Correlated noise estimation and error correction in parallel storage channels

The evolution of new technologies frequently results to the need of formulating old problems in new way. The design of MEMS-based storage systems that use multiple, simultaneously accessed channels, has to confront, among others, the problem of correlated noise in different channels. This work presents a method that exploits the inherent parallelism found in such communication/storage systems in order to estimate the correlated error vector and to improve the data decoding process. The method uses the parallelism of the multiple channels both to flag symbols as erasures using the error locations revealed by the initial errors-only decoding attempt, and to estimate the introduced channel noise vector and provide soft information to the demodulators of all channels. Numerical results demonstrate the performance improvement that is achieved by the proposed method, thus improving the reliability of such systems/devices.