Hierarchical Packet Pre-dropping Approach for Improved MPEG-4 Video Transmission in Heavily Loaded Wireless Networks

Abstract

In this study, we develop a hierarchical packet pre-dropping (HPPD) approach that exploits different methods to process intra-coded (I)/predictive-coded (P)/bidirectionally predictive-coded (B) video frame packets to improve the video transmission quality in heavily loaded IEEE 802.11e wireless local area networks (WLANs). It is not easy to transmit packets in WLANs with a heavy traffic load. Efficient transmission of video packets is important for improving the quality of video transmission. If the network becomes congested, important packets may be dropped. Further, video packets tend to be affected considerably more than data packets. Therefore, controlling the traffic of insignificant packets to avoid the transmission of useless data can improve the quality of video transmission. On the basis of this concept, in this paper, we propose a strategy that is different from the approaches proposed in earlier studies. First, identification of the dependence of video frames in a group of pictures (GOP) can prevent the transmission of useless data. Thus, the transmission channel resources are economized by avoiding the transmission of undecodable video frames. Second, in heavily loaded WLANs, it is not sufficient to utilize the queues to transmit important video frame packets. HPPD adopts the method of an aggressive drop of unimportant packets to retain the transmission channel resources for important video packets. Experiments have demonstrated that in heavily loaded WLANs, the Peak Signal to Noise Ratio (PSNR) for the case when the HPPD mechanism is adopted is better than that when enhanced distributed channel access (EDCA), adaptive mapping, or a novel mapping mechanism is used.