Cross Layer Techniques To Enhance Link Performance In Wireless Networks

In this thesis, we improve the link performance by employing cross-layer techniques in two totally different wireless environments: a) IEEE 802.11 wireless networks and b) ultra-wideband wireless ad hoc networks. The cross-layer information exchange between the MAC and the PHY layer is used to maximize the link throughput. In the first part, we achieve the throughput gain by solving the problem of rate adaptation in IEEE 802.11 networks using the proposed Stochastic Automata Rate Adaptation (SARA) algorithm. SARA is inspired from Stochastic Learning Automata (SLA), a machine learning technique for adaptation in random environments and is fully compatible with IEEE 802.11 MAC. Next, we focus on improving the link performance in ultra-wideband wireless networks by exposing the MAC layer to the presence of multiple orthogonal pulse shapes at the physical layer. These pulse shapes have been recently proposed for ultra-wideband systems and have been used to improve the system performance at the physical layer. However, at the MAC layer, very few designs based on utilizing these pulse shapes have been proposed. In this thesis, we first propose an alternative way to use these pulse shapes at the physical layer to achieve high data rates. Next, we show that these pulse shapes can increase both the rate and the reliability of packet transmission at the MAC layer. To achieve better reliability, we introduce and develop the idea of 'pulse shape redundancy'. Two throughput optimization techniques based on these ideas and findings are then analytically developed. After this, we propose a new MAC protocol, namely Redundancy-Enabled MAC (REMAC), which is capable of employing either of the two proposed schemes. Finally, we illustrate the performance gain achieved by REMAC and the role of redundancy through extensive simulation results and discussions.