Improving Energy Efficiency of Centrally Controlled Wireless Data Networks

Abstract

Wireless network access protocols can assist nodes to conserve energy by identifying when they can enter low energy states. The goal is to put all nodes not involved in a transmission into the doze state. However, in doing so, one must tradeoff the energy and other costs associated with the overhead of coordinating dozing with the energy savings of putting nodes to sleep. In this paper, we define three alternative directory protocols that may be used by a central node to coordinate the transmission of data and the dozing of nodes. We attempt to optimize their performance by using scheduling and protocol parameter tuning. In addition, we consider the impact of errors and error recovery methods on energy consumption. Although one can argue that carefully scheduling transmissions will improve performance, ultimately, appropriately tuning protocols reduces scheduling's significance. In most cases, scheduling transmissions between the same nodes contiguously and ordering such transmissions shortest processing time first results in good performance. The most critical feature that contributes to an access protocol's effectiveness is its ability to minimize the time it takes to inform nodes that they may doze. However, the ability of our protocols to conserve energy is highly dependent on (1) network size, (2) traffic type (e.g., down/uplink, and peer-to-peer) and (3) channel bit error rate. In particular, we show that when protocols are faced with packet errors, more elaborate schemes to coordinate the dozing of nodes can pay-off. We conclude by recommending an energy conserving implementation of the IEEE 802.11 Point Coordination Function.