SOURCE IEEE INFOCOM’20 AoI Workshop, Toronto, ON, Canada, Canada, Jul. 6-9, 2020
Published Date: Jul. 6-9, 2020
For many time-critical Internet of Things applications, the performance depends heavily on the freshness of information. A fundamental problem in this scenario is how to support heterogeneous information traffic whose freshness may be measured by different metrics. This paper studies a single-link wireless communication system where a sender supports two types of traffic—status update and delay-constrained traffic. The sender decides whether to serve the delay-constrained traffic or sample an underlying status process and update the receiver (central controller) on the status. The optimal scheduling policy is designed to minimize the long-term average age of information (AoI) of the status at the central controller while guaranteeing minimum timely-throughput of the delay-constrained traffic. This problem is first formulated as a Constrained Markov Decision Process (CMDP) and then converted into unconstrained MDP by Lagrangian relaxation. The structural property of the optimal policy for the CMDP is derived and an optimal policy is developed. Moreover, considering the computation overhead of MDP, we develop a rather simple scheduling policy based on the Lyapunov-drift method. The performance is analyzed theoretically and verified by simulations.