Selecting a Common Video Compression Algorithm for Use in Error-Resilient Wireless Video

MPEG-4 is designed for multimedia and web compression, using object-based compression. Abstract This paper contains an analysis of the four most common video compression algorithms and their errorresilience capabilities when applied to wireless video compression. The algorithms examined are: MPEG-2, H.263, MJPEG-2000, and MPEG-4. Each of these compression methods has specific strengths and capabilities. However, their application in the wireless environment stretches beyond their original design criteria. Recent papers have covered specific compression techniques (e.g., Trellis decoding, soft decoding) or a general overview of error resilient video coding [3]. This paper focuses instead on the overall error resiliency tools of the most common video compression methods, and compares their effectiveness in optimizing video transmission reliability over a wireless network. The primary characteristics that affect selection of an appropriate algorithm include: 1) In compressed video, the largest number of bits from an image appears where the most complex amount of video compression is taking place. Typically, this is the part of the video where a lot of motion is taking place. Statistically, therefore, this is the part of the video where the most errors would occur. 2.0 Categories of error resilience functions 2) Channel errors in the bitstream typically occur due to interference and therefore happen statistically in groups. These categories were selected from the overall tools and optional modes available in common compression methods, specifically because their application in the compression method will have a definitive impact (positive or negative) on the error resiliency of wireless video. For example, synchronization was included because in bursty, error-prone environments, once synchronization has been lost, quickly re-synchronizing data transmission is necessary in order to begin reacquiring data. However, quantization was not considered because although it has an effect on error resilience, application of quantization levels contain an unstable tradeoff between bandwidth and resiliency: higher quantization levels means that fewer bits will be used to describe a fixed amount of data. Therefore, there is less data sent, and less time for it to be vulnerable to errors. However, the data that is sent is much more compressed and therefore more susceptible to error. 3) Video transmission is a time-dependent process. More complex amounts of error control reduce the amount of bandwidth available to send the actual bitstream. After thorough consideration and evaluation, MPEG4 was determined to have the most advanced capacity for error-resilient, wireless video transmission. Its major features, as well as minor capacities, apply extremely well to wireless technology error resilience, although not necessarily to wireless bandwidth and portable device capacity requirements.