table 1 Measurement equipment Equipment ESG vector signal generator Agilent E4438C RF transceiver MAXIM MAX2829 A/D converter Analog devices AD9248 Processor Intel Core i7 930 Specifications Frequency: from 250 to 6 GHz maximum output power: +17 dbm Frequency: GHz GHz Gain control range : 93 db Dynamic range: 60 db Speed: 20, 40, and 65 MS/s Resolution: 14 bit Clock: 2.8 GHz Core: quad core

7 Experimental Investigation of Signal Sensing 559 Table 2 Experiment conditions Parameter Modulation Condition QPSK/OFDM Number of subcarriers 64 Number of data subcarriers 52 Transmit filter Fourier transform of the spectrum mask with hamming window Filter length 150T s Filter bandwidth 20 MHz Center frequency of the channel GHz Channel AWGN Indoor Residential-A Bandwidth of received signal 40 MHz Observation period 24 µs Sampling interval 25 ns FFT size 16 PFA 0.1 Detection bin 8 Number of trials 10 4 Table 3 Indoor Residential-A (RMS delay spread = 18 ns) Number of taps Delay (ns) Average power (db) Received Signal Bandwidth:40MHz Transmit Signal Bandwidth:20MHz Center Frequency Frequency MHz Detection bin Fig. 4 Relationship between bandwidth and detection bin signal sensing is set to 24 μs. The sampling interval is 25 ns and the FFT size is 16. The PFA is set to 0.1 [15]. The 8th detection bin is used for performance evaluation while the carrier frequency is located in the middle of the 9th bin as shown in Fig. 4. The number of trials is set to 10 4.

8 560 R.Takaietal. Correlation between the Frames Theory Experiment Delay of Frames (sample) Fig. 5 Correlation between FFT frames with noise and fixed distortion component (RF gain = 16.0dB, BB gain = 8.0dB) 0.14 Convergence Value of Correlation Frequency Bin Fig. 6 Correlation between non-overlapping FFT frames (RF gain = 16.0 db, BB gain = 8.0dB) 3.2 Correlation with Fixed Distortion Component The correlation between the FFT frames in the 8th bin is shown in Fig. 5. The RF gain and the BB gain of the receiver are set to 16.0 and 8.0 db, respectively. From Fig. 5, the correlation at the delay of more than 16 is From Eq. (15), the ratio of the noise variance, σ 2 n,and the power of the fixed distortion component, C 2, is calculated as follows: σ 2 n : C 2 = 85 : 15. (16) The correlations between the non-overlapped FFT frames in each bin is shown in Fig. 6. The RF gain and the BB gain of the receiver are set to 16.0 and 8.0 db, respectively. The magnitude of the fixed distortion component increases in proportion as the FFT bin approaches to the DC offset.

9 Experimental Investigation of Signal Sensing 561 Correlation between the Frames BB Gain=8.0dB, RF Gain=16.0dB BB Gain=32.0dB, RF Gain=16.0dB BB Gain=8.0dB, RF Gain=32.0dB Delay of the Frames (sample) Fig. 7 Correlation between FFT frames in 8th bin with different gain values Table 4 Variance Of squared FFT outputs with noise (8th bin, RF gain = 16.0dB, BB gain = 32.0dB) Overlap ratio Variance of the summed outputs ( 10 7 ) 0/ / / / / / / / / / / / / / / / Variance of the averaged outputs ( 10 5 ) The measured correlations between the FFT frames in the 8th bin with different gain values are presented in Fig. 7. From the correlation between the non-overlapped FFT frames, it is confirmed that the effect of the fixed distortion component is suppressed when the BB gain of the receiver increases to 32 db. In this case the power of the noise grows and the fixed distortion component is negligible. 3.3 Variance of Squared Noise Table 4 shows the measured variance of the squared noise at the FFT outputs when the fixed distortion is negligible. For the measurement, 14 bit ADC is used. The RF gain and the BB

10 562 R.Takaietal. Normalized Variance of Squared Noise Theorey Experiment Overlap Ratio(/16) Fig. 8 Variance of squared FFT outputs with noise (8th bin, RF gain = 16.0 db, BB gain = 32.0dB) Table 5 Variance Of squared FFT outputs with noise and fixed distortion component (8th bin, RF gain = 16.0dB, BB gain = 8.0dB) Overlap ratio Variance of the summed outputs ( 10 3 ) 0/ / / / / / / / / / / / / / / / Variance of the averaged outputs gain of the receiver are set to 16.0 and 32.0 db, respectively. The variance of the summed outputs is given in Eq. (9), and the averaged outputs corresponding to Eq. (4). The normalized variance of the averaged FFT outputs is shown in Fig. 8. The variance is normalized by the value without overlapping. The measurement results indicate that the variance reduces as the overlap ratio increases. The variance of the squared FFT outputs with the noise and the fixed distortion component is shown in Table 5. The RF gain and the BB gain of the receiver are set to 16.0 and 8.0 db, respectively. Figure 9 shows the normalized variance of the averaged outputs. It is shown that the overlapped FFT scheme also reduces the variance effectively under the existence of the fixed distortion component.

