Narrow Band Interference (NBI) Mitigation Technique for TH-PPM UWB Systems in IEEE 82.15.3a Channel Using Wavelet Pacet Transform Brijesh Kumbhani, K. Sanara Sastry, T. Sujit Reddy and Rahesh Singh Kshetrimayum Department of Electronics and Electrical Engineering Indian Institute of Technology Guwahati Guwahati, India E-mail:.brijesh@iitg.ac.in, rs@iitg.ac.in Abstract Impulse Radio Ultra Wideband (IR-UWB) systems have been studied for their inherent advantages of coexistence with narrowband systems with high data rate over short distances with sufficiently small amount of transmitted power. UWB systems are highly susceptible to interference from coexisting narrowband systems because of very low power transmission. In this paper, we have studied the effect of narrowband interference (NBI) on the bit error rate (BER) performance of time hopping pulse position modulated ultra wide band (TH-PPM UWB) systems in IEEE 82.15.3a channel model. The method of wavelet pacet is employed in pulse shaping to reduce the effect of NBI on the bit error rate performance. The NBI taen in consideration is the interference from IEEE 82.11a wireless local area networ (WLAN) at 5.2GHz band. The simulation results show that wavelet pacet based pulse shaping is an effective approach to reduce the effect of narrowband systems in coexistence with UWB systems. Keywords- IR-UWB, IEEE 82.11a, Wavelet pacets, Narrow Band Interference(NBI) I. INTRODUCTION Ultra Wide band (UWB) technology has been accepted to be the solution for high speed transmission over short distances. Time hopping pulse position modulated UWB (TH- PPM UWB) is one of the popular multiple access techniques which uses pulse position modulation (PPM) scheme with time hopping used for accommodating multiple users. Impulse radio (IR) UWB is the carrierless communication system in which narrow pulses are used for transmission. The pulses are designed to agree with the transmission limits of Federal Communications Commission (FCC) [1]. The transmitted power is limited to -41.3dBm/MHz in the band of 3.1GHz to 1.6GHz and a -1dB lower outside that band. This low power spectral density maes UWB signals highly prone to interference from the narrowband systems operating in the frequency range coinciding with the band allocated to UWB. In [2] and [3], the effect of interference from IEEE 82.11a wireless local area networ (WLAN) which is a licensed wireless technology that operates at 5.2GHz band is studied and the degradation in bit error rate (BER) performance due to narrowband interference (NBI) is reported. The methods of NBI suppression based on modulation parameter, notch filtering and minimum mean square error (MMSE) combining have been reported in [2]. The NBI mitigation technique based on template waveform processing using Gabor Transform is proposed in [4] which used Gabor Transform to approximate the template waveform or the pulse shape to ignore the NBI frequency. Other interference suppression schemes for UWB applications have been proposed using multiple antennas and the selection diversity techniques [5], [6]. In this paper, we used template waveform processing by wavelet pacet approximation [7] to approximate the transmitted pulse shape. The TH-PPM UWB system is analyzed in Modified Saleh-Valenzuela multipath channel model (IEEE 82.15.3a) [8], [9] and single user scenario is considered. RAKE reception scheme is used to receive and combine multipath components of the received signal. We have also analyzed the effect of RAKE receiver combining all the multipath and combining first L multipath components. II. UWB SYSTEM MODEL Fig. 1 shows UWB system model incorporating wavelet pacet approximation for NBI cancellation. At transmitter, the originally generated Gaussian doublet pulse is approximated using wavelet pacet transform to remove the frequency components corresponding to NBI. The data is then modulated by TH-PPM modulator and the approximated pulse is used for transmission. A correlation type receiver is considered at receiver that uses template to correlate with the received signal. The template is also approximated using wavelet pacet transform to cancel the effect of NBI signals present in the system. Output of correlator is integrated the data is regenerated by detector which thresholds the sampled output of integrator. Fig. 2 shows the signal waveforms obtained at different stages of the UWB system model described in Fig. 1 when data bit is transmitted. The title of each subfigure signifies the point at which the waveform is obtained. For illustration purpose, the time hopping (TH) code assumed is (4,,2) which gives sufficient separation of pulses in the transmitted signal. The values of interference power to signal power ratio (ISR) and SNR considered are 2dB. A. UWB Transmitter We have considered TH-PPM UWB transmission in which multiple pulses are transmitted at random instant specified
