NOTE: Restrictions are in place to limit access to one or more of the files associated with this item. Authorized users must log in to gain access. Non-authorized users do not have access to these files.
Decentralized Learning for Wireless Video Streaming with Delayed Feedback
Abstract
We study the optimal control of multiple video streams over a wireless downlink from a base-station (BS)/access point to N end-devices. The BS sends video packets to each end-device under a joint transmission power constraint, the end-devices choose how to play out the received packets, and the collective goal is to provide a high Quality-of-Experience (QoE) to the end-users. All the end-devices send feedback about their states and actions to the BS which is assumed to reach it with a fixed deterministic delay. We analyze this team problem with delayed information sharing by casting it as a cooperative Multi-Agent Constrained Partially Observed Markov Decision Process (MA-C-POMDP). First, the original team problem is decomposed into N independent unconstrained team problems, using recent theoretical developments for MA-C-POMDPs. Thereafter, the common information approach and the formalism of approximate information states (AISs) are used to develop approximately optimal solutions. Computationally feasible data-driven implementations using neural networks are then employed to obtain such solutions. Numerical simulations using AISs demonstrate vastly improved performance when compared to a scheme without AISs. We compare performance, power costs, QoE and robustness to channel variations.
Subject
Reinforcement LearningMarkov Decision Process
Wireless Systems
Video Streaming
Constrained optimization
Collections
Citation
Arunachalam, Subrahmanyam (2023). Decentralized Learning for Wireless Video Streaming with Delayed Feedback. Master's thesis, Texas A&M University. Available electronically from https : / /hdl .handle .net /1969 .1 /202999.