11 Experimental Investigation of Signal Sensing 563 Variance of the Squared Noise Simulation Experiment Overlap Ratio(/16) Fig. 9 Variance of squared FFT outputs with noise and fixed distortion component (8th bin, RF gain = 16.0 db, BB gain = 8.0dB) Table 6 Measurement conditions with noise Parameter RF gain BB gain Transmission power Condition 16.0 db 32.0 db 36.0 to 25.0 dbm Probability of Detection Overlap Ratio=0/16 Overlap Ratio=5/16 Overlap Ratio=11/16 Overlap Ratio=15/ SINR (db) Fig. 10 Probability of detection versus SINR with noise (PFA = 0.1, AWGN channel) 3.4 Detection Probability with Noise The measurement results when the Gaussian noise is the major disturbance in energy detection are shown in this section. Table 6 presents the measurement conditions of the gain values and the transmission power of the signal generator. The RF gain and the BB gain of the receiver are 16.0 and 32.0 db, respectively. In order to vary the signal-to-interference and noise power ratio (SINR) from 10 to 0 db, the power of the transmit signal is set from 36.0 to 25.0 dbm.

12 564 R.Takaietal. Probability of Detection Overlap ratio=0/16 Overlap ratio=5/16 Overlap ratio=11/16 Overlap ratio=15/ PFA Fig. 11 ROC with noise (SINR = 6.0 db, AWGN channel) Probability of Detection Overlap Ratio=0/16 Overlap Ratio=5/16 Overlap Ratio=11/16 Overlap Ratio=15/ SINR (db) Fig. 12 Probability of detection versus SINR with noise (PFA = 0.1, Indoor Residential-A model) Figure 10 shows the probability of detection versus the SINR on the AWGN channel for the PFA of 0.1. The detection probability improves with the increase of the overlap ratio over the whole range of the SINR. Especially, the overlapped FFT scheme improves the probability of detection up to about 0.09 when the SINR is 6.8dB. Figure 11 presents the receiver operating characteristic (ROC) curves for the SINR of 6.0 db. The ROC also improves by the overlapped FFT scheme. Figure 12 indicates the probability of detection versus the SINR on the Indoor Residential- A model for the PFA of 0.1. As the overlap ratio increases, the detection probability improves. In particular, the detection probability increases about by the overlapped FFT scheme at the SINR of 3.7 db. Figure 13 shows the ROC curves for SINR = 1.2dB.TheROC also shows better performance for the overlapped FFT scheme.

13 Experimental Investigation of Signal Sensing Probability of Detection Overlap ratio=0/16 5 Overlap ratio=5/16 Overlap ratio=11/16 Overlap ratio=15/ PFA Fig. 13 ROC with noise (SINR = 1.2 db, Indoor Residential-A model) Table 7 Measurement conditions with noise and fixed distortion component Parameter RF gain BB gain Transmission power Condition 16.0 db 8.0 db 30.0 to 19.0 dbm Probability of Detection Overlap Ratio=0/16 Overlap Ratio=5/16 Overlap Ratio=11/16 Overlap Ratio=15/ SINR (db) Fig. 14 Probability of detection versus SINR with noise and fixed distortion component (PFA = 0.1, AWGN channel) 3.5 Detection Probability with Noise and Fixed Distortion Component The measurement results under the existence of the fixed distortion component are shown in this section. Table 7 presents the measurement conditions. The RF gain and the BB gain of the receiver are 16.0 and 8.0 db, respectively. In order to vary the SINR from 10 to 0 db, the power of the transmit signal is set from 30.0 to 19.0 dbm.

14 566 R.Takaietal. Probability of Detection Overlap ratio=0/16 Overlap ratio=5/16 Overlap ratio=11/16 Overlap ratio=15/ PFA Fig. 15 ROC with noise and fixed distortion component (SINR = 6.7 db, AWGN channel) 0 Probability of Detection Overlap ratio=0/16 5 Overlap ratio=5/16 Overlap ratio=11/16 Overlap ratio=15/ Fig. 16 ROC with noise and fixed distortion component (SINR = 2.0 db, Indoor Residential-A model) PFA Figure 14 shows the probability of detection versus the SINR on the AWGN channel for the PFA of 0.1. It is shown that the overlapped FFT scheme effectively improves the probability of detection by about 0.15 when the SINR is 7.4 db. Figure 15 shows the ROC curves for the SINR of 6.7 db. The ROC also improves by the overlapped FFT scheme. The ROC curves on the Indoor Residential-A channel for SINR = 2.0 db is shown in Fig. 16. The overlapped FFT scheme increases the probability of detection on the Indoor Residential-A channel. 4 Conclusions This paper evaluates the performance of the overlapped FFT based energy detection scheme through the experiments. In the measurements, the fixed distortion component caused by the

