Two-Phase GA Based Area and SER Trade-off Algorithm for MPRM Circuits

With the increasing susceptibility of combinational circuits to transient errors, it is essential to consider soft error rate（SER） as one crucial design constraint when optimizing mixed-polarity Reed-Muller（MPRM） circuits, and efficiently obtain reasonable trade-off solutions between area and SER. Two-phase genetic algorithm with history buffer（TPGAHB） is proposed for area and SER optimization of MPRM circuits. TPGAHB performs area and SER trade-off in two phases by using a genetic algorithm（GA） model with history buffer. Phase-one of TPGAHB obtains area minimized solution by optimizing area. Basing on the concept of Pareto optimality, phase-two of TPGAHB quantifies preference by using the area minimized solution as reference and computing efficiency factor of area and SER for MPRM circuits, and obtains reasonable trade-off solutions between area and SER by maximizing the efficiency factor. Experimental results obtained by optimizing several MCNC benchmark circuits show that, TPGAHB has better optimization capability, can obtain MPRM circuits that can provide a good trade-off between area and SER at high time efficiency, and can be applied to circuits with more than 14 inputs.