Multiple-antenna systems in ad-hoc wireless networks

• Main Author: Govindasamy, Siddhartan, 1975-
• Other Authors: David H. Staelin and Daniel W. Bliss.
• Format: Others
• Language: English
• Published: Massachusetts Institute of Technology 2009
• Subjects: Electrical Engineering and Computer Science.
• Online Access: http://hdl.handle.net/1721.1/44410

Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 2008. === Includes bibliographical references (leaves 223-229). === The increasing demand for wireless communication services has resulted in crowding of the electromagnetic spectrum. The "spectral-commons" model, where a portion of the electromagnetic spectrum is public and used on an ad-hoc basis, has been proposed to free up spectrum that has been allocated but underutilized. Ad-hoc wireless networks (networks with no central control) are also interesting in their own right as they do not require costly infrastructure, are robust to single-node failures, and can be deployed in environments where it is difficult to deploy infrastructure. The main contributions of this thesis are expressions for the mean and in some cases the variance of the spectral efficiency (bits/second/Hz) of single-hop links in random wireless networks as a function of the number of antennas per node, link-length, interferer density, and path-loss-exponent (an environmental parameter that determines signal decay with distance), under assumptions chosen for realistic implementability in the near future. These results improve our understanding of such systems as they indicate the data rates achievable as a function of tangible parameters like user density and environmental characteristics, and are useful for designers of wireless networks to trade-off hardware costs, data-rates, and user densities. We found that constant mean spectral efficiencies can be maintained in wireless networks with increasing user density by linearly increasing the number of antenna elements per user, or by maintaining a constant fraction of nodes connected to high capacity infrastructure like optical fiber, equipped with antenna arrays. These are promising ways to serve an increasing density of users without increasing bandwidth. Additionally, several interesting features of such networks have been highlighted. === (cont.) For instance we found that the mean and variance of spectral efficiencies can be characterized in terms of a parameter called the link rank, which on average equals the number of interferers whose signal power is stronger at a representative receiver than its target transmitter. Rank thus combines the effects of node density and link lengths. Another interesting finding is that mean spectral efficiency in networks with rank-1 links, and equal numbers of antennas at transmit and receive sides can be improved if nodes turn off two thirds of their transmit antennas. These results were derived using infinite random matrix theory and validated using Monte Carlo simulations which were also used to characterize the distribution of spectral efficiencies in such networks. === by Siddhartan Govindasamy. === Ph.D.