Evolving cellular automata for self-testing hardware

Abstract

Testing is a key issue in the design and production of digital circuits: the adoption of BIST (Built-In Self-Test) techniques is increasingly popular, but requires efficient algorithms for the generation of the logic which generates the test vectors applied to the Unit Under Test. This paper addresses the issue of identifying a Cellular Automaton able to generate input patterns to detect stuck-at faults inside a Finite State Machine (FSM) circuit. Previous results already proposed a solution based on a Genetic Algorithm which directly identifies a Cellular Automaton able to reach good Fault Coverage of the stuck-at faults. However, such method requires 2-bit cells in the Cellular Automaton, thus resulting in a high area overhead. This paper presents a new solution, with an area occupation limited to 1 bit per cell; the improved results are possible due to the adoption of a new optimization algorithm, the Selfish Gene algorithm. Experimental results are provided, which show that in most of the standard benchmark circuits the Cellular Automaton selected by the Selfish Gene algorithm is able to reach a Fault Coverage higher that what can be obtained with current engineering practice with comparable area occupation.