This research investigates a test data compression method which fully utilizes the unique characteristics of test data. The key idea is to perform the Burrows-Wheeler (BW) transformation on individual test sequences, and then to apply two different compression schemes. The BW transformation reduces the number of transitions for test sequences, which leads to a higher compression ratio compared with the original test sequences. Research topics are: 1. study characteristics of test data that form a basis for the work and to measure redundancy of test sequences along pins and across pins; 2. determine how best to merge correlated test sequences to achieve a higher compression ratio; 3. study the BW transformation and the strings generated by successive applications of the BW transformation to enable one to construct a test set with a minimal number of transitions; 4. integrate the research findings into the P.I.'s existing tool for test data compression.
