NEW DOUBLE-BYTE ERROR-CORRECTING CODES FOR MEMORY-SYSTEMS

Double-byte error-correcting codes over GF(q) were constructed by Dume r, which have the parameters n = g(m-1), r less than or equal to 2m [m-1/3], m = 2, 3, ..., when q is even, and have me parameters n = q(m ), r less than or equal to 2m + [m/3] + 1, m = 2, 3, ..., when q is od d, respectively. In this paper, we construct a class of double-byte er ror-correcting codes over GF(2(i)), which have the following parameter s: n = q(m), r less than or equal to 2m + [m/3] + 1, m = 3, 4, .... So our constructions reduce the code redundancy of Dumer by one symbol, and we eliminate the disparity in code redundancies obtained for even and odd q. A decoding procedure for our codes is also considered.