Ultra Wideband System Performance Studies in AWGN Channel with Intentional Interference

Transcription

1 Ultra Wideband System Performance Studies in AWGN Channel with Intentional Interference Matti Hämäläinen, Raffaello Tesi, Veikko Hovinen, Niina Laine, Jari Iinatti Centre for Wireless Communications, University of Oulu P.O.Box 45, FIN-94 University of Oulu, FINLAND Abstract: This aer studies the erformance degradation of ultra wideband (UWB) radio system in intentionally interfered. Simulated UWB systems utilize ulse shae (PSM), ulse osition (PPM) and ulse amlitude (PAM) data modulation schemes. All the modulation schemes are binary. The ulse waveforms used in the study are based on a Gaussian ulse, being its first four derivatives. Interference model is coloured Gaussian noise, having 6 MHz bandwidth at both UMTS ulink and downlink bands. The results showed that PAM modulation outerforms the other studied modulation schemes. If PSM is used, the selection of the used ulses set also the sensitivity against interference. PPM modulated system retains its erformance while the interfering ower is increased. However, its overall erformance is the worst among the imlemented modulations.. Introduction According to the regulations, radio system is classified as an ultra wideband (UWB) if its bandwidth is greater than 5 MHz, or its fractional bandwidth is greater than 2% []. In this study, the ulse waveforms used are satisfying these requirements. The most common carrierless UWB data modulation schemes used in the roosed communication alications nowadays are ulse osition (PPM) and ulse amlitude (PAM) modulations. Including these schemes, this aer studies also a erformance of an UWB system based on ulse shae modulation (PSM) that also allows one to get an orthogonal signalling. Multilevel (M>2) modulation schemes are excluded from this study. The system comarison has been made by biterror-rate (BER) simulations. The goal of the study is to rank the erformance of different UWB system concets in interfered environment. Interference in our case is located both at the UMTS-FDD ulink and downlink bands at the same time. Reference results where only one of the listed bands is interfering can be found from [2]. 2. UWB System Model The studied hysical layers are based on time hoing (TH) and direct sequence (DS) techniques alied to UWB context. In [2], a more detailed descrition of the studied system concets can be found. The TH-UWB model used in this study corresonds to the conventional imulse radio [3] where the discontinuous transmission instants are controlled by a seudo random time hoing code. Because the ulse reetition interval in TH system is much longer than the ulse widths, received noise ower level can be reduced by keeing the receiver off during the eriods when the user of interest is in silent mode. In multi-user case, all the users have different time hoing attern to avoid collisions. The studied ulse waveforms are the first four derivatives of the Gaussian ulse reresented in Figure. Based on [4] the antennas are modelled as a differentiation oeration. During the simulations, all the timing and synchronization rocedures are assumed to be ideal. DS-UWB model corresonds to the conventional DS sread sectrum (SS) technique having chi waveforms from Figure. In DS-UWB, however, high chi rate is not used for the sectral sreading, but a DS code is used for user searation. The following results are resented for a single user scenario. The effects of multile access interference in the corresonding UWB systems are studied in [5]. To decrease the simulation times the average ower and the total rocessing gain (PG) of the studied UWB link are fixed to P UWB = dbm and PG = 2 db, resectively. Because the relative rank order is studied the PG limitation does not imair the generality of the results. A. Pulse Position Modulation In a binary PPM modulation scheme the information is conveyed with the time shifts between the nominal and actual ulse transmission instants. If the ulse is sent during the actual transmission time that is defined by the user secific seudo

