Compressive Video Streaming: Design and Rate-Energy-Distortion Analysis

Real-time encoding and error-resilient wireless transmission of multimedia content using traditional encoding techniques requires relatively high processing and transmission power, while pervasive surveillance and monitoring systems often referred to as wireless multimedia sensor networks (WMSNs) are generally composed of low-power, low-complexity devices. To bridge this gap, this article introduces and analyzes a compressive video sensing (CVS) encoder designed to reduce the required energy and computational complexity at the source node. The proposed encoder leverages the properties of compressed sensing (CS) to overcome many of the limitations of traditional encoding techniques, specifically lack of resilience to channel errors, and high computational complexity. Recognizing the inadequacy of traditional rate-distortion analysis to account for the constraints introduced by resource-limited devices, we introduce the notion of rate-energy-distortion, based on which we develop an analytical/empirical model that predicts the received video quality when the overall energy available for both encoding and transmission of each frame of a video is fixed and limited and the transmissions are affected by channel errors. The model allows comparing the received video quality, computation time, and energy consumption per frame of different wireless streaming systems, and can be used to determine the optimal allocation of encoded video rate and channel encoding rate for a given available energy budget. Based on the proposed model, we show that the CVS video encoder outperforms (in an energy constrained system) two common encoders suitable for a wireless multimedia sensor network environment; H.264/AVC intra and motion JPEG (MJPEG). Extensive results show that CVS is able to deliver video at good quality (an SSIM value of 0.8) through lossy wireless networks with lower energy consumption per frame than competing encoders.