Deep reinforcement learning based quality of experience aware for multimedia video streaming

Video streaming involves the continuous delivery of video files from a server to a client, where multimedia streaming is employed for playback through an online or offline media player. Video streaming uses live broadcasts to enhance direct communication with community partners and customers. The existing methods have less video streaming quality and are unable to efficiently adapt to the dynamic conditions of the network. In this research, an adaptive bit rate (ABR) method depending on dynamic adaptive video streaming over hypertext transfer protocol or HTTP (DASH) based deep reinforcement learning (DRL) named DASH-based DRL is proposed to determine the following segment’s quality in DASH video streaming with wireless networks. The proposed algorithm significantly improves the quality of experience (QoE) performance by providing a highly stable video quality, reducing the distance factor, and enduring smooth streaming sessions. The performance of the proposed method is analyzed based on performance measures of performance improvement, QoE metrics, mean opinion score, normalized value of QoE, average of normalized value of QoE, switching quality, and rebuffering time. The suggested algorithm obtains a high average normalized QoE of 0.72, average switching quality of 0.15, and an average rebuffering time of 0.16 sec, which is comparatively superior to other algorithms like real-time streaming protocol (RTSP), HTTP live streaming (HLS) and reinforcement learning (RL).