25.12.2013 Views

SLAMorris Final Thesis After Corrections.pdf - Cranfield University

SLAMorris Final Thesis After Corrections.pdf - Cranfield University

SLAMorris Final Thesis After Corrections.pdf - Cranfield University

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

or more fragments, showing that fragmentation is likely to be encountered on<br />

NTFS systems, this is in contrast to FAT systems where almost 97% of files<br />

were found to be contiguous. Approximately 44% of the total drives in the<br />

corpus contained no fragmented files, however the total number of files found<br />

from these drives accounted for less than 1% of the files recovered from the<br />

entire corpus, suggesting the more files found on the drive the greater the<br />

chance of fragmentation occurring.<br />

Garfinkel [2007] makes an interesting statement when he discusses the<br />

probability of encountering fragmentation is related to the individual type of the<br />

file; user generated files such as word documents were found to have higher<br />

fragmentation rates, possibly due to the tendency to work on such files over a<br />

prolonged period. These results show that whilst tools such as Disk<br />

Defragmenter exist it is still likely an analyst will find fragmented user files on<br />

the disk. This suggests that the file types an analyst is likely to be interested in<br />

are those most at risk of fragmentation; this suggests that fragmented file<br />

carving may be necessary to recover relevant artefacts. Since this study was<br />

conducted, the use of computers has altered; users are now storing vast<br />

quantities of media such as music and videos on their machines. The media<br />

can take up a significant amount of space; whilst the size of storage media is<br />

also increasing, the large volume of data being stored still suggests a greater<br />

chance of encountering fragmented files. Since the thumbnail cache stores<br />

information relating to user files, it is likely to become fragmented if enough user<br />

files are added that generate thumbnail cache content. This is particularly likely<br />

if the additions are done over a prolonged period as the additional clusters<br />

required to store the new subrecords may not be available contiguously on the<br />

storage media.<br />

Fragments can be classified based on pattern recognition statistics; a prototypic<br />

histogram is used in a supervised learning environment to attempt to model file<br />

types, and research has shown this to be reasonably successful with a low error<br />

Page<br />

163

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!