A Bandwidth Variation Pattern-Differentiated Rate Adaptation for HTTP Adaptive Streaming Over an LTE Cellular Network

Abstract

Currently, HTTP adaptive streaming (HAS) is the state-of-the-art technology for mobile video streaming. The rate adaptation of HAS has been designed to make a trade-off between two contrasting requirements, i.e., enhancing the quality of a video, and reducing the probability of video freezes, by adaptively switching between different video bitrates during a video playback session. This process becomes more challenging when moving onto an long term evolution (LTE) cellular network due to the unstable nature of the wireless channel. In this paper, we propose the bandwidth variation pattern-differentiated rate adaptation (BVPDRA) algorithm for LTE cellular networks. Unlike prior works, BVPDRA does not strike a balance between the stableness and responsiveness of bitrate switching in the case of bandwidth capacity variations. BVPDRA differentiates between bandwidth variation patterns of the LTE cellular network as either constant bandwidth fluctuations or instantaneous bandwidth hopping. Accordingly, BVPDRA operates with a dual character: for the constant bandwidth fluctuations, BVPDRA performs smoothed bandwidth prediction and conservative rate switching to minimize video quality version oscillations; for the instantaneous bandwidth hopping, BVPDRA performs positive bandwidth prediction and aggressive rate switching to maximize the bandwidth utilization and minimize the risk of playback stalling. We empirically evaluate the performance of BVPDRA on an LTE cellular network testbed. The results demonstrate that BVPDRA achieves a higher average bitrate, and lower rebuffering ratio with a reduced bitrate switching frequency.