Energy Conscious Interactive Communication for Sensor Networks

Abstract

In this work, we are concerned with maximizing the lifetime of a cluster of sensors engaged in single-hop communication with a base-station. In a data-gathering network, the spatio-temporal correlation in sensor data induces data-redundancy. Also, the interaction between two communicating parties is well-known to reduce the communication complexity. This paper proposes a formalism that exploits these two opportunities to reduce the number of bits transmitted by a sensor node in a cluster, hence enhancing its lifetime. We argue that our approach has several inherent advantages in scenarios where the sensor nodes are acutely energy and computing-power constrained, but the base-station is not so. This provides us an opportunity to develop communication protocols, where most of the computing and communication is done by the base-station. The proposed framework casts the sensor nodes and base-station communication problem as the problem of multiple informants with correlated information communicating with a recipient and attempts to extend extant work on interactive communication between an informant-recipient pair to such scenarios. Our work makes four major contributions. Firstly, we explicitly show that in such scenarios interaction can help in reducing the communication complexity. Secondly, we show that the order in which the informants communicate with the recipient may determine the communication complexity. Thirdly, we provide the framework to compute the $m$-message communication complexity in such scenarios. Lastly, we prove that in a typical sensor network scenario, the proposed formalism significantly reduces the communication and computational complexities.