Multi Objective Optimization Of Ternary FPRM Circuits

With the rapid development of VLSI(Very Large-Scale Integration Circuit)technology,the chip's integration and clock frequency increase sharply,which leads to the increasing power consumption of the chip.The increasing power consumption not only brings about the problem of chip power supply,but also accelerates the aging of the circuit and shortens the life of the chip.Therefore,it is necessary to consider the optimal relationship among circuit area,power consumption and delay in the process of chip design.Compared with the traditional binary circuit,the multi-valued logic circuit increases the information carrying capacity of single line,which not only improves the information density of digital circuit,reduces the circuit area,but also reduces the circuit lead-out line and reduces the complexity of the interconnection within the circuit.Among them,the ternary logic has the smallest cardinality and is easy to implement,and it is representative in the multi-valued logic system.Boolean logic and Reed-Muller(RM)logic are two main forms of ternary logic functions.Compared with traditional Boolean logic circuits,RM-based circuits(such as arithmetic logic circuits,communication circuits,etc.)have more obvious advantages in terms of road area,power consumption and speed.The ternary Fixed Polarity Reed-Muller(FPRM)expression is a common logical expression in the ternary RM logic function.The ternary FPRM expressions of n variables have 3n polarities.Different polarities correspond to different ternary FPRM expressions,and the corresponding circuit area,power consumption and delay are also different.In view of this,this paper first establishes the area,power consumption and delay estimation model of ternary FPRM circuit,and then uses multi-objective swarm intelligence algorithm to search for the optimal polarity solution set of circuit,so as to achieve the goal of integrated optimization of circuit area,power consumption and delay.The research contents include the following four parts:1.Improvement of mathematical model of ternary FPRM circuit polarity optimization: By studying the logic expression of ternary FPRM,the polarity conversion algorithm of ternary FPRM circuit based on list technology and the area,power consumption and delay estimation model of ternary FPRM circuit are proposed respectively.On this basis,combined with exhaustive method,ternary FPRM circuit is proposed separately.Delay optimization and power optimization.2.Multi-objective Discrete Competitive Particle Swarm Optimization(MODCPSO)algorithm based on competitive behavior mechanism for integrated optimization of area and delay of ternary FPRM circuit: By studying Multi-objective Discrete Particle Swarm Optimization(MODPSO)algorithm,introducing competitive behavior mechanism and mutation mechanism,MODCPSO algorithm is established to ternary FPRM circuit.Based on the mapping of area and delay optimization and the area and delay estimation model of ternary FPRM circuit,an area and delay optimization scheme of ternary FPRM circuit based on MODCPSO algorithm is proposed.3.Comprehensive optimization of area and power consumption of ternary FPRM circuits based on Differential Non-dominated Sort Genetic Algorithm II(DNSGA-II)algorithm: By introducing differential evolution strategy into Non-dominated Sort Genetic Algorithm II(NSGA-II)algorithm,a DNSGA-II algorithm with stronger search ability is proposed,and the area and power consumption of ternary FPRM circuits from DNSGA-II algorithm are optimized.Combining with the area and power estimation model of ternary FPRM circuit,the optimal polarity search scheme of DNSGA-II algorithm is proposed.4.Comprehensive optimization of area,power consumption and delay of ternary FPRM circuit based on Multi-Objective Teaching-Learning-Based Optimization(MOTLBO)algorithm: By studying the logic expression of ternary FPRM,a delay and power decomposition algorithm for ternary FPRM circuit is designed,and combined with MOTLBO algorithm and ternary FPRM circuit.Value FPRM circuit polarity transformation technology,and proