This paper considers a network of energy-constrained sensors, wherein each of sensors periodically produces information that is transmitted to a base station for further processing. During data gathering, sensors have the ability to perform in-network aggregation (fusion) of data packets en route to the base station. A key challenge in such data gathering is to maximize the system lifetime, given the energy constraints of each sensor. This paper develops an algorithm that, given the location of each sensor and a base station, together with available energy at each sensor, provides a schema in which data could be collected such that the system lifetime is maximized. Further, this algorithm is shown to provide a worst-case running time polynomial in , which is said to significantly outperform previous algorithms in terms of system lifetime.

Kalpakis, Dasgupta, and Namjoshi address the more general construct of distributed data gathering and aggregation in sensor networks, in terms of the maximum lifetime data aggregation (MLDA) problem. They first develop a near-optimal polynomial-time algorithm for solving the MLDA problem that, while said to perform significantly better than existing protocols in terms of system lifetime, is computationally expensive for large sensor networks. They also develop a clustering-based heuristic approach for maximum lifetime data gathering and aggregation in large-scale sensor networks. Finally, the authors present experimental results to show that this latter approach for solving the MLDA problem for smaller networks achieves system lifetimes that are 1.1 to 2.3 times better, when compared to an existing data gathering protocol. Moreover, for larger networks, this clustering-based heuristics approach is shown to achieve up to a 2.6 times increase in system lifetime, when compared to the same protocol.

This contribution toward development of a polynomial-time near-optimal algorithm for addressing the MLDA problem and the experimental verification of its performance is a useful step toward improved understanding of the design and application of distributed sensor networks. A simplifying assumption is made whereby each sensor is allowed to aggregate its own data packets with those of any other sensor in the network. A more useful, but far more complex scenario would constrain certain sensors to aggregate their own data with only certain sensors, while acting as routers for other incoming packets. In addition, an extension to this algorithm incorporating variable processing time delays within individual sensors, as experienced in real-world sensor networks, would permit improved characterization of design tradeoffs between allowable sensor delays and the lifetime achieved by the distributed sensor network. The authors indicate they plan to investigate these and related issues in future work.