2 random code, bit is. If the transmission instant is delayed of a certain time instant that is related to the modulation index of the system δ, the transmitted bit is. In an the erformance of PPM modulated signal is related to the used δ due to the cross-correlation roerties of the ulse waveforms [6]. In this study, the otimal δ that gives a better erformance than the theoretical robability of error of orthogonal signals. As a matter of fact, if δ <, the cross-correlation of the two ulses related to bit and can give negative values. Otimal δ is related to the radiated ulse waveform, and it corresonds to the time shift of the ulses that gives the minimum value of correlation. Values for the used waveforms are resented in Table [6]. PPM modulation is imlemented only to TH-UWB concet. Table. Otimal PPM modulation indexes δ for different ulse waveforms. Waveform Otimal δ st ulse * T 2 nd ulse * T 3 rd ulse.2952 * T 4 th ulse.9522 * T B. Pulse Shae Modulation The radiated ulse waveform airs used in the PSM case have been resented in Figure. The ulse widths associated to the both data bits are equal. In PSM, the selection of the waveform air affects the sectral allocation of the transmitted signal, and due to that, the tolerance against the interference. PSM is not an original invention even in UWB context. Further information can be found in [7]. As one can notice from the cross-correlation results in Figure 2, some of the ulse airs form an orthogonal set, and almost antiodal sets can be found as well. In addition to the coherent methods, an orthogonal ulse air that has crosscorrelation zero can be used also in non-coherent systems. BER curves will follow the shae of the cross-correlation function if the signal-to-noise ratio (SNR) is fixed. Using orthogonal ulses in, the erformance of PSM modulated UWB system equals the erformance of PPM. In the interference case, the erformance is related to the sectral roerties of the different ulses used in the UWB system. In this aer, both TH- and DS-UWB concets are imlemented using PSM data modulation. C. Pulse Amlitude Modulation In a binary PAM data modulation scheme, different ulse waveforms, or their amlitude reversed versions are sent, giving an antiodal ulse set if the timing is ideal. This modulation scheme can be used in coherent UWB systems. PAM is studied both in TH- and DS-UWB concets. 3. Interference Model Interference in the study is modelled using coloured Gaussian noise (CGN) that is a bandlimited version of the Gaussian white noise (see Figure 3). The interference is simultaneously located to the UMTS FDD bands, both in ulink (UL, f c =.95 GHz) and downlink (DL, f c = 2.4 GHz) bands at the same time. Interfering signal having 6 MHz bandwidth at both bands is reresenting a fully loaded UMTS system. Unless otherwise mentioned, the total interference ower in fixed to P I = + dbm, that is equally distributed to the UL and DL bands, keeing the signal-to-interference ratio (SIR) in the simulation as SIR = - db. 4. Simulation results This chater reresents the simulation results based on the assumtions introduced in the revious chater. UWB system erformances (BER) are studied as a function of SNR or SIR. The rocessing gain of the studied UWB system is fixed to PG UWB = 2 db to fasten the simulations. Keeing T =.5 ns and PG tot = 2 db, the data rate in the studied cases is R d = 2 Mbs. In time hoing case, the rocessing gain is evenly divided between the gain coming from reetition coding and the gain coming from the low duty cycle. Due to the similar simulation assumtions the resented results are comarable to each others. The simulation results for the UWB system utilizing PSM modulation are resented in Figure 4. If one comares the BER results to the crosscorrelation values from Figure 2, a straightforward connection can be seen. The ulse airs based on the Gaussian st and 3 rd ulses and 2 nd and 4 th ulses (n th derivatives of the Gaussian ulse) are forming almost an antiodal ulse air, and the corresonding erformances give relatively the best results. With the used SIR assumtion the theoretical curves are almost reached with those ulse airs. If the ulses are selected otherwise, an orthogonal erformance can be reached. Figure 4 shows that indeendently of the used ulse airs, significant difference between TH and DS concet cannot be seen.

