ABSTRACT
To be more effective, FEC techniques can take into account the importance of the different parts of a video bitstream, thereby applying unequal error protec tion (UEP) (4,5). The video bitstream is first partitioned into classes of different sensitivities. Important information such as frame size, temporal location of a frame, spatial location indication, coding mode, and quantization information must be protected to ensure a minimum level of video reproduction quality. Er rors in these header bits will not only affect the video reproduction quality, but also can cause the decoder to reset in response to resource allocation problems. For example, if the encoder changes picture size from QCIF to CIF and this in formation is not correctly received, the decoder may have to reset to accommo date higher than expected number of blocks and display resolution. FEC coding can be applied in a cascade manner, where the same FEC code is first applied to
all header information, and then reapplied to the more important part of the same information. Since header information is usually a relatively small part of the bitstream, FEC overhead can be minimal.
