Performance of wireless networks with directional antennas.

Abstract

Directional antennas have been considered to be a promising technology that can improve user capacity and power efficiency for wireless networks. In this thesis, we analyze the performance of a wireless network using directional antennas with an antenna beam pattern of arbitrary shape under a realistic channel model which includes the effects of path loss, shadowing, and Rayleigh fading. The performance is analyzed for both slow fading and fast fading cases. The performance measure considered is the outage probability. A mobile using directional antenna needs to know the direction of its communicating party to achieve good performance. However in reality, direction estimation is imperfect and the error affects the system performance. This effect is analyzed in this thesis. Other than direction estimation error, we also consider the effect of power control which is often used to combat the power loss in wireless environments. We analyze the system performance under power control and combine the analysis with a simulated annealing algorithm to find the optimal power control scheme for any given wireless network. In order to generalize our outage probability analysis to wireless networks with different physical layers, we need to determine the threshold of signal-to-interference-noise ratio (SINR) for the outage probability computation according to the coding and modulation scheme used in the physical layer. Bit-interleaved coded modulation (BICM) and turbo codes are commonly used in wireless communications and their performance is analyzed in the thesis. This enables us to find the SINR threshold and thus compute the outage probability when BICM or turbo codes are used in the physical layer of a wireless network. Since bit-interleaved space-time (BI-ST) code is very effective in combating fading due to its time diversity and transmit diversity, we also generalize our BICM performance analysis to analyze the performance of BI-ST coded systems under various fading distributions.