By: Xinmin Deng
Advisor: Dr. Alexander M. Haimovich
Department of Electrical and Computer Engineering

Time: 11:00 AM, Monday, May 2nd, 2005
Place: Room 202, ECE Center, New Jersey Institute of Technology (NJIT), Newark NJ. Directions

Abstract

In this work, the effects of system parameters upon the performance are quantified under the assumption that some statistical information of the wireless channels is available. These results are useful in determining the optimal design of system parameters. Suboptimal receivers are designed for systems that are constraint in terms of implementation complexity.

The achievable rates are investigated for a wireless communication system when neither the transmitter nor the receiver has prior knowledge of the channel state information (CSI). The dynamics of the fading channel are characterized by a Doppler spectrum with fading rate known to the receiver. Quantitative results are provided for independent and identically distributed (i.i.d.) Gaussian signals. Expressions for the achievable rates include a lower bound on mutual information obtained by upper-bounding the penalty due to unknown CSI, and the achievable rates of pilot-aided systems with optimized resource allocation. A simple, low-duty-cycle signaling scheme is proposed to improve the information rates for low signal-to-noise ratio (SNR), and the optimal duty cycle is expressed as a function of the fading rate and SNR. It is demonstrated that the resource allocations and duty cycles developed for Gaussian signals can also be applied to systems using other signaling formats.

The average SNR and outage probabilities are examined for amplify-and-forward cooperative relaying schemes in Rayleigh fading channels for various cases of available CSI. For optimal receivers with global knowledge of CSI, it is shown that constant-amplification relaying has a higher average SNR gain and a lower diversity order than channel-inversion relaying. Optimal power allocation strategies are developed that optimize the average SNR and outage performance, respectively. When only local knowledge of CSI is available, it is demonstrated through simulations that suboptimal coherent receivers can still achieve substantial power and diversity gains.

Suboptimal algorithms are proposed for cases that optimal receivers are difficult to implement. For systems with multiple transmit antennas, an iterative method is used to avoid the inversion of a data-dependent matrix in decision-directed channel estimation. Convergence conditions are found as a function of the number of transmit antennas and the noise power. When CSI is not available, two noncoherent detection algorithms are formulated based on the generalized likelihood ratio test (GLRT). When used in systems with cooperative diversity, the GLRT-based detectors are less sensitive to the locations of the relay nodes than other suboptimal detectors.

Committee Members:

Dr. Alexander M. Haimovich (Advisor), Professor, NJIT
Dr. Ali Abdi, Assistant Professor, NJIT
Dr. Yeheskel Bar-Ness, Distinguished Professor, NJIT
Dr. Leonard J. Cimini, Professor, University of Delaware
Dr. Roy You, Assistant Professor, NJIT