Performance of TH and DS UWB Multiaccess Systems in Presence of Multipath Channel and Narrowband Interference G. Durisi Istituto Superiore Mario Boella Corso Trento 21 I-10129 Torino, Italy E-mail: durisi@ismb.it J. Romme IMST GmbH Carl Friedrich Gauss Str. 2 D-47475 Kamp-Lintfort, Germany E-mail: romme@imst.de S. Benedetto CERCOM-Politecnico di Torino Corso Duca degli Abruzzi 24 I-10129 Torino, Italy E-mail: benedetto@polito.it Abstract— The probability of error of UWB systems, employ- ing two different multiaccess techniques (Direct Sequence - DS, and Time Hopping - TH) is analytically evaluated, in presence of multiuser, narrowband interference and multipath channel. Rake receiver and MMSE equalizer performance is compared. An uplink scenario is considered, with completely asynchronous transmissions; the effect of path loss is also taken into account. TH is shown to be as robust as DS in presence of strong narrowband and multiuser interference and dense multipath channel. I. I NTRODUCTION Ultra-wideband (UWB) systems development for high speed indoor communications requires an analysis to establish which type of multiaccess and modulation format and which receiver structure offers the best compromise between complexity and robustness against multipath, multiuser and narrowband interference. A comparison is presented between Time Hopping (TH) and Direct Sequence (DS) multiple access schemes, both for Rake and MMSE receiver structures. TH is part of the original proposal for UWB communications (see, for example, [1]), while DS is an established multiuser technique, that was recently considered also for UWB applications [2]. As shown in [3], DS can better deal with multiuser interference (MUI) than TH, on AWGN channels, representing therefore a promising candidate. In this paper a more detailed analysis is illustrated, in which the presence of dense multipath channels and strong narrowband interference is taken into account. The equal power user assumption in [2], has been removed in our analysis, through the introduction of a user dependent path loss attenuation. Furthermore, in those cases where the use of Gaussian approximation for the interference term is potentially loose, the results have been validated by a semi- analytical method. This work was partially sponsored by MIUR (Italian Ministry of Education and Research) under the project CERCOM and PRIMO and by the European Union under project number IST-2000-25197-whyless.com II. DS-2PAM The signal transmitted by the user k in a DS-2PAM system can be written in the following way: s (k) (t)=  E (k) c +∞  i=−∞ b (k) i Nc−1  l=0 c (k) l x (t − lT c − iT f ) (1) where x(t) is the transmitted pulse, normalized to have unitary energy and with time duration equal to T x . In our analysis, it is modelled with the second derivative of a gaussian pulse, the typical waveform considered in the literature [1]. The sequence  b (k) i  represents the stream of equiprobable binary information bits transmitted by the source. As binary PAM is employed as modulation technique, then b (k) i ∈ {±1}. The time axis is divided into frames of length T f , each corresponding to one bit interval. Each frame is subdivided into N c chips of length T c . It is also assumed that T c ≥ T x .A vector of length N c , c k =[c (k) 0 ,c (k) 1 ,...,c (k) Nc−1 ] T ,  c (k) i  ∈ {±1}, describes the spreading code assigned to user k. Finally, E (k) c is the energy per transmitted pulse for user k. The energy per bit E (k) b can be obtained noting that E (k) b = N c E (k) c . Assuming that N u users are transmitting, then the signal at the receiver can be written as r(t) = Nu  k=1  A k E (k) c  i b (k) i Nc−1  l=0 c (k) l q (k) (t − lT c − iT f )+ n(t)+  A b n b (t) (2) where n(t) is a white Gaussian noise process with two-sided power spectral density N 0 /2 and n b (t) is the narrowband interference. Furthermore, q (k) (t)= x(t) ∗ h (k) (t) (the symbol “∗” denotes convolution), where h (k) (t) is the time-invariant, asynchronous multipath channel impulse response for user k. Each channel impulse response h (k) (t) is assumed to have a maximum delay spread of τ (k) max seconds. Finally A k , A b represent the attenuations due to path loss, which are function of the transmitter receiver (TX-RX) distance. In order to evaluate the bit error probability (BER) ana- lytically, we construct a discrete time equivalent model of