Conclusion and Future Work

The value of this research is not merely to show that one can optimize a given algorithm to achieve a certain reduction in energy, but to show that the choice of how and whether to compress is not obvious. It is dependent on hardware factors such as relative energy of CPU, memory, and network, as well as software factors including compression ratio and memory access patterns. These factors can change, so techniques for lossless compression prior to transmission/reception of data must be re-evaluated with each new generation of hardware and software. On our StrongARM computing platform, measuring these factors allows an energy savings of up to 57% compared with a popular default compressor and decompressor. Compression and decompression often have different energy requirements. When one's usage supports the use of asymmetric compression and decompression, up to 12% of energy can be saved compared with a system using a single optimized application for both compression and decompression.

When looking at an entire system of wireless devices, it may be reasonable to allow some to individually use more energy in order to minimize the total energy used by the collection. Designing a low-overhead method for devices to cooperate in this manner would be a worthwhile endeavor. To facilitate such dynamic energy adjustment, we are working on EProf: a portable, realtime, energy profiler which plugs into the PC-Card socket of a portable device [22]. EProf could be used to create feedback-driven compression software which dynamically tunes its parameters or choice of algorithms based on the measured energy of a system.