Research On The Principal Component Generation Method In Decomposing Compression Technology

With the development of integrated circuits,the size of transistors is getting smaller and smaller,chip developers can use more and more transistors on the chip,and the chip design size is getting smaller and smaller,and the probability of defects in the manufacturing process is increasing.As the number of transistors increases,the amount of test data also increases,and the increase in the amount of test data inevitably leads to an increase in hardware storage costs and an increase in test application time.Therefore,the amount of test data and test application time should be effectively reduced..Based on the decomposing compression technique,this paper studies how to generatio the principal component set,which mainly has the following two aspects:(1)A method of generating a principal component based on test set vector is proposed.Using the correlation between test set vector,first find the required compatible classes through the greedy algorithm and the relational matrix method,then select the base vector from the compatible class,then the base vector is XORed to obtain a matrix.This matrix and inverted matrix form a new matrix.The new matrix is compared with each column in test set.The column closest to test set in matrix is selected as a principal component to obtain a principal component set,and finally principal component is XORed with test set,and a residue set is obtained and encoded and compressed.Compared with the hadamard transform,the number of 1s is reduced in residue set,because vectors generating principal component is related to test set and column vectors can replace some vectors in test set.Experimental results show that the compression ratio of principal component method generated by test set's own vector is as high as 77.94%,and the average compression rate is 73.83%.The average compression ratio of this method is 14.41% higher than the direct compression of the test set,and compared with hadamard transformation,the compression rate has an average increase of0.08%.(2)A method of generating a principal component based on hadamard transformation and test set self vector is proposed.The basic idea of this method is to use hadamard matrix and the vector of test set itself to compose a new matrix,so as to better match test set,and the number of 1s is reduced in residue set.Firstly,according to data set's size,the order of the hadamard matrix required by the hadamard transformation is selected,then X is added in the end of bit stream to make data set's size the same as hadamard matrix's size,and then the hadamard matrix and its vector generated matrix combined into a new matrix,and then theprincipal component is extracted according to the principal component generation algorithm.Finally,principal component is XORed with data set to obtain a residual component and compressed.The experiment show that compression rate of the method of combining data set itself and hadamard transform to generate the principal component is up to 77.99%,the average compression rate is 73.92%,and compared with hadamard transformation,the compression rate has an average increase of 14.83%,the average compression rate is 0.17%higher than direct compression.