Research On Interval Extraction And Difference Coding Based Test Data Compression Method

The chip is the underlying support of the digital age,and it is increasingly strategic in international competition.Governments support the development of their chip industry through financial subsidies,market support and patent protection.As chip integration continues to increase,the amount of test data increases exponentially,and the time and cost required to analyze and process test data skyrockets.Coding compression is an effective method to reduce test data,which can realize lossless compression of test data and achieve the purpose of reducing test cost.This thesis is based on coding compression to start the research,starting from the coding rules of test sets and the correlation between code words.The main research results and innovation points are as follows:(1)A coding compression and decompression scheme based on interval extraction with alternating extracted coding strings and interval character strings is proposed.The encoding method firstly performs adaptive filling of irrelevant bits;secondly divides the test data according to double tours to obtain the length and type of each tour;again extracts the even bits of the divided tours to form a new test data set;and finally encodes the test data set according to the newly obtained test data set.Since only the extracted interval sequences are encoded,the encoding corresponding to the interval extraction length is shorter than the encoding corresponding to the original tours.Also the shortest length of the original division is 2,so there is no case where the interval extraction length is 0.The new encoding method adopted in this scheme does not need to read prefix and suffix sequentially,and the decompression structure is simpler and the hardware overhead of decoding circuit is smaller.(2)A reference number difference encoding method based on dividing groups is proposed.The encoding method firstly adaptively fills irrelevant bits;secondly divides the tours into several groups equally by a fixed number;again selects the representative tour length in the current group as the reference number,and encodes the original tour length after making a difference with the reference number.The encoding method of this scheme reasonably takes into account the correlation between tours,and can effectively shorten the length of the data to be compressed without changing the current number of tour sequences,and the compression efficiency is obviously improved,and it is also effective in reducing the testing cost.Theoretical analysis and experimental results show that both test compression schemes have good compression effects and can effectively deal with the conflict between the large amount of test data and the limited memory and bandwidth of automatic test equipment,and can achieve the purpose of reducing test costs.