Dispersive Video Frame Importance Driven Probabilistic Packet Mapping for 802.11e Based Video Transmission

The FIFO characteristic of the access category queue (AC_VI) reserved for video in the 802.11e MAC limits its performance for high quality video transmission during traffic congestions. To enhance video transmission quality, further service differentiation among the video packets is preferred. To achieve this, this paper presents a novel cross-layer design by proposing a dispersive video frame importance (DVFI) scheme from the Application layer and a dispersive random early detection (DRED) algorithm for video packet mapping among ACs. DVFI is calculated from a single frame loss induced video quality impairment, including the effects of imperfect concealment for the lost video frame itself and its error propagation to the surrounding video frames. Whereas DRED can dynamically determine the packet mapping probabilities among ACs based on the dispersive values of packet importance. Our results show the superiority of the proposed (DVFI+DRED) framework over the existing ones, including EDCA and (I/P/B+AM), and it can reach a largest gain of 3.2 dB over EDCA, and 2.6 dB over (I/P/B+AM).