DEST: Distributed endurant spanning tree for data aggregation on wireless sensor networks

In this paper, an energy-aware distributed algorithm is proposed to construct an endurant spanning tree for data aggregation on wireless sensor networks. On the constructed aggregation tree, nodes with higher residual energy are arranged close to the trunk to maximize the lifetime of the tree, and maintain the integrity of aggregated data. Data aggregation, in which a node processes the data collected from all its children nodes and then transmits the processed result to its parent node by a single message, is an energy efficient way to collect data from sensor nodes. However, in the tree hierarchy for data aggregation, a node failure close to the root of the tree (the sink node of a sensor network) will cause severe data loss (all data from the downstream of this failure node will be lost.). This motivates our algorithm to arrange those nodes with higher residual energy close to the root of aggregation tree and relieve the responsibility of nodes with less residual energy. Our algorithm for tree construction is distributed; each node makes its own decision by exchanging information with its neighbouring nodes. The experiment results show that the constructed aggregation tree by our algorithm, named Distributed Endurant Spanning Tree (DEST) is the most endurant, i.e., of the longest time to reach a certain level of data loss due to node failures from running out of energy, compared to other representative energy-aware aggregation trees in the literature. This also indicates a less frequent re-construction of data aggregation tree, which further reduces the energy consumption on sensor networks.