Distortion estimation algorithms for real-time video streaming: An application scenario

This paper investigates further on the possible application scenarios for three low complexity video distortion estimation algorithms previously proposed by the authors. Thanks to their low computational complexity, these algorithms allow a real-time estimation of the impact of transmission errors and/or packet losses on the ongoing video transmission. Thus, instead of using an “a-priori-knowledge-based” method for assigning priorities to video datagrams, an ensemble of priorities can be defined on the fly, as the video transmission proceeds, through a real-time evaluation of the end-to-end distortion of each priority class. This priority assignment algorithm can be used to get a gain in performance in many different scenarios. In a previous work of the authors, two possible video streaming scenarios were presented, which were shown to take a significant advantage from the application of the above mentioned video distortion estimation algorithms. In this paper, we recall these advantages and sketch a possible further application introducing a method for assigning priorities to video datagrams in view of designing a real-time adaptive Unequal Error Protection (UEP) of video data. We present some results about the possibility of encapsulating the distortion, estimated by the encoder, in each video packet, so that to allow a time-varying tuning of Reed Solomon Forward Error Correction (FEC) according to the priority assignment algorithm and to the crossover probability of the channel.