Cross-Layer Optimization for Wireless Communication Networks

Cross-Layer Design (CLD) has been widely studied and developed to optimize resource allocation decisions in wireless communication networks. Various design schemes have been proposed for the purpose of enhancing performance of resource restricted, error-prone wireless networks. This dissertation presents two CLDs for wireless networks; one promotes network energy efficiency, the other provides less resource-intensive optimization and maximizes system reward.;The energy efficient routing algorithm keeps records of wireless sensors' energy levels, utilizes this information to determine routing decisions at an upper layer. With this algorithm, network power consumption is distributed among all sensors, which in turn prolongs the lifetime of the network. Simulations and comparisons shown in the paper of energy conserving routing algorithm demonstrate that this algorithm gains more than 40% improvement on energy saving over the energy-efficient m-coverage and n-connectivity routing algorithm. This algorithm is a CLD in which the network layer gets information about the physical layer, including remaining energy and transmission power, to make better routing decisions in real time.;By contrast, the autonomous CLD accumulates information from lower layers, passes it to upper layers. And vice versa. Each layer only communicates with its neighbors. Each layer has its own optimizer which can be run in real-time to determine the best transmission parameters in order to maximize the wireless user's system utility. This optimization accounts for different Quality of Service (QoS) requirements involving different communication protocols. Comparing to centralized optimization, which uses a centralized optimizer to gather the environment dynamics of all layers before calculating best transmission strategies for each layer, this solution requires less computation time on the resource constrained wireless devices and adapts to various data sources quickly. This theoretical model describes many practical networks. Ns-3 simulations of this algorithm for a Wi-Fi network demonstrate improved network performance, including maximizing network throughput, which is almost doubled over an unmanaged solution, lowering transmission cost by more than one half and reducing transmission delay. Comparing the improved autonomous CLD with the original design proposed by F. Fu et al., the improved version has more robust performance with more transmission cost tolerance and faster optimization calculation.