Adaptive video streaming over HTTP through 3G/4G wireless networks employing dynamic on the fly bitrate analysis

The smooth video streaming over HTTP through 3G/4G wireless network is challenging as available bit rate in the internet changes due to sharing of network resources and time varying nature of wireless channels. The present popular technique Dynamic Adaptive Streaming over HTTP (DASH)provides solution up to some extent to stored-video, but the effective adaptive streaming of a live video remains a challenge in a high fluctuating bit rate environment. In this paper, an intelligent algorithm based on client server model-where client system analyses the incoming bit rate on the fly and periodically sends report to server which in turns adapts the outgoing stream as per the feed-back, is proposed. The bit rate analysis process at the receiver estimates the link data rate dynamically by comparing it with some pre-defined pattern. The proposed system was implemented and tested in real-time in CDMA lxEVDO Rev-A network using internet dongle. An improvement of 37.53% in average PSNR and 5.7% increase in mean SSIM index over traditional buffer filling algorithm was observed on a live video stream. The proposed system was also evaluated on a stored video.