Self Synchronising T-codes to Replace Huffman Codes

This paper describes recent work on the T-Codes, which are a new class of variable length codes with superlative selfsynchronizing properties. The T-Code construction algorithm i s outlined, and it is shown that in situations where codeword synchronization is important the T-Codes can be used instead of Huffman codes, giving excellent self-synchronizing properties without sacrificing coding efficiency. Background When corruption occurs in a streani of data which is coded with variable length codes, the decoder can lose track of where codeword boundaries are located in the data stream, and so. the effect of the corruption can extend over a large number of received symbols. Variable length code sets can be chosen such that the receiver is able to determine the correct location of these codeword boundaries relatively quickly after a corruption, but this is usually done by choosing codes in which certain bit sequences occur only at the end of codewords. The receiver must look for these special sequences in the received data stream. In some cases it is possible to choose codes which will self-synchronize as a result of the normal decoding process, but these are relatively difficult to find. T-Codes The original discovery of the T-Codes was published in 1984 by Titchener [l]. This work gave an algorithm for-generating families i f T-Code code sets, and showed that all code sets generated in accordance with this algorithm are self-synchronizing by nature. These properties are not derived from having specific synchronizing bit sequences, but rather the synchronizing information is spread throughout the code as an inherent part of its construction. The construction of the codes is such that when codeword synchronization is lost as a result of a corruption, the receiver will automatically resynchronize on a subsequent codeword boundary without any special measures being taken. This will happen even when the loss or corruption is not recognized as such. When a data loss or corruption is known to have occurred, an algorithm is available for the receiver to determine the point at which subsequent codeword synchronization is re-established. T-Code Construction Algorithm The T-Code construction algorithm is very simple. Code sets are constructed by augmenting lower level T-Code code sets, with the lowest level being the code set 0 and 1. The augmentation process consists of writing out a list with two copies of the lower level code