A Dec-POMDP Model for Congestion Avoidance and Fair Allocation of Network Bandwidth in Rate-Adaptive Video Streaming

We consider the problem of distributed rate adaptation among multiple video streaming sessions over the Internet from a decision-theoretic and computational intelligence point of view, and we design a multi-objective optimization model for network resources, seeking a fair and efficient distribution of end-users´ Quality of Experience (QoE). A social welfare function is developed to capture both fairness and efficiency objectives at the same time. Then, assuming a common altruistic goal for all network users, we propose a Decentralized Partially Observable Markov Decision Process (Dec-POMDP) model for finding the optimal network bandwidth allocation that leads to social welfare maximization. We show that the resulting optimal policy for the proposed model outperforms TCP-Friendly Rate Control (TFRC) protocol in terms of total utility and fairness.