Video compression in the neighborhood: An opportunistic approach

The proliferation of mobile devices combined with advances in the area of low-power wireless communication, such as Wi-Fi Direct and Bluetooth 4.0, gave rise to a new computation paradigm known as Device-to-Device (D2D) offloading. In this scenario, devices collaborate with each other using short wireless links to create ad-hoc P2P networks for distributed task execution. Experiments on human movement, a non-negligible factor in the D2D context, have shown that people move in group or meet frequently, which suggests that D2D is possible. In this work, we examine the case of parallel compression of smartphone recorded videos with the help of nearby devices. First, we present a mathematical formulation of the problem that optimizes the compression time on the number of nearby helping devices, and show that the problem can be mapped as a water-filling problem. Then, we present real results of the compression time and energy when the compression is performed on one device and when it is parallelized among collaborating devices. To obtain these results, we implemented an Android application that is able to detect nearby devices, connect with them using Wi-Fi Direct, send video chunks for compression, receive and merge compressed chunks into one full compressed video.