Software Decoding of Legacy Tape Signals

There is new value in decades-old data, if it can be accessed with modern information technology tools and methods. Some of the data provide snapshots-in-time that cannot be duplicated, such as seismic tests or climate data. The Department of Energy has identified that it, and many other companies and institutions worldwide, have large amounts of data stored on obsolete, legacy, magnetic tapes, much of which is deteriorating. Commercial data recovery and migration services use vintage tape drives to read the data. These old machines have limited signal processing capabilities and the signals sensed from degraded portions of the tapes are unreadable. Modern, adaptive signal processing can be applied to these signals and recover many of the lost bits, however, this hand-crafted approach to each tape is cost-prohibitive, especially for organizations that have thousands of tapes to recover and migrate. This Phase I project addresses these issues by developing a “universal” read channel model that can be used by commercial recovery companies, or the tape repository owners, at their facility. The head/media signal output from the tape drive’s preamplifier is digitized. This waveform file is input to the channel model. The characteristics of the digitized head/media signal are automatically analyzed and the channel parameters set to optimize recovery. Degradations will be identified and specialized algorithms will be applied to maximize the likelihood of correct recovery of the data. In Phase I, the biggest risks for successfully developing this universal read channel model will be surmounted. First, a robust method of automatically characterizing the channel and optimizing model parameters will be developed. Then, a software decoder for the channel codes and error correction codes will be written. Furthermore, vintage 9-track tapes will be degraded in known ways to develop algorithms targeted to getting the best fidelity from them, enabling more tapes to be recoverable. In Phase II and beyond, these vital technology capabilities will be made more robust, accurate, faster, and cheaper. They may be migrated to hardware or parallel processing in the cloud. They will be applied to formats beyond the original target of 1960s to 1980s era 9-track tapes to include higher density cartridge tapes and later popular formats, such as R-DAT. From this foundation, the technology could be applied to optical communication channels and to a variety of sensors used in the Internet-of-Things (IoT) and edge computing.