Energy efficient hardware and software solutions for wireless network access

Abstract

To develop a solution for load-proportional energy usage during wireless network access, this thesis considers networking hardware with support for dynamic frequency scaling (DFS). The proposed system is based on changing the operating clock frequency of the network interface cards as a function of the offered load, in order to reduce to energy consumption. The first contribution of the thesis is the design of a frequency selection mechanism that tries to minimize the energy consumption. The second contribution of the paper is the design of a pipelined architecture for the implementation of DFS that reduces the delays associated with the use of DFS. The third contribution of the thesis is the development of a queuing model to evaluate the energy-delay tradeoff associated with use of DFS on wireless network interface cards. Finally, the last contribution of this thesis is the development and analysis of a threshold based sleep scheduling mechanism for energy savings in a wireless network.

One of the fundamental engineering challenges for wireless networks is the development of hardware and software solutions for energy efficiency. While significant effort has been devoted towards the development of energy efficient protocols at all levels of the protocol stack, a number of challenges still remain unsolved. Among these, an important issue concerns the rate of energy consumption by network hardware such as network interface cards. In most cases, the power consumed by such devices is independent of the offered load. As a result, the power consumption of the device stays constant and at levels close to the maximum, even at low traffic loads. Thus an issue of particular importance is the development of wireless network access mechanisms that provide load-proportional energy usage. The objective of this thesis is to address the issue of load-proportional energy use in wireless networks.