15 Experimental Investigation of Signal Sensing 567 DC offset has been observed. The effect of the fixed distortion component is also investigated in this paper. The variance of the squared FFT outputs with the noise reduces as the overlapped ratio between the FFT frames increases. The numerical results measured through the experiments have shown that the detection probability improves by about 0.15 when the PFA is 0.1 and the SINR is 7.4 db with the fixed distortion component. When the fixed distortion is negligible, the probability of detection improves by about 0.09 with the PFA of 0.1 at the SINR of 6.8 db. Acknowledgments This work is supported in part by the Keio Gijuku Fukuzawa Memorial Fund for the Advancement of Education and Research from Keio University in Japan. References 1. Mitra III, J., & Maguire, G. Q, Jr. (1999). Cognitive radio: Making software radios more personal. IEEE Personal Communications, 6, Mitra III, J. (1999). Cognitive Radio for Flexible Mobile Multimedia Communications. In 1999 IEEE International Workshop on Mobile Multimedia Communications (pp. 3 10). 3. Haykin, S. (2005). Cognitive radio: Brain-empowered wireless communications. IEEE Journal on Selected Areas in Communications, 23(2), Zhao, Q., & Sadler, B. M. (2007). A survey of dynamic spectrum access. IEEE Signal Processing Magazine, 24(3), Maranda, B. H. (1996). On the false alarm probability for an overlapped FFT processor. IEEE Transactions on Aerospace and Electronic System, 32(4), Fan, H., Wang, S., Chen, F., & Zhu, Y. (2006). Frame-overlapped zoom-fft optimization in PD radar application. In International conference on radar (pp. 1 5). 7. Wang, S., Inkol, R., Rajan, S., & Patenaude, F. (2010). Detection of narrow-band signals through the FFT and polyphase FFT filter banks: Noncoherent versus coherent integration. IEEE Transactions on Instrumentation and Measurement, 59(5), Tomioka, T., Tomizawa, T., & Kobayashi, T. (2009). High-sensitivity carrier sensing using overlapped FFT for cognitive radio transceivers. In IEEE 69th Vehicular Technology Conference (pp. 1 5). 9. Uchida, S., Sato, A., Inamori, M., & Sanada, Y. (2012). Overlapped FFT scheme with frames of noncontinuous samples. In IEEE International Conference on Communication Systems (pp ). 10. Proakis, J. (1989). Digital communications. New York: McGraw-Hill. 11. Golub, G. H., & Loan, C. F. V. (1996). Matrix computations (3rd ed.). Baltimore, MD: The Johns Hopkins University Press. 12. Chong, J. W., Sung, Y., & Sung, D. K. (2008). Cross-layer performance analysis for CSMA/CA system: Impact of imperfect sensing. In IEEE 9th Workshop on Signal Processing Advances in Wireless Communications. 13. Tan, K., Zhang, J., Fang, J., Liu, H., Ye, Y., Wang, S., et al. (2009). Sora: High performance software radio using general purpose multi-core processors. In 6th USENIX Symposium on Networked Systems Design and Implementation. 14. Joint Technical Committee of Committee T1 R1P1. 4 and TIA TR /TR on Wireless Access. (1994). Draft final report on RF channel characterization. Paper no. JTC(AIR)/ R IEEE Part 22: Cognitive wireless RAN medium access control (MAC) and physical layer (PHY) specifications: Policies and procedures for operation in the TV bands.

16 568 R.Takaietal. Author Biographies Ryo Takai was born in Shizuoka, Japan in He received his B.E. degree in electronics engineering from Keio University, Japan in Since April 2013, he has been a graduate student in School of Integrated Design Engineering, Graduate School of Science and Technology, Keio University. His research interests are mainly concentrated on software defined radio. Shoya Uchida was born in Yamanashi, Japan in He received his B.E. degree in electronics engineering from Keio University, Japan in Since April 2012, he has been a graduate student in School of Integrated Design Engineering, Graduate School of Science and Technology, Keio University. His research interests are mainly concentrated on software defined radio. Akihiro Sato was born in Kagawa, Japan in He received his B.E. degree in electronics engineering from Keio University, Japan in Since April 2011, he has been a graduate student in School of Integrated Design Engineering, Graduate School of Science and Technology, Keio University. His research interests are mainly concentrated on software defined radio.

17 Experimental Investigation of Signal Sensing 569 Mamiko Inamori was born in Kagoshima, Japan in She received her B.E., M.E., and Ph.D degrees in electronics engineering from Keio University, Japan in 2005, 2007, and 2009, respectively. Since April 2013, she has been a lecturer in Tokai University. She received the Young Scientist Award from Ericsson Japan in Her research interests are mainly concentrated on wireless communication and power electronics. Yukitoshi Sanada was born in Tokyo in He received his B.E. degree in electrical engineering from Keio University, Yokohama Japan, his M.A.Sc. degree in electrical engineering from the University of Victoria, B.C., Canada, and his Ph.D. degree in electrical engineering from Keio University, Yokohama Japan, in 1992, 1995, and 1997, respectively. In 1997 he joined the Faculty of Engineering, Tokyo Institute of Technology as a Research Associate. In 2000 he jointed Advanced Telecommunication Laboratory, Sony Computer Science Laboratories, Inc, as an associate researcher. In 2001 he jointed Faculty of Science and Engineering, Keio University, where he is now a professor. He received the Young Engineer Award from IEICE Japan in His current research interest is in software defined radio, cognitive radio, and OFDM systems.