Fall 2010 IAB
September 15 to 17
Protocol-Agnostic Compression for Resource-Constrained Wireless Networks
In wireless devices, reducing the time the radio is on results in lower power consumption. In resource-constrained wireless networks, then, sending the same data in fewer bytes can greatly extend the lifetime of the network. In this paper, we explore the use of protocol agnostic packet compression, a technique orthogonal to current explicit compaction techniques. Such a compression algorithm functions as a transparent layer inside a communication stack. Because it makes no assumption on the specific protocols used, it is generic enough to be used on multiple technologies.
Compression is done by identifying patterns in recently sent packets and replacing those patterns with bit flags, effectively reducing the size of the packet to be sent. We discuss the tradeoffs between computation, memory costs and power savings in such an algorithm. We then present the results of compressing actual packet traces collected from several commercial networks using this algorithm. Results indicate compression ratios between 40% and 80%, which yield savings of 30-70% in the average power consumption of a typical network.