MIMO Link Adaptation for Real Systems

Omar Ahmed Nasr
Advisor: Babak Daneshrad

Wednesday, November 25, 2009 at 10:00am
Engr IV Maxwell Room 57-124

Abstract:
The most important advancement in the field of physical layer wireless communication in the last twenty years is the use of Multiple Input Multiple Output (MIMO) transmission techniques. In this thesis, we explore few possible gains by using MIMO technology in both MAC layer and the physical layer.

From MAC layer point of view, we use game theory to show that adding multiple antennas at the base station can be used to punish any selfish user that tries to manipulate the network. We show that the baseband can use its spatial selectivity to force the network reaching a fair and unique Stackelberg equilibrium and punish any selfish user.

From physical layer point of view, we tackle two problems. The first is the optimal bit assignment on different subcarriers in a MIMO-OFDM system. In our thesis, we introduce new discrete bit loading algorithms to solve the margin maximization and the rate maximization problems, with total energy defined as the sum of transmission energy and circuit energy. Compared to the conventional discrete bit loading algorithms, the new algorithms can achieve savings in the order of 9 dB in the energy minimization problem. 180% increase in the number of transmitted bits can also be achieved with the same total energy when compared to the existing algorithms.

The second physical layer problem considered is performance estimation for MIMO systems when linear decoders are used in the receiver. The general rule of thumb is to use spatial multiplexing when the channel is "good" and use diversity when the channel is "bad". The quality of the channel though is not well defined in the literature. We clarify that the channel singular values, which have been proposed in the literature as a quality metric, can not be a reliable estimator of the quality of the channel. We propose the use of "scaled singular values" to accurately estimate the quality of the MIMO channel and estimate the system performance in both coded and uncoded MIMO systems. A channel emulator and a complete hardware implementation of a coded MIMO system were used to validate our results.

Biography:
Omar Ahmed Nasr received the B.S. and M.S. degrees with honors in EE from Cairo University, Egypt, in 2003 and 2005, respectively. Since 2005, he has been with UCLA where he is working towards the Ph.D. degree.