Optimal in-network packet aggregation policy for maximum information freshness

The Internet of Things is emerging as more wireless sensor networks get deployed every day, pushing decision making and processing towards the edge onto more constrained devices. In these constrained networks, using a single packet to transmit small sensed data is an inefficient use of both bandwidth and energy. Aggregating multiple measurements and packets into a single packet increases efficiency and resource utilization, but it also introduces the problem of deciding when to transmit aggregated data. In this work, we show theoretically that the classical performance metrics of energy consumption and expiration rate create a balanced tradeoff, where their combined metric of energy consumption per non-expired measurement is a constant, independent of policy selection. We introduce a new metric, information freshness, and derive an optimal policy to maximize it under both single hop and multi-hop cases. We provide a distributed algorithm to implement our optimal policy. Our case studies show that our algorithm outperforms state-of-the-art policies by more than 3.3x for information freshness and reduces energy consumption by more than 62%.