Optimizing Information Freshness in Uplink Multiuser SIMO Systems: Low-Complexity Scheduling Algorithms

This paper develops scheduling policies to optimize the information freshness, quantified by the age of information (AoI) metric, in an uplink multi-user SIMO status update system. The multi-user scheduling problem is formulated as a Markov decision process (MDP) to derive the optimal policy that minimizes the average AoI across devices. However, the optimal policy suffers from high complexity due to dimensionality in large networks. To address this, a low-complexity max-weight (MW) policy is developed for large-scale networks using the Lyapunov optimization framework. The MW policy dynamically determines the subset of devices to schedule in each time slot by maximizing the expected AoI drop of the subsequent time slot. Simulations are conducted to compare the performance of the optimal policy, the MW policy, and the baseline fixed scheduling (FS) policy that always schedules a fixed number of devices with the highest AoI. The results show that the MW policy achieves close-to-optimal performance. Moreover, for a given network setup, there exists an FS policy with a particular number of scheduled devices that can approach the MW policy. This observation inspired the development of another low-complexity scheduling policy, termed optimized FS (OFS). This policy further optimizes the number of devices scheduled under the FS policy based on specific network configurations. Closed-form expressions for the average peak AoI and the approximated average AoI of the FS policy with a given number of scheduled devices are derived to determine the optimal number of scheduled devices for the OFS policy under different network setups. Simulation results validate the theoretical analysis and show that the OFS policy achieves performance comparable to the MW policy while circumventing the need for per-slot optimization.