Identifying Energy Holes in Randomly Deployed Hierarchical Wireless Sensor Networks

Abstract

This paper proposes a novel protocol, called an aggregation-based topology learning (ATL) protocol, to identify energy holes in a randomly deployed hierarchical wireless sensor network (HWSN). The approach taken in the protocol design is to learn the routing topology of a tree-structured HWSN in real-time, as an integral part of the sensed data collection and aggregation process in the network. The learnt topology is then examined to identify high energy-consuming nodes, that are more likely to create energy holes in the network. The major challenge in designing this protocol is to code topology data in such a way that it can be carried in length-constrained messages supported by current sensor technologies. To address this challenge, three topology coding methods are proposed. A theoretical analysis of the three topology coding methods is carried out to find the optimum method among the three, and this optimum method is used in the ATL protocol. The ATL protocol is tested and evaluated on a real WSN test bed in terms of completeness, correctness and energy costs. Based on the evaluation results, we have identified two classes of high energy-consuming nodes, which are: 1) nodes that carry topology data from more downstream nodes and 2) nodes that more frequently switch between different upstream nodes. This finding is significant as it provides an insight as how topology-learning, as well as data collection, may be used to prolong the life-time of a HWSN. In addition, the evaluation results also show that the energy cost incurred in a data collection process integrated with our proposed topology-learning facility is at a similar level as for the process without the facility, thereby implying that the cost incurred in topology-learning by using our proposed method is negligible. These findings indicate that, by integrating the topology-learning process with the sensed data collection and aggregation process, the ATL protocol can identify high energy-consuming nodes, i.e., nodes that are more likely to create energy holes, in a random HWSN deployment, in an effective and cost-efficient manner.