PRATM - Derived PRASAN Error Concealment Model for Efficient Transmission of MPEG Video Streams Over ATM Networks

Abstract

Efficient recovery of video sequences from compressed video, transmitted over data networks is one of the biggest challenges to researchers in the field. This paper addresses this problem in context of asynchronous transfer mode networks (ATM) and proposes a model for error free recovery of damaged/lost data. In ATM networks channel errors or congestion cause data to be dropped, which results in the loss of entire macro blocks of MPEG video being transmitted. The proposed model PRATM employs the bit mapping procedure defined in PRASAN error concealment model and applies the mapping algorithm to the intra frames of GOPs for providing enhanced compression by generating correlation polynomials. The major contribution of this paper is the inclusion of additional header block, containing the correlation information derived from the principle of bit level mapping for the payload data of high priority ATM packets carrying the video information. The high priority data, taken two (termed inverse pair) at a time are subjected to bit level mapping and the obtained correlation polynomial information is provided as the header in both the considered data. This is a measure taken to cope with extreme congestion condition when even a high priority packet could be lost in the channel