STAC: a spatio-temporal approximate method in data collection applications

Abstract

Wireless sensor networks (WSNs) and IoT are often deployed for long-term monitoring. However, the network lifetime of these applications is limited by non-rechargeable battery-powered. To vastly reduce energy consumption, this paper proposes a spatio-temporal approximate data collection (STAC) method to prolong the network lifetime. Under the tolerable accuracy, STAC utilizes spatial correlation among neighbors to select partial network for data collection with balanced energy distribution, and takes advantage of temporal redundancy to dynamically adjust the sampling interval by Q-learning based method. With the spatio-temporal approximate and correlation-variation verification mechanism, STAC prolongs the network lifetime with error-bounded data precision. Simulation results demonstrate STAC significantly improves network lifetime in various circumstances.