Strategies for adaptive smoothing and rebuffering via dynamic network calculus

This paper proposes an adaptive video smoothing and rebuffering algorithm for video transmission in best-effort, low-bandwidth networks. Our algorithm, which is derived from a “dynamic” extension of the network calculus, (1) detects whether playback disruptions are inevitable and, if so, schedules a rebuffering action with minimal rebuffering delay and (2) computes a conservative transmission schedule that ensures continuous playback while minimizing consumption of network available bandwidth. NS-2 simulation results show that, compared to an aggressive transmission strategy, our method can greatly reduce bandwidth consumption and, compared to a simple rebuffering strategy with the same bandwidth consumption, our method can effectively reduce total playback delay while also reducing the number of rebuffering events.