Improvements in the capacity of computer-generated holographic storage using the Lee method with sparse multivalued reconstructions

Abstract

The use of sparse multivalued data encodings for the purpose of increasing the capacity of parallel-access optical memories based on Fourier-transform computer-generated holography is discussed. Results based on the Lee method indicate that a sparse encoding of nonbinary data words can be used to increase the storage-area utilization efficiency from 35% to > 70%. It is also found that for signal-to-noise ratios greater than 200, five-level data encoding can be used to achieve bit error rates less than 10(-12) reliably with 60% area efficiency.

<title>Storage capacity of pixelated paraxial diffractive elements</title>

“…Hence, the mutual information I(S, 5) giving the effectively transmitted and correctly received information at the channel output is determined by -I(S;S) = K(S) -K(S) . (5) In a diffractive storage system information vectors m E S2 are transmitted in one step. Consequently, the mutual information is expressed as I(S'2 ; fl2) = K(S'2) -K(S'2 fl2) .…”

“…The reason is that there exist a variety of efilcient computer-aided design methods which can be used with high flexibility to encode digital data into a data representation that can be directly recorded in some specified optical storage medium. We would also like to draw attention to related publications by Neifield [4,5], where the information capacity of pixelated diffractive Fourier elements is examined considering a certain sparse multivalued encoding of data. However, this approach is based on statistical simulation studies and does not take advantage of information-theoretical methods.…”