next up previous contents index
Next: Error Concealment for Speech Up: Multimedia Transport Protocols Previous: Challenges of Multimedia Transmission   Contents   Index


Working Around Loss

To protect against packet drop (channel error) there are two types for coding error resiliency: Forward Error Correction (FEC) and Automatic Repeat reQuest (ARQ). FEC is used for error correction at data link layer. The goal is to localize and to correct errors (packet drops) by adding redundancy codes to the original data. However, it cannot handle burst error. In addition, encoding delay is very large as the encoder should process block of packets before start sending them. Some existing FEC coding schemes are Reed Solomon, Turbo Codes, Hamming Codes, etc. On the other hand, ARQ is done by requesting the retransmission of the data (dropped packets) when an incorrect packet is received. This can be detected by negative acknowledgment (NACK) after a timer expiration, receiving packets out of order, or by implicit acknowledgment (ACK). It maybe suitable for one way real time video applications such as VoD. However, it is not very effective for interactive services such as video conferencing, due to the too long delay that can be required to retransmit the lost packets. There is also a hybrid channel coding between FEC and ARQ. For a survey of packet loss protection techniques, namely ARQ and FEC for video transmission, see [160]. Unlike FEC and ARQ, which try to protect against error occurrence or assure the transmission of data correctly, error concealment techniques' goal is to mask or to minimize the effect of error after its occurrence. There are many error concealment techniques. However, due to the different nature of speech or audio and video signals, there are specific techniques for each media type.

Subsections
next up previous contents index
Next: Error Concealment for Speech Up: Multimedia Transport Protocols Previous: Challenges of Multimedia Transmission   Contents   Index
Samir Mohamed 2003-01-08