Tree-based Redundancy Control for Achieving Desired Level of Reliability in Wireless Sensor Networks

Abstract

Data aggregation, a process of delivering data from measurement sources (sensors) to the sink, is a major task in wireless sensor networks (WSNs). A data aggregation protocol often has two goals: reliability and energy-efficiency. Reliability of data aggregation considers how many packets from sources successfully arrive at the sink, whereas energy-efficiency depends on how many transmissions (and receptions) are needed in order to collect data at the sink. The two goals are contradictory, since more transmissions are required to achieve high reliability using unreliable links. In this paper, we consider WSN applications where a level of reliability is given as a requirement. For these applications, the goal of an aggregation protocol is to use minimum packet transmissions to achieve the desired level of reliability. We propose a data aggregation protocol which controls redundancy of packets based on observed reliability level. The proposed protocol is tree-based, but also uses multipath forwarding opportunistically in order to increase reliability with minimal cost. Simulation results show that the proposed protocol successfully achieves the desired level of reliability using significantly less packet transmissions compared to existing major data aggregation protocols.