Generalized concatenated codes for channels where unidirectional byte errors are predominant

Codes are considered for correction and detection of unidirectional byte errors. A code construction based on the generalized concatenated code construction is proposed. This construction gives a large number of efficient codes. For example, from this construction, a 72-input-bit encoder of the triple unidirectional 8-bit-byte error-correcting and fourfold unidirectional 8-bit-byte error-detecting code with a length of 112 bits and rate 9/14 is obtained, whereas the ordinary triple 8-bit-byte error-correcting and fourfold 8-bit-byte error-detecting code of the same length has only 56 information bits and is of rate 1/2. The proposed construction is generalized to one that gives efficient short-length codes