3 Figure 5 gives a summary on the erformance of PAM, PPM and PSM modulated systems as a function of increasing interference ower. In PPM case, the difference in the erformance based on the orthogonal (δ = ) and the otimal (δ = δ ot) modulation index values can be seen. In otimal case, δ ot is taken from Table. The UWB erformance starts to degrade when the UMTS interference signals are about 2 db higher than UWB signal level. 2 db is also the value of the rocessing gain. Figure 6 reresents the simulated erformance curves for the ulse shae modulated (PSM) ultra wideband systems using different ulse widths. In different subfigures the ulse airs associated to the bits and bit are changed. The best simulated erformance can be reached when the ulse air is based on the Gaussian 2 nd and 4 th ulses. The airs using Gaussian st and 2 nd, st and 4 th and 3 rd and 4 th ulses do not differ significantly to each others. The reasons for the difference in erformance are related to the sectral roerties of the ulses and interference. If the interfering signals and the used UWB ulses are overlaing in the frequency domain the degradation in the system erformances is highest. Figures 7a-b resent the erformance of PPM and PAM modulated UWB systems in when the UMTS ulink and downlink bands are interfering. In PPM case, ulse widths T =.5 ns and T =. ns are studied. As a reference, also the results for the case when only UMTS UL band is interfering with the same total ower level have been resented. It can be noticed that PPM systems using ulses related to higher orders of derivatives of the Gaussian ulse outerform the basic ones ( st, 2 nd ulses). Comaring now the results between the two ulse widths, one could notice that the system erformance is worse with the longer ulse. The reason is that in this case the sectrum of the desired signal moves towards the interfering band, that is, to lower frequencies. With a certain T the UWB sectrum will eventually overla the sectrum of the interfering signal. Due to the sectral allocation the UWB system can tolerate even high dual band interference because the sectra are not overlaing although the UWB bandwidth is in a class of GHz. Figure 7b reresents the comarison between the ulse waveforms both in DS and TH cases for different waveforms when modulation scheme is PAM. Based on the results obeying the simulation assumtions used, the imact of the interference on UWB system is insignificant. However, also here the higher order ulses give better erformance if comared to the lower order ulses. 5. Conclusion This aer studies single user UWB system erformance in when interference is coming from both the UMTS ulink and downlink bands. Time hoing and direct sequence based UWB systems having ulse waveforms based on first four derivatives of Gaussian ulse are used in the study. The modulation schemes imlemented are ulse osition, ulse amlitude and ulse shae modulations. All studied modulation schemes are binary. The simulation results indicate that the dual band interference degrades the PSM system worst. However, if the ulse air is based on the 2 nd and 4 th derivatives of the Gaussian ulse, the effect of the interference is only db in a BER level of -3 if comared to the erformance in ure. In the case of the other ulse airs the erformance degradation is 3-5 db. PAM outerforms the other studied modulation schemes and there cannot be seen significant difference between TH and DS concets. PAM outerforms also the other modulation schemes if the sensitivity against the interference ower is taken into account. The erformance starts to degrade after the interference is 24 db higher than the UWB signal. The most sensitive modulation scheme is PSM. However, PPM modulation retains the same erformance level while the interference increases, although its overall erformance is the worst. 6. Acknowledgements This study has been funded by the National Technology Agency of Finland (Tekes), Nokia, Elektrobit and the Finnish Defence Forces, and also in the framework of the roject UL- TRAWAVES: IST , which is artly funded by the Euroean Community. The authors would like to thank the sonsors for their suort. 7. References [] Federal Communications Commission: First Reort and Order. Ar 22, 22. FCC [2] M. Hämäläinen, V. Hovinen, R. Tesi, J. Iinatti, M. Latva-aho, On the UWB System Coexistence with GSM9, UMTS/WCDMA, and GPS Systems, IEEE Journal on Selected Areas in Communications, Dec.22, Vol. 2, No. 9, [3] R.A. Scholtz, M.Z. Win, Imulse Radio, Wireless Communications, TDMA versus CDMA (Ed.

4 S. Glisic, P. Leänen), Kluwer Academic Publisher, London 997, [4] H.F.Harmuth, Antennas and Waveguides for Nonsinusoidal Waves. Academic Press, Inc., 984 [5] R.Tesi, M.Hämäläinen, I.Oermann, V.Hovinen: On the Multi-User Interference Study for Ultra Wideband Communication Systems in AWGN Channel. International Worksho on Ultra Wideband Systems, IWUWBS, Oulu, Finland [6] M.Hämäläinen, R.Tesi, J.Iinatti, V.Hovinen: On the Performance Comarison of Different UWB Data Modulation Schemes in AWGN Channel in the Presence of Jamming, 2 IEEE Radio and Wireless Conference, RAWCON22, Boston, USA [7] J.A.N. da Silva, M.L.R. de Camos, Orthogonal Pulse Shae Modulation for Imulse Radio, International Telecommunications Symosium, ITS 22, Natal, Brazil Figure 3. Sectrum of the coloured Gaussian noise located at UMTS DL band PSM ds&th,umts: UL/DL,P I = dbm, f c =.95GHz, f c2 = 2.4GHz.8 =6 MHz Amlitude.2.2 Amlitude.4.5 Amlitude.6.8 Gaussian ulse the 2 nd derivative of Gaussian ulse.5.5 Time [ns] the st derivative of Gaussian ulse the 3 rd derivative of Gaussian ulse.5.5 Time [ns] Figure. Pulse waveforms and waveform airs for PSM modulation used in this study..5.5 Amlitude =.5ns, st & 3 rd PSM ds,t =.5ns, st & 3 rd =.5ns, 2 nd & 3 rd PSM ds,t =.5ns, 2 nd & 3 rd PSM th,t =.5ns, st & 4 th PSM ds,t =.5ns, st & 4 th =.5ns, 2 nd & 4 th PSM ds,t =.5ns, 2 nd & 4 th Theor. BPSK Theor. BPPM Figure 4. PSM modulated UWB system erformances, T =.5 ns Correlation factor.2.2 Pulses and 2 Pulses and 3 Pulses and 4 Pulses 2 and 3 Pulses 2 and 4 Pulses 3 and 4 Interference at UMTS UL&DL bands TH BPAM, 4 th ulse DS BPAM, 4 th ulse TH PPM, 4 th ulse, waveform otimised δ TH PPM, 4 th ulses, orthogonal ulses, δ= TH PSM, 2 nd and 4 th ulses DS PSM, 2 nd and 4 th ulses Time shift between two ulses [ns] Figure 2. Cross-correlations between different ulse waveforms in PSM. SNR = 8 db, single user case,t =.5 ns Total interfence ower [db] Figure 5. Bit-error-rates as a function of interference ower, SNR = 8 db.

