Abstract
The neuro-interpolator (NI) integrating an interpolator with a neural network equalizer in holographic data storage is proposed to reduce the influence of interpixel interference due to the aperture restriction which limits the effective recording area on the holographic medium. The simplification of NI is performed by using a hybrid genetic algorithm based on a new mutation method introducing a pruning probability, and its bit error rate (BER) performance is studied by computer simulation. The results show that the BER performance of simplified NI is superior to that of the interpolator alone. Furthermore, the number of connections in the simplified NI is reduced to about half of that obtained by using a conventional mutation method. In addition, the proposed mutation method shows the slightly-improved performance over the conventional one.