Pulse Generator Wavelet Pacet Approximation Tx Rx Data TH-PPM Modulator Integrator Detector Template Generator Wavelet Pacet Approximation Regenerated Data Fig. 1: UWB System Model by the TH code. TH code serves the purpose of identifying the user and adding security. The transmitted TH-PPM UWB signal can be given as s(t)= i (i+1)n s ( ) ETX p t nt f C n (i) T c δd i in s T f n=in s where i is bit index, n is pulse index, N s is length of pseudo code used in TH-PPM in repetitive coder, p(t) is energy normalized pulse, T c is chip time, T f is frame time, δ is the PPM shift and d i is ith data bit and E TX is transmitted energy per pulse. In this case, our focus being on interference mitigation, we have considered single user scenario despite the fact that TH- PPM UWB systems are ideal for multiple transmitting user scenario. B. IEEE 82.15.3a Channel Model IEEE 82.15.3a discrete channel impulse response is given by [9] h(t)=x L 1 l= K 1 = (1) α,l δ(t T l τ,l ) (2) where l is the cluster index, is the index of a ray within the cluster, X is the log-normal shadowing parameter, α,l is the multipath gain of the th ray of the l th cluster. The total number of clusters and rays are denoted by L and K respectively. The arrival time of the l th cluster is denoted by T l, and that of the th ray within the l th cluster is represented by τ,l. The arrival time of clusters and rays are modeled as Poisson processes. C. RAKE Reception The transmitted signal as in (1) passes through IEEE 82.15.3a channel equation (2). In addition, additive white Gaussian noise (AWGN) and NBI gets added to it. Thus the signal at receiver may be given as r (t)=x L 1 K 1 (i+1)n s ETX α,l ( l= = i n=in s ) p t nt f C n (i) T c δd i T l τ,l in s T f +n(t)+i(t) (3) where i(t) is NBI and n(t) is AWGN. Received signal is collection of LK multipath components. RAKE receiver is used to combine all or some of the multipath components to improve the performance. In all RAKE (ARae), all the multipath components are combined and in partial RAKE (PRae), first P paths are combined. D. Narrow Band Interference NBI in consideration for this wor is IEEE 82.11a WLAN interference around 5.2GHz band. It uses OFDM in which multiple bits are transmitted simultaneously using orthogonal carriers. In this paper, we have used single tone sinusoidal interferer at frequency 5.2GHz. III. WAVELET PACKET TRANSFORM AND TEMPLATE WAVEFORM PROCESSING A. Wavelet Pacet Transform Wavelet pacet transform represent any signal as pacet of wavelets. Wavelet pacet transform does decomposition of both the low frequency components and the high frequency components. Wavelet pacet representation of a signal m(t) can be given as [7] m i (t) ( c i, ψ i, (t)=c i, ψ i, 2 i t ) (4) where m i (t) is the wavelet pacet approximation of the signal at the i th decomposition level, decomposition level i is chosen based on the desired accuracy level of approximation. ψ i, (t) is chosen wavelet and c i, is th coefficient of wavelet pacet transform of m(t) calculated as [7] ( c i, = s )ψ ( 2 i i, 2 i t ) (5) Our aim is to generate the template or pulse as per FCC spectrum mas. The template thus generated should be able to reject the frequencies corresponding to NBI at the correlator of the receiver end. The components corresponding to 5.2 GHz band are identified and the coefficients corresponding to them are made
.1 Approximated Pulse.1.2.4.6.8 1 1 2 x 1 Signal Passed Through Channel 2 5 1 15 2 25 3 5 x 1 Transmitted Signal (Bit ) 5 1 15 2 25 3 5 x 1 Effect of Channel + NBI 5 1 15 2 25 3 2.5 2.5 5 x 1 Effect of Channel + NBI +AWGN 5 1 15 2 25 3.1 Template Multiplied with Received Signal.1 5 1 15 2 25 3 5 25 25 Approximated Template at Receiver 5 1 15 2 25 3.1.5 Correlation Sample at Receiver 5 1 15 2 25 3 Fig. 2: Signal waveforms at different stages of the UWB system for time hopping (TH) code (4,,2) Amplitude 12 1 8 6 4 2 2 4 Reconstruction using Wavelet Pacet Transform Original pulse reconstructed pulse 1 2 3 4 5 6 7 8 TIME x 1 1 Fig. 3: Original and reconstructed pulse after removing the interfering components zero to avoid the contribution of those components at output of correlators at the receiver. For approximation of pulse we Amplitude 25 2 15 1 5 Reconstruction using Wavelet Pacet 2 4 6 8 Frequency Original Reconstructed x 1 9 Fig. 4: Frequency spectra of original and reconstructed pulses have used 5 th level wavelet pacet decomposition and Remez exchange algorithm [1]. The original and approximated pulses are shown in Fig.3 and their corresponding frequency
