Date of Award
2008
Degree Type
Thesis
Degree Name
Master of Engineering Science
Program
Electrical and Computer Engineering
Abstract
In the information theory, the channel capacity states the maximum amount of in formation which can be reliably transmitted over the communication channel. In the specific case of multiple-input multiple-output (MIMO) wireless systems, it is well recognized that the instantaneous capacity of MIMO systems is a random Gaussian process. Time variation of the capacity leads to the outages at instances when it falls below the transmission rate. The frequency of such events is known as outage probability.
The cross-layer approach proposed in this work focuses on the effects of MIMO capacity outages on the network performance, providing a joint optimization of the MIMO communication system. For a constant rate transmission, the outage prob ability sensibly affects the amount of information correctly received at destination. Theoretically, the limit of the ergodic capacity in MIMO time-variant channels can be achieved by adapting the transmission rate to the capacity variation. With an accu rate channel state information, the capacity evolution can be predicted by a suitable autoregressive model based on the capacity time correlation. Taking into consider ation the joint effects of channel outage at the physical layer and buffer overflow at the medium access control (MAC) layer, the optimal transmission strategy is derived analytically through the Markov decision processes (MDP) theory. The adaptive pol icy obtained by MDP is optimal and maximizes the amount of information correctly received at the destination MAC layer (throughput of the system). Analytical results demonstrate the significant improvements of the optimal variable rate strategy com pared to a constant transmission rate strategy, in terms of both system throughput and probability of data loss
Recommended Citation
Luccini, Marco, "Cross-Layer Optimization of Network Performance over MIMO Wireless Mobile Channels" (2008). Digitized Theses. 4688.
https://ir.lib.uwo.ca/digitizedtheses/4688