Deep Reinforcement Learning for Age and Energy Tradeoff in Internet of Things Networks

It is critical to reach a balance between information freshness and energy consumption in energy-limited Internet of Things networks. In this paper, we consider that the update scheduling decisions are affected by the limited spectrum resources. Each sensor is assumed to have the caching capability to store the generated packets and send updates to the receiver after spectrum resources are available. We utilize Age of Information (AoI) to measure the information timeliness and construct a sensor update scheduling problem to minimize the cost function, which simultaneously considers AoI and transmission energy consumption. Then, we rephrase the problem as a Markov decision process and attempt to obtain an optimal scheduling strategy to strike AoI and energy tradeoff via deep reinforcement learning. Finally, simulation results demonstrate that our proposed algorithm can significantly decrease the transmission energy consumption of sensors while maintaining near-optimal AoI performance.