Comparing structures of two-dimensional error correction codes

Abstract

Advances in integrated circuit production technologies have reduced device sizes, leading to corresponding scaling in electrical characteristics, such as threshold voltage. This scaling has increased the susceptibility of devices to electromagnetic radiation, raising the bitflip probability. Systems requiring a certain level of fault tolerance employ techniques like Error Correction Codes (ECC), providing a degree of reliability in mitigating this issue. The error correction and detection efficacies and ECC scalability vary based on the encoding and codestruct employed. This study employs four Hamming and parity code organizations for performing four Two-Dimensional (2D)-ECCs (N × 4p, N × ExHam, N × Ham_p, and N × Ham2_2p). We investigated the scalability, synthesis results, and correction and detection rates employing the same number of check and data bits for the four 2D-ECCs. The results point to the advantages for ECCs that employ cross-checking using radiation-hardened memories for checkbits, especially when ECCs scale to large codestructs.