5 PSM ds&th,umts: UL/DL, st & 2 nd, P I,tot = dbm, f c =.95GHz, f c2 = 2.4GHz PSM ds&th,umts: UL/DL, st & 4 th, P I,tot = dbm, f c =.95GHz, f c2 = 2.4GHz = 6 MHz = 6 MHz PSM ds,t =.25ns, st & 2 nd PSM ds,t =.5ns, st & 2 nd PSM ds,t =.75ns, st & 2 nd PSM th,t =.25ns, st & 2 nd =.5ns, st & 2 nd PSM th,t =.75ns, st & 2 nd PSM ds,t =.25ns, st & 4 th PSM ds,t =.5ns, st & 4 th PSM ds,t =.75ns, st & 4 th =.25ns, st & 4 th =.5ns, st & 4 th =.75ns, st & 4 th PSM ds&th,umts: UL/DL,2 nd & 4 th, P I,tot = dbm, f c =.95GHz, f c2 = 2.4GHz rd PSM ds&th,umts: UL/DL,3 & 4 th, PI,tot = dbm, f =.95GHz, f = 2.4GHz c c2 = 6 MHz PSM ds,t =.25ns, 2 nd & 4 th PSM ds,t =.5ns, 2 nd & 4 th PSM ds,t =.75ns, 2 nd & 4 th =.25ns, 2 nd & 4 th =.5ns, 2 nd & 4 th =.75ns, 2 nd & 4 th Theor. antiodal in AWGN Channel E b /N o [db] PSM ds,t =.25ns, 3 rd & 4 th, P I PSM ds,t =.5ns, 3 rd & 4 th, P I PSM ds,t =.75ns, 3 rd & 4 th, P I =.25ns, 3 rd & 4 th, P I =.5ns, 3 rd & 4 th, P I =.75ns, 3 rd & 4 th, P I E b /N o [db] Figure 6. Performances of ulse shae modulated UWB systems with the resence of coloured Gaussian noise interference at UMTS ulink and downlink bands in. Different ulse airs are used. Binary TH PPM with coloured Gaussian noise P, T=.5 ns P2, T=.5 ns P3, T=.5 ns P4, T=.5 ns P3, T=. ns P4, T=. ns P, T=.5 ns, UL P3, T=.5 ns, UL Theor. orthogonal Theor. antiodal T =.5 ns BPAM with coloured Gaussian noise DS Pulse DS Pulse2 DS Pulse3 DS Pulse4 TH Pulse TH Pulse2 TH Pulse3 TH Pulse4 Theor. antiodal BER T =.5 ns Multiband interference: B J =UMTS UL B J2 =UMTS DL P = db J,tot PG =2 db tot BER T =.5 ns Multiband interference: =UMTS UL 2 =UMTS DL P = db I,tot PG =2 db tot 5 Otimum PPM modulation index in is used Better erformance could be achieved with overlaing ulses than with orthogonal ulses 5 6 E /N b 6 E /N b a) b) Figure 7. Performances of PPM and PAM modulated UWB systems with the resence of coloured Gaussian noise interference at UMTS ulink and downlink bands in.