Video Content Redundancy Elimination Based on the Convergence of Computing, Communication and Cache

The tsunami of video services brings huge pressure on communication systems, especially on some bandwidth-limited scenarios like wireless networks. Represented by H.264, current video compression approaches can save the transmission bandwidth by reducing the intra-and inter-frame redundancy. However, the huge amount of video content redundancy (VCR) in the transmission is still existing. In this paper, we propose a novel Content-Slimming System (CSS) framework based on the convergence of Computing, Communication and Cache to avoid the transmission of VCRs. The main idea of CSS is to detect VCRs, generate VCR models and clip them from the original frames by Computing, then transmit the necessary video content and semantic difference description by Communication, finally reconstruct the full video based on the received video content, semantic difference description and VCR models stored by Cache. Moreover, we also investigate the Video Monitoring application based on the CSS framework (CSS-VM) in which the background of monitoring frames is modeled, detected and sheared to reduce the transmission bandwidth consumption. The simulation results show that the bandwidth consumption of CSS-VM can be reduced at least by half compared to H.264, while the video quality and visual experience of CSS-VM are even better.