Neither Shortest Path Nor Dominating Set: Aggregation Scheduling by Greedy Growing Tree in Multihop Wireless Sensor Networks

Data aggregation is a fundamental task in multihop wireless sensor networks. Minimum-latency aggregation scheduling (MLAS) seeks to minimize the number of scheduled time slots to perform an aggregation. In this paper, we present the first work on a solvable mathematical formulation of the MLAS problem. The optimal solution of small example networks suggests that an optimal scheduling can be neither shortest path nor dominating set based. Instead of yet another theoretical analysis with provable bounds, our work focuses on reducing the average latency of general random topologies. Inspired by backward induction theory, we propose to schedule the aggregation in a reverse order, and the tree construction of Greedy Growing Tree (GGT) is directly guided by the scheduling algorithm in a step-by-step way. By following priority rules when we schedule candidate 〈sender, receiver 〉 pairs, the opportunity of parallel transmission is maximized. As a result, the aggregation latency can be minimized. Our extensive evaluation results demonstrate the superiority of the GGT algorithm: For sparse networks, the resultant latency is comparable with the best practice, whereas for high-degree networks, the latency is only half of that using state-of-art competitors.