An efficient test vector compression scheme using base vector extraction and Huffman coding
This thesis presents a methodology for achieving high test data compression efficiency by extracting base vectors from the pool of uncompressed test vectors combined with Huffman Coding. The base vectors are used to generate coefficient vectors, which are associated with each test vector and are used to span the whole test space. Resultant base vectors and coefficient vectors can represent the entire test space. However, a high compression ratio cannot be achieved by just replacing the initial test vectors directly by base vectors and coefficient vectors. To achieve a high compression ratio, coefficient vectors, which consist of highly repetitive sequences, are further compressed by Huffman Coding. Moreover, since each base vector contains only a single "1", it is much more efficient to store the bit position at which there is a "1" instead of the entire base vector. Experimental data is also presented in this thesis to support the effectiveness of the proposed compression scheme.