By: Hongsan Sheng
Advisor: Dr. Alexander Haimovich
Department of Electrical and Computer Engineering

Time: 10:30 AM, Tuesday, April 26th, 2005
Place: Room 202, ECE Center, New Jersey Institute of Technology (NJIT), Newark NJ. Directions

Abstract

Ultra-wideband (UWB) technology has received significant interest in recent years. Many questions still exist in the design and implementation for UWB systems. Transceiver implementations are particularly challenging due to regulation, interference, and channel estimation issues. This dissertation investigates the potential promises and proposes possible solutions to the challenges of designing transceivers and optimizing system parameters.

In February 2002, the Federal Communications Commission (FCC) authorized UWB devices to operate in a wide bandwidth under restrictions: a low power spectral density and a specific spectral mask. Large bandwidth enables not only large spreading ratios, but also very low coding rates. Low power constraint leads to a short range for UWB communications. Impulse radio UWB systems operate by sending sequences of baseband short pulses. The spectral mask challenges the design of pulse waveforms employed by impulse radio UWB systems. In this dissertation, a new UWB pulse shape is designed that satisfies the FCC spectral mask. Coding-spreading tradeoff is investigated, and the information theoretic channel capacity is evaluated as a function of the transmission distance under specific UWB regimes.

A core idea using UWB is that the UWB signal can cohabit in the frequency band currently occupied by existing applications without causing harmful interference. However, narrowband interference emitted by existing networks in close proximity has to be suppressed. In this dissertation, a reduced-rank adaptive filtering technique, in particular the eigencanceler, is applied to the problem of interference suppression in UWB communications.

One promise of UWB is that a UWB channel can be resolved into a significant number of multipath components. A Rake receiver can be employed to exploit the high multipath diversity. However, practical Rake receivers require knowledge of multipath delays and amplitudes. In practice, that is estimated through pilot aided channel estimation, producing imperfect channel state information, which leads to degraded performance. This dissertation investigates the impact of non-ideal estimates on system performance when path delays and path amplitudes are jointly estimated. Further, this analysis is applied to determine optimal transceiver parameters, such as the fraction of a transmitted packet constituted by the pilot symbols, signal bandwidth, and the number of diversity paths combined at the receiver.

Committee Members:

Prof. Alexander Haimovich, Advisor, NJIT
Prof. Ali Abdi, NJIT
Prof. Yeheskel Bar-Ness, NJIT
Prof. Larry Greenstein, Rutgers
Prof. Roy You, NJIT