Energy-Efficient Wireless Communication

Rapidly, the world is getting more and more connected. Internet plays a prominent role in this trend. Furthermore, people are becoming increasingly mobile. The combination of both trends results in a huge demand for wireless communication. This thesis addresses the problem of energy-efficient wireless communication with an adequate quality of service. The scarce energy resources of a mobile require an energy-efficient operation. The dynamic environment of a mobile causes large quality variation of the transport medium and requires a continuous monitoring of the quality of service of the wireless link. This thesis provides a general framework to build an energy-efficient wireless link. The central idea is to determine the exact amount of effort that is necessary to obtain the required quality given a current wireless medium. So the key is to be effective, only doing the work that is really necessary. Doing more work than necessary means waste of resources, doing less work means that the required quality of the link can not be achieved. The information about the required amount of effort for the current circumstances is used by a control system to adapt the receiver so that it will operate with minimal energy consumption. The framework is illustrated by an example of a UMTS link that is worked out in detail. A simulation environment for this example was built to illustrate the proper working of the proposed methods and to show that substantial savings can be achieved by applying these methods. A second example, which uses a completely different technology for the transmission, HiperLAN/2, is also presented. It is shown that the proposed methods can be applied to this example as well. Contributions are made on different areas: First, an algorithm is developed that provides a good indication of the quality of the medium (estimation of the BER) based on statistical analysis of the received symbols, without transmission of additional information or knowledge of the environment. Simulations show that the mean error of the