Scheduling and Power Control for Wireless Multicast Systems via Deep Reinforcement Learning

Multicasting in wireless systems is a natural way to exploit the redundancy in user requests in a content centric network. Power control and optimal scheduling can significantly improve the wireless multicast network’s performance under fading. However, the model-based approaches for power control and scheduling studied earlier are not scalable to large state spaces or changing system dynamics. In this paper, we use deep reinforcement learning, where we use function approximation of the Q-function via a deep neural network to obtain a power control policy that matches the optimal policy for a small network. We show that power control policy can be learned for reasonably large systems via this approach. Further, we use multi-timescale stochastic optimization to maintain the average power constraint. We demonstrate that a slight modification of the learning algorithm allows tracking of time varying system statistics. Finally, we extend the multi-time scale approach to simultaneously learn the optimal queuing strategy along with power control. We demonstrate the scalability, tracking and cross-layer optimization capabilities of our algorithms via simulations. The proposed multi-time scale approach can be used in general large state-space dynamical systems with multiple objectives and constraints, and may be of independent interest.

Transmission Rate Sampling and Selection for Reliable Wireless Multicast

The multicast communication concept offers a scalable and efficient method for many classes of applications; however, its potential remains largely unexploited when it comes to link-layer multicasting in wireless local area networks. The fundamental lacking feature for this is a transmission rate control mechanism that offers higher transmission performance and lower channel utilization, while ensuring the reliability of wireless multicast transmissions. This is much harder to achieve in a scalable manner for multicast when compared with unicast transmissions, which employs explicit acknowledgment mechanisms for rate control. This article introduces EWRiM, a reliable multicast transmission rate control protocol for IEEE 802.11 networks. It adapts the transmission rate sampling concept to multicast through an aggregated receiver feedback scheme and combines it with a sliding window forward error correction (FEC) mechanism for ensuring reliability at the link layer. An inherent novelty of EWRiM is the close interaction of its FEC and transmission rate selection components to address the performance-reliability tradeoff in multicast communications. The performance of EWRiM was tested in three scenarios with intrinsically different traffic patterns; namely, music streaming scenario, large data frame delivery scenario, and an IoT scenario with frequent distribution of small data packets. Evaluation results demonstrate that the proposed approach adapts well to all of these realistic multicast traffic scenarios and provides significant improvements over the legacy multicast- and unicast-based transmissions.