1 1 1 Comparison of NBI mitigation techniques No mitigation Mitigation (Remez) Mitigation (Coiflet) 1 1 1 Effect of ARae and PRae on BER No Mitigation Mitigation ARae Mitigation PRae(L=1) 1 2 1 3 1 2 1 3 1 4 1 4 1 3 6 9 12 15 18 21 2425 SNR (db) 1 3 6 9 12 15 18 21 2425 SNR (db) Fig. 5: Comparison of BER performance for different mitigation algorithms for ISR=1dB Fig. 7: SNR vs. BER to show effect of RAKE selection on performance 1 1 1 1 2 1 3 1 4 1 ISR Vs BER for different values of SNR SNR=dB SNR=5dB SNR=1dB SNR=12dB SNR=15dB 2 4 6 8 1 12 14 16 18 2 ISR (db) Fig. 6: BER vs. ISR for different values of SNR spectra are depicted in Fig. 4. A slight degradation in the pulse shape is observed and a deep at 5.2GHz in spectrum indicates that there is no power transmitted in that band. B. Template Waveform The pulse we have assumed originally is second derivative of Gaussian pulse being widely used in analysis of UWB systems [2]. It can be given as [ ( ) ] t 2 p(t)= 1 4π e 2π( τm t )2 (6) τ m where τ m is pulse shaping factor. It is approximated by wavelet pacet transform. The approximated pulse can be represented in the form of summation of wavelets as ( p i (t) c i,ψ i, 2 i t ) (7) / B n where B n indicate the set of wavelets that have been removed from the wavelet pacet to mitigate NBI. This approximated pulse is then used at the receiver as local signal for correlation which can be given as v(t)= L 1 l = K 1 = N s 1 n = α,l {p i (t τ) p i (t τ δ)} (8) where τ = n T f +Cn m T c + mnt f + T l + τ,l is the arrival delay of n th pulse in th path of l th cluster and v(t) is the modified template. IV. SIMULATION RESULTS The UWB TH-PPM system was simulated for different cases. Improvement in performance is measured in terms of average bit error rate (BER). Fig. 5 compares the effect of NBI mitigation using wavelet pacet transform with Coiflet wavelet transform and Remez algorithm. The different values of error floor without NBI mitigation and with mitigation are observed. The simulation is done for interference power to signal power ratio of 1 db. Simulations are also performed for the study of effect of interference on BER performance with NBI mitigation technique using Remez algorithm. The UWB system was simulated for different values of ISR for constant SNR. Fig. 6 shows the BER performance over ISR for different values of SNR. The BER performance has also been studied to observe the effect of ARae over PRae for ISR of 1dB. Fig. 7 shows the BER vs. SNR plot for ARae and the RAKE receiver combining only first 1 multipath components. It is observed that ARae gives good performance at cost of hardware complexity.
V. CONCLUSION NBI from narrowband systems coexisting with UWB spectrum affects the UWB performance severely. The effect of NBI on BER performance of TH-PPM UWB system is studied and a technique to improve the performance in presence of NBI is studied and simulated. The simulation results show that the NBI mitigation technique of template waveform approximation with Remez algorithm gives almost 1 db SNR gain and that with Coiflet wavelet gives SNR gain of 5dB at BER of 1 2 (Fig. 5). The effectiveness of NBI mitigation increases for higher SNR conditions (Fig. 6). Wavelet pacet approximation with Remez algorithm is more effective in NBI mitigation. REFERENCES [1] FCC, First Report and Order, FCC-2-48, Revision of Part 15 of the Commissions rules Regarding Ultra-Wideband Transmission Systems, Federal Communications Commission, vol. 11, pp. 991 11, September 1993. [2] X. Chu and R. Murch, The effect of NBI on UWB time-hopping systems, IEEE Transactions on Wireless Communications, vol. 3, no. 5, pp. 1431 1436, 24. [3] B. Hu and N. Beaulieu, Effects of IEEE 82.11a narrowband interference on a UWB communication system, in Proceedings IEEE International Conference on Communications, vol. 4, 25, pp. 2818 2824 Vol. 4. [4] K. Ohno and T. Iegami, Interference mitigation study for UWB radio using template waveform processing, IEEE Transactions on Microwave Theory and Techniques, vol. 54, no. 4, pp. 1782 1792, April 26. [5] V. Bharadwaj and R. Buehrer, An interference suppression scheme for UWB signals using multiple receive antennas, IEEE Communications Letters, vol. 9, no. 6, pp. 529 531, 25. [6] J. Ibrahim and R. Buehrer, NBI mitigation for UWB systems using multiple antenna selection diversity, IEEE Transactions on Vehicular Technology, vol. 56, no. 4, pp. 2363 2374, 27. [7] M. K. Lashmanan and H. Niooar, Wavelet pacet based strategy to mitigate wideband interference on impulse radio, in Proceedings IEEE 18th International Symposium on Personal, Indoor and Mobile Radio Communications, 27, pp. 1 5. [8] A. Saleh and R. Valenzuela, A statistical model for indoor multipath propagation, IEEE Journal on Selected Areas in Communications, vol. 5, no. 2, pp. 128 137, February 1987. [9] J. R. Foerster, Channel modeling Sub-committee Report Final, IEEE P82.15-2/49r1-SG3a, February 23. [1] O. Rioul and P. Duhamel, A remez exchange algorithm for orthonormal wavelets, IEEE Transactions on Circuits and Systems II: Analog and Digital Signal Processing, vol. 41, no. 8, pp. 55 56, 1994.