Digital storage media failure is a common occurrence, but an understanding of the mechanisms for the failure can be elusive and have far reaching implications. This project utilizes a digital laser microscope to investigate the impact of physical and environmental manipulation of digital storage media on failure rates and data recoverability. This investigation into the physical characteristics of digital media is establishing failure thresholds and their suitability for data recovery. The results of the project are being stored in a taxonomy of digital media failure characteristics, potential data recovery techniques, and microscopic image maps of the media failures and interventions. An additional aspect of the project is an extensive outreach component, which includes the K-12 and community college environments. Further impacts of this investigation include the ability to utilize physical media manipulation as a security mechanism, as well as improve digital media reliability.

Project Report

This project incorporated two distinct areas of research: 1) utilize a Digital Laser Microscope (DLM) to analyze optical storage media for data recovery potential should the media become damaged. Optical disks analyzed included CD-ROM, DVD, and Blu-Ray, and 2) utilize a Motion Analysis Microscope to determine the feasibility of introducing a physical stimulus to an optical disk as a security mechanism by temporarily impacting readability. We expected to find that the destructive techniques commonly used to destroy optical media – things like cutting a disk in half or bending the disk to break it – would produce markedly different impacts on recoverability by having different zones of destruction. The reason for the differing zone of destruction line of thinking was that the stretching of storage media by bending a disk would ruin more data than simply cutting the disk with a fairly precise instrument such as a scissors or tin snips. What we have found is that the typical physical destruction techniques employed have similar – and limited – impact on the data where the media is actually broken. In other words, nearly all data on the optical disks was intact. We also explored the notion that certain types of information written to an optical disk may be more intrusive to the media and impact the strength of the media. We could not find evidence to support this notion and found that the type of information on the disk had no impact on the size of the zone of destruction, therefore no impact on recoverability. A subsequent aspect of researching the state of data on damaged optical media was to determine recoverability potential. The DLM provided the ability to take snapshot images of regions of data on the disks and stitch the regions together. A prototype method of reading the snapshot images was developed by using a computer program to process the images based on patterns of pits and lands. The program could recover encoded data, but it should be emphasized that the prototype program was not developed to read all data types, nor were complete file header and file footer information for the various file types incorporated into the prototype program. However, we were able to reach a proof of concept that data could be recovered from optical disks by using a DLM and having the DLM images processed programmatically for data recovery may be possible. We found that the broken optical disk media often tends to separate or flake away from the plastic surface of the disk over a relatively short period of time without further stimulus. This lead to two discoveries: 1) it is far easier to read the information on the optical disk storage media without having to read through the plastic layer, and 2) we came up with the concept of simply reading the optical disk from the top by removing the disk label. A Motion Analysis Microscope was used to determine if it is possible to introduce a physical stimulus to a readable optical disk to make the disk unreadable, and then remove the stimulus and restore readability. We found that this is in fact possible and that the location of the stimulus on the disk can be varied. The implication of this is that optical disks may be able to be secured by the use of a physical stimulus.

Agency
National Science Foundation (NSF)
Institute
Division of Computer and Network Systems (CNS)
Type
Standard Grant (Standard)
Application #
1116268
Program Officer
Jeremy Epstein
Project Start
Project End
Budget Start
2011-09-01
Budget End
2014-08-31
Support Year
Fiscal Year
2011
Total Cost
$330,860
Indirect Cost
Name
Ferris State University
Department
Type
DUNS #
City
Big Rapids
State
MI
Country
United States
Zip Code
49307