Ultra-wideband technology and wireless sensor networks: Acquisition and distributed inference.

摘要

The explosive growth in demand for wireless applications propels breakthroughs in wireless technologies and is changing the way we access, exchange and process information. This dissertation consists of three thrusts in the following two main areas of wireless communications which have attracted considerable attention in recent years: impulse radio (IR) ultra-wideband (UWB) and distributed inference in wireless sensor networks (WSNs).;The first thrust studies the coarse acquisition of UWB signals in low-data-rate applications, such as WSNs. In particular, the tradeoff between the multipath energy combining capabilities of three fundamental detection schemes, including the single pulse correlator (SPC), the energy detector (ED), and the transmitted-reference (TR), is investigated for the coarse acquisition. A systematic comparison with full analysis is carried out in terms of mean acquisition time and false acquisition rate. The results show that the multipath energy combining capabilities of the ED and TR schemes can improve the acquisition performance greatly compared to the SPC scheme. Useful guidelines for important design parameters are provided for practical systems.;The second thrust provides a comprehensive study of distributed detection in UWB WSNs over frequency selective channels. These include schemes with different requirements on channel state information at the sensors and fusion center. The error exponent and asymptotic optimality for these schemes under different energy, and time-bandwidth product requirements are derived. The study reveals important tradeoffs between feedback overhead, synchronization requirements, energy and time-bandwidth product.;The third thrust explores scaling laws of the outage for distributed inference problems over fading channels with respect to the total power and the number of sensors. Tight upper and lower bounds on outage diversity are derived and shown to depend on not only the number of sensors but also the sensing signal-to-noise ratio (SNR) of the sensors. The results indicate that adding new sensors might not add to the diversity order. A large class of envelope distributions for the wireless channel is studied. Finally, it is shown that the outage decays faster than exponentially if fixed power per sensor with an asymptotically large number of sensors is considered.