Multi-objective Optimization Of Mixed Polarity Reed-Muller Logic Circuits

With the rapid development of integrated circuits,the integration of integrated circuits is increasing,the speed is getting faster and faster,and the power consumption is also increasing rapidly.Integrated circuit optimization design is becoming more and more complex.At present,integrated circuit optimization design is mainly based on Boolean logic.However,a large number of studies have shown that circuits represented by Reed-Muller(RM)logic have advantages over circuits represented by traditional Boolean logic in terms of power consumption,area,speed,and testability.Fixed-Polarity RM(FPRM)expansion and Mixed-Polarity RM(MPRM)expansion are two common RM logic expansions.In terms of multi-objective optimization of RM circuits,MPRM logic circuits have a larger polarity search space and better optimization effects than FPRM logic circuits.The optimization of MPRM circuits is to search for one or more polarities in a specific space to optimize the corresponding area,power consumption and other goals.However,the polarity search space of MPRM logic circuits is huge and complex,and it takes a lot of time to perform multi-objective optimization of circuits.In view of this,this thesis will combine intelligent algorithms and multi-objective optimization problems to study the multi-objective optimization of MPRM logic circuits.This thesis mainly does the following four aspects:1.Logic synthesis for circuits and multi-objective optimization: This thesis introduces logic synthesis for integrated circuits and conducts in-depth research on multi-objective optimization problems,including the definition of multi-objective optimization problems,related concepts,tasks and goals.It provides theoretical basis for multi-objective optimization of MPRM logic circuits.2.Multi-objective optimization model of MPRM logic circuits: The thesis conducts in-depth research on XNOR/OR expansions and mixed polarity conversion method of MPRM logic circuits.The area and power consumption models of MPRM logic circuits are established.It has laid a solid foundation for multi-objective optimization of MPRM logic circuits.3.Multi-objective optimization of MPRM circuits based on the Multi-Objective Ternary Diversity PSO(MOTDPSO)algorithm: On the basis of the Ternary Diversity PSO(TDPSO)algorithm,boundary constraint processing is performed on particles that exceed the defined boundary range.And combined with the Pareto domination concept to improve algorithm,the MOTDPSO algorithm is proposed.Then combined with the area and power consumption models and mixed polarity conversion method.The MOTDPSO algorithm is applied to multi-objective optimization of MPRM circuits.The MCNC Benchmark circuit test shows that,compared with the Discrete PSO(DPSO)algorithm,the average optimization rates of circuit area and power consumption of the MOTDPSO algorithm are 8.57% and 11.12% respectively.Compared with the TDPSO algorithm,the average optimization rates of circuit area and power consumption of the MOTDPSO algorithm are 3.49% and 5.65% respectively.Compared with DPSO and TDPSO algorithms,MOTDPSO algorithm has better optimization effect and robustness.4.Multi-objective optimization of MPRM circuits based on the Pareto Dominance Ternary Diversity PSO(PDTDPSO)algorithm: Based on the MOTDPSO algorithm,a mutation operator is introduced to perturb particles,and the particle boundary processing method and the selection of the global optimal position are improved,then combined with the area and power consumption models and mixed polarity conversion method,a multi-objective optimization algorithm for MPRM circuits based on PDTDPSO is proposed.The MCNC Benchmark circuit test shows that,compared with NSGA-II,the average optimization rates of circuit area and power consumption of PDTDPSO algorithm are 11.10% and 13.71% respectively.Compared with the MOTDPSO algorithm,the average optimization rates of circuit area and power consumption of the PDTDPSO algorithm are 5.84% and 8.08% respectively.Compared with NSGA-II and MOTDPSO algorithms,PDTDPSO algorithm has better optimization effect and robustness.