Font Size: a A A

A Coding Compression Based On Hadamard Transform And Technique Of Main Component Enhancement

Posted on:2015-11-10Degree:MasterType:Thesis
Country:ChinaCandidate:Y L HeFull Text:PDF
GTID:2428330488999699Subject:Computer technology
Abstract/Summary:PDF Full Text Request
Since the integrated circuit(IC)feature sizes entered into the nanometer scale,the number of transistors within a chip has reached billions,and the operating frequency has run to GHz.The changes lead to a huge amount of test data volume for IC testing.The enormous test data volume can burden the capability of the tester,in addition,it will result in the waste of the test time as well as the test power.Optimize compression methods,such as test data compression,can help to reduce the amount of the test data and save the power consumption and the test time.The test data compression has been a hot topic in the field of IC testing currently.This paper proposes a coding compression method based on Hadamard Transform and a further optimization of the method to enhance the main component by using KM algorithm.Experiments on ISCAS'89 benchmark circuits show that the method has achieved remarkable results in improving the test data compression rate.The main innovations of this paper are as follows.A coding compression method is proposed based on Hadamard transform.In this method,the main spectral component of the test set is obtained by using Hadamard transform.Then the test set is split into two parts by XOR:the main component and the residuals.The main component can be generated by the simple hardware within the chip,so only the residuals need to store into the tester.During the way,it isn't compressed with the test set,but the residuals.FDR coding is highly effective because of the less number of 1 in the residuals.The experiment results on the reference circuits ISCAS'89 show that the compression rate of FDR coding has achieved 75.41%,and improve 18.19%.An enhancement method on the main component is proposed in this paper.Based on the test set and its main component matrix,a bipartite graph and a weight matrix are constructed.Then the enhancement on the main component is transferred into the optimal bipartite graph matching,and solved using KM algorithm.After the order adjustment among the test set based on the matching,the relevance between the main component and the test set has been increased.So the main component is well enhanced.Finally the experiments on the ISCAS'89 benchmark circuits show that this method enhances the main component 13.42%,and the number of 1 in the residuals decreases 20.54%,and the average compression ratio of FDR coding reaches 78.52,and improves 3.11%.
Keywords/Search Tags:Hadamard transform, significant components, bipartite graph match, KM algorithm, correlation
PDF Full Text Request
Related items