Test Data Compression Method Based On Binary Dictionary Code

In the context of the rapid development of electronic technology and continuous innovation in recent years,the scale of integrated circuits has expanded,and ultra-large-scale integrated circuits have emerged as the times require.However,as the scale of integrated circuits is getting larger and larger,and its complexity and integration are also increasing,it is inevitable that various defects will appear in the design and production links,and these defects will eventually lead to the failure and problems of integrated circuits.Therefore,it is very critical to test the integrated circuits.At present,while Chinese integrated circuit testing technology is developing rapidly,there are practical problems such as too large amount of test data,too long testing time,and high technical requirements for testing.In response to this situation,domestic scholars have been pursuing the development of an efficient testing technology.Integrated circuits currently include three test technologies: Test Set Compaction,Built-in Self-Test(BIST)and Built-off-Self-Test(BOST).At the present stage,the coding compression technology in BOST is relatively common at home and abroad.The coding compression technology is a lossless compression technology,which will not affect the failure rate of the circuit,and can detect the existing problems of the circuit in time.There are four main coding compression methods: variable-fixed length coding,fixed-fixed length coding,fixed-variable length coding,and variable-variable length coding.The more common ones are:run-length coding,dictionary coding,statistical coding,and indirect coding.This thesis focuses on the study of coding compression technology.The combination of run-length coding and dictionary coding not only retains the advantages of dictionary coding no need of memory for storage and no need of hardware decompressor,but also retains the advantages of run-length coding,which can be easily implemented with good compression effect.In addition,the bisection algorithm was used to establish the dictionary,and a test data compression method based on the bisection dictionary coding was proposed.This method breaks through the limitation of dictionary coding to build a dictionary on the basis of test data,and proposes to control the length of dictionary coding with N-digit system,that is,taking decimal Y,which is divided into N equal parts in the(0,1)interval,as an intermediate parameter,and expressing it as a regular data set in the form of 0 runs,finally preparing the dictionary by the binary search method.This method can not only reduce the test time and cost,but also can restore the test data through simple calculation during decompression and find faults existing in the integrated circuit in time.Considering whether the length of the dictionary will have an impact on the compression effect,in this thesis,some binary dictionaries of 2 and non-2 power positions were established,and compression experiments were carried out one by one for the seven test circuits in the ISCAS89 standard circuit Mintest set.The experimental results showed that the 8-bit binary dictionary compression method has an average compression rate of 63.06%.Compared with the compression ratio of FDR code and Golomb code,it was found that the compression method proposed in this thesis has better effect,which further confirms the feasibility and effectiveness of this scheme.