一种有效的片上系统测试数据压缩算法

Study of an efficient SOC test vector compression scheme

测试数据的规模和容量直接影响了片上系统的测试成本,故提出了一种测试数据编码的压缩算法———M in Comp.该方法采用不等间距的编码方式,根据测试数据中游程长度的统计分布情况来调整各组数据的大小,从而提高测试数据的压缩率,降低了测试成本.为了使编码算法对应的解码电路的硬件开销最小化,该算法还引入了前后缀标识位的概念,这样可减小解码电路的规模和复杂度.对ISCAS89benchm ark电路的实验结果表明,采用M in Comp编码方式的压缩效率要比Golomb等编码方法好,而且实现方式简单.

Test volume is an important factor affecting the cost of the SOC Testing. In order to reduce the cost of SOC Testing, the paper proposes a method for compression/decompression of test data, that is the Min_Comp. Based on the analysis of the different sizes of the run-length coding, the method determines the grouping of test data and it can improve the compression ratio and reduce the test cost. Furthermore, by introducing the prefix and tail flag, the decoder leads to lower hardware overhead compared with those in the references. Experiments were performed on the ISCAS 89 bechmark circuits, and the results show that the Min_Comp code has a better compression ratio than the Gololnb code. Also, the decompression circuit has a low area overhead and is easy to design.