<title>Data recovery during two-dimensional readout of optical tape</title>

Abstract

Optical tape offers the greatest volumetric storage density of any commercially available storage medium, and thus is a cost effective alternative to traditional forms of archival storage. Traditional optical tape mechanisms read a 1D line of data marks. We propose a 2D read-out mechanism that can improve the data transfer rate of optical tape systems to meet future demands. However, such a mechanism relies on a CCD array for retrieving an image of data bits stored on the tape. In order to recover the binary data, some processing of the image data must be performed, which should not be detrimental to the throughput of the system. We have experimented with two data recovery techniques, one based on image processing and one on an artificial neural network. We describe an experimental set up of the read-out head in our laboratory for obtaining data that are then processed by the two data recovery methods. Our results show that the neural network performs better in almost every case. The resulting RBERs indicate that with the expected improvements in the hardware of the read-out mechanism any errors in the array of bits will be corrected by the error control process.