Multiple undirectional byte error-correcting codes

Codes correcting/detecting unidirectional byte errors are investigated. It is shown that such codes have the capability of correcting combinations of bidirectional and unidirectional byte errors and code constructions. These are derived from the combination of two codes. One is a code for encoding the Hamming weights of data bytes and is used to estimate error locations. The other is a byte-error-correcting code for error evaluation. Moreover, the authors describe a decoding procedure for the codes, correcting combinations of bidirectional and unidirectional byte errors. The constructions provide many efficient codes of short length. For example, when a byte consists of 32 bits, a systematic code can be constructed with 61 information bytes and 3 check bytes that has the capability of correcting a single bidirectional byte error for double unidirectional byte errors, whereas the best known double-byte-error-correcting code requires 4 check bytes.