From rate-distortion analysis to resource-distortion analysis

The ultimate goal in communication system design is to control and optimize the system performance under resource constraints. As the communication paradigm evolves from the conventional desktop computing, wired, and centralized communication to current mobile, wireless, distributed, and massive communication, video encoding and transmission operate under more and more resource constraints. In traditional video communication applications, such as digital TV broadcast, the major constraint is in the form of transmission bandwidth or storage space, which determines the encoding bit rate. Rate-distortion (R-D) theories have been developed to model the relationship between the coding bit rate and signal distortion. For video communication over mobile devices, the video encoding and transmission operate under additional resource constraints, such as energy supply and on-board computation capability. Therefore, there is a need to extend the traditional R-D analysis to resource-distortion analysis by incorporating the new resource constraints into the R-D analysis framework. In distributed and massive wireless video sensor networks, the resource utilization behaviors of individual video sensors should be well-coordinated through network-level rate allocation and optimum routing so as to maximize the overall performance. In this paper, we start from the classical R-D theory developed by Shannon over 50 years ago, and then review the R-D modelling techniques for modern image and video compression systems. We study the resource-distortion analysis framework for video communication over wireless devices. As one step further, we present the research problem of resource allocation and performance optimization for video compression and communication over a network of wireless communication devices.