IBM Journal of Research and Development · 1970 · 26 citations · 2 references
EngineeringComputer EngineeringComputer ScienceFault ToleranceFault-tolerant ComputingFault-tolerant SystemsError Control CodingError CorrectionError DetectionError-correcting CodesMagnetic Recording MaterialsBinary CodesBurst Error CorrectionBurst-error-correcting CodesErasure CorrectionError Correction CodesError Correction In RecordsError Detection And CorrectionParity Bits
The 800‑bit‑per‑inch magnetic tape units of IBM System/360 can correct unlimited‑length error bursts in any one of their nine tape tracks. The technique uses a modified cyclic code with a parity bit per nine‑bit character, nine check bits forming a record‑end check character, and requires rereading a record with detected errors; it works during both read‑forward and read‑backward operations on records of any length. Errors involving more than one track within the same record are detected but are not correctable.
The 800-bit-per-inch magnetic tape units that are components of IBM System/360 can correct error bursts of unlimited length in any one of their nine tape tracks. The correction technique employs a modified cyclic code in conjunction with a parity bit in each nine-bit character. There are nine check bits in the modified cyclic code and these form a check character at the end of every record. To perform the correction, a record in which an error has been detected must be reread. Errors involving more than one track within the same record are detected but are not correctable. This error correction technique operates during both read-forward and read-backward operations, and on records of any length.
2
R. G. Gallager, W. W. Peterson · Mathematics of Computation · 1962 · 2.1K citations
Cyclic Codes for Error Detection
W. W. Peterson, Daniel G. Brown · Proceedings of the IRE · 1961 · 606 citations