Channel exploration for wireless media streaming with handoff and rebuffering control

In wireless media streaming from multiple access points, handoff control ensures optimal connectivity as the wireless environment fluctuates, while rebuffering control seeks to maximize user experience by avoiding playout jitters and prolonged freeze times. A joint policy aims to capture the interactions between these controls, but requires a fluctuating wireless environment to be estimated frequently, a process which can interrupt packet retrieval in many mobile applications. In this paper we develop a model for maximizing user QoE by optimally scanning wireless channels while making handoff and rebuffering decisions. We design a control scheme that utilizes partial knowledge of the wireless environment to decide when to explore alternative connections, when to handover to them, and when to exploit an access point while employing channel-aware rebuffering control. We develop a model for the general case of multiple access points, then computationally evaluate the system in specific relevant cases. We find that the resulting optimal control takes the form of threshold-based policies, and that it compares favorably to the omniscient policy with full knowledge of all channel states. We also find that the optimal decision regions can be greatly simplified with only a small decrease in optimality.