Scheduling for energy conservation and quality enhancement in multi-hop wireless networks

Wireless networks and devices are ubiquitous today. This is primarily due to the increased mobility, exibility and connectivity as well as high speed and low cost deployment that they provide. However there are some major concerns of wireless devices and networks viz.; quality of service, capacity and spectral efficiency, battery lifetime and security. These concerns result from broadcasting using radio waves by the wireless devices. The broadcast nature of wireless media makes them prone to losses and high bit error rates caused by interference and multipath fading among other reasons. The variabilities and uncertainties make it challenging to guarantee traditional quality of service parameters like rate, latency and jitter. Hence an additional dimension of quality is introduced in the context of wireless, viz. adaptivity. Wireless devices also consume more power than corresponding wired devices. These devices function on limited battery capacity to facilitate portability and their mobile nature precludes frequent recharge of the battery. The operational lifecycle of these devices is governed by the efficiency of their battery usage. The objective of this thesis is to analyze and apply novel application aware cross-layer scheduling techniques for energy conservation and quality enhancement in multi-hop wireless networks. We focus on scheduling techniques as many of these can integrated with existing systems and incorporated in new designs and deployments. Application awareness helps to identify the constraints and requirements at lower network layers (network and MAC layer). This identification facilitates the lower layers to perform prioritized scheduling to optimize bandwidth utilization for class differentiated ows and reduce the impact of sleep-awake scheduling on the quality of service. Specifically, we identify three applications:OFDMA based multi-hop networks, Sensor networks and Voice over Internet Protocol (VoIP).;We model the QoS aware centralized slot allocation scheme for prioritized ows in OFDMA based networks. The model is an ILP that maximizes the described Flow Admittance metric. Several computationally tractable heuristics are compared for efficiency using different topologies and various workloads. It is seen that the performance of the 'Schedule from Frame End' heuristic matches that of the optimal solution closely in most cases. We enhance this solution for IEEE 802.16j networks to make it scalable and adaptive to rate changes. We also include variable zone allocation and adaptive scheduling period based on channel conditions and link demands.;We illustrate how a multi-channel sensor network scheduling scheme can result in considerable energy savings when used to model a moving phenomenon or to deploy a heterogenous sensor network. Moreover, these savings are achieved without any significant degradation in data gathering: quality or latency. Finally, we show how an adaptive scanning technique utilizing the sleep-awake scheduling in Wi-Fi networks can be used to conserve energy and increase the battery lifetime in VoIP Wi-Fi devices without harming the user experience.