A novel trade-off between communication and computation costs for data aggregation in wireless sensor networks

摘要

Wireless Sensor Networks (WSNs) consist of inexpensive low-power miniature sensing devices with severe power constraints, necessitating energy-efficient solutions for networking operations. Major prior art proposals have been primarily directed towards minimising the communication cost, either implicitly assuming away the computation overhead as being negligible, or radically trading against it. However, in computation-bound scenarios, dealing with a large volume of data, such simplifying assumptions or radical measures tend to be inefficient. In this paper, we investigate the problem of minimising the overall energy needed to send data from a set of sensor nodes to a single destination, where each node is in charge of a mission. Two types of missions are defined: sensing and decision making; while source nodes are only in charge of sensing, relay nodes can carry out both missions simultaneously. More specifically, given a node's current backlog and its latest view on the relevant portion of the data-gathering tree, taking on the decision-making mission involves deriving an online trade-off between energy costs of compression and communication, and deciding between sending data either in the raw mode or alternatively compressed with a feasible optimal compression ratio. The used data compression technique depends on the type of application and the spatiotemporal correlation in the packets. Simulation experiments reveal that, compared with previous methods, the proposed scheme exhibits superior energy efficiency with an additional 36% reduction of the costs.