利用整数存储无理数的测试数据编码压缩方法

Test Data Compression Coding-based Scheme Storing Integers Represented for Irrational Numbers

针对集成电路测试过程中自动测试设备需要传输大量测试数据到被测芯片,浪费了大量的测试数据传输时间,不能降低芯片测试成本的情况,提出一种整数存储无理数的测试数据编码压缩方法.首先将测试数据按游程长度划分,默认第1个游程长度为小数的个位,其他游程长度依次为小数的小数位,将测试数据转换成小数;然后提出用二分查找无理数的方法,将该小数转化成可以整数表示的无理数;最后存储无理数对应的整数表示m,l,k.该方法采取传输测试数据规律而不是测试数据本身的方法,理论上可以将整个测试集的存储转化成对单个或若干个无理数对应整数表示的存储.对部分ISCAS89标准电路中规模较大的时序电路进行实验,结果表明,在同样实验环境下,其压缩效果方面优于Golomb码、FDR码、EFDR码、MFVRCVB码等成熟的编码方法.

A compression method for test data coding is presented, by storing irrational numbers in integer form against large-amount time waste of transmission of test data and high cost of chip testing due to large-amount test data transmitted to chips for test via automatic testing device in integrated circuit test in procedures as follows. Firstly, dividing test data according to run lengths and setting the first run length as the default digit before the decimal point, other lengths as decimal part respectively, before transforming test data into its decimal form. Secondly, transforming the decimal into an irrational number that can be expressed in integer form via binary search of irrational numbers. Thirdly, storing integers corresponding to their irrational form for expressing m, l and k. The method is theoretically available to transform the storage of whole test unit into that expressed in integer form corresponding to single or several irrational numbers because it transmits the rule of test data but not the test data itself. The experimental results on larger sequential circuits in ISCAS89 benchmark circuits show that, the test data can be compressed with higher efficiency than that compressed by commonly-applied coding such as Golomb, FDR, EFDR and MFVRCVB under the same experimental circumstances.