A learning-based method to match a test pattern generator to a circuit-under-test

LFSRs are used as low-cost test pattern generators for circuits testable by pseudo-random patterns. We demonstrate that different LFSRs of the same degree, started from different initial states, may yield significantly different fault coverages and test lengths when used as test pattern generators for a given circuit, especially when the circuit is not fully testable by a practical number of pseudo-random patterns. A method to tailor an LFSR to a circuit-under-test is proposed, that attempts to select the most effective LFSR and initial state for the circuit. The method is based on a learning process that can be applied directly to certain classes of circuits. The learning process is also used to establish a collection of (primitive and non-primitive) LFSRs and initial states, effective for arbitrary circuits. Experimental results demonstrate the applicability of the proposed approach