Test wrapper algorithm for scan chain balance based on entropy increase hybrid differential evolution

Abstract

A well cores reused-based wrapper design is an important approach to minimize SOC test application time and test costs. The combinatorial optimization problem of core wrapper design has been proven to be a NP-hard problem. In this paper, a wrapper scan chain balance algorithm with entropy increase hybrid discrete differential evolution (EIHDE) is proposed to solve the core wrapper problem, which is inspired by thermodynamic system principle of entropy increase and outstanding global searching ability of Differential Evolution (DE). The proposed approach develops a cooperative mutation strategy based on entropy increase for the problem to preserving its interesting search mechanism for discrete domains. In the proposed model, two cooperative encode modes of individuals are introduced for standard differential mutation and the cooperative entropy increase mutation: integer encode mode and binary encode mode. EIHDE controls the search space by differential mutation, and search for superior individual in local space by entropy increase mutation. The combination of two kinds of mutation operations promotes the optimization ability considerably and achieves a better tradeoff between exploitation and exploration. The experimental results of the ITC'02 SOC test benchmarks show that EIHDE can achieve more balanced results compared with other algorithms.