Analysis of signaling and coding schemes for non-coherent ultra-wideband systems

Ultra-wideband (UWB) radio is a new emerging technology which promises to bring a real revolution in the field of local area wireless communications. UWB is based on a shift in spectrum management paradigm which consists on allowing users to transmit over a shared bandwidth of several GHz rather than allocating private bandwidth to each user. No significant interference is caused to other UWB users nor to other systems coexisting on the same frequency bandwidth thanks to the fact that the transmitted power is constrained to have a very low spectral density. Given the huge bandwidth used by UWB systems, the latter operates in the low spectral efficiency regime. In our work we investigate the impact of UWB characteristics on the design of adequate signalling and coding schemes. Motivated by the fact that channel knowledge is not required to achieve channel capacity for vanishing spectral efficiency, we consider non-coherent type of detection. We first evaluate the performance of practical non-coherent schemes using on-off signalling. We then investigate the impact of channel estimation, made possible by channel stationary, on system performance and show at which extinct practical coherent UWB systems can outperform non-coherent ones. Later we introduce a multi-carrier UWB signalling scheme which generalizes the concept of on-off signalling to the time-frequency 2-dimentional signalling space. We analyze the performance of this signalling scheme by deriving lower and upper bounds on its achievable data rates over the set of all frequency taps correlation profiles. We then consider UWB on-off signalling in the context of peer-to-peer multiple access networks. We propose a quantized threshold-based non-coherent receiver whose performance is shown to approach the performance of a genie aided receiver. Finally we propose some practical channel code constructions that are specially designed for non-coherent UWB m-ary PPM systems. The code design uses an exit chart analysis. Code performance is then measured through simulations.