Spreading Aggregation: A distributed collision-free approach for data aggregation in large-scale wireless sensor networks

Abstract

Recently, numerous works have shown that serial aggregation in large wireless sensor networks is scalable and very efficient, in terms of avoiding collisions and conserving energy and, more importantly, in terms of reducing response time. In this paper, a novel serial data aggregation approach, called Spreading Aggregation (SA), is proposed with the aim of shortening the traversal path and further reducing communications. First, given the fact that it is not based on a pre-established itinerary, SA is data structure maintenance-free and does not require any communications in this regard. Each time an aggregation process is launched, a new path is built, which decreases vulnerability to failure in links and nodes and allows the approach to handle topology changes. Second, SA is a localized approach that relies only on the one-hop neighbors table of each node to gradually construct the path, which makes it very scalable. A third interesting feature of SA is the merge of path construction and data processing. While the path is progressively constructed, data is simultaneously aggregated, saving a considerable amount of time and energy. In addition to all that, SA also saves energy and time due to its collision-free nature. In fact, in SA, only one packet is present in the entire network at any given time. In this paper, we formally prove the correctness of SA (i.e., free of looping and ensures the traversal of all connected nodes). Furthermore, the extensive OMNeT++ simulations, we performed, confirm that the proposed approach reduces communications, scales well in large networks, and conserves time and energy. The obtained results also show that SA outperforms state-of-the-art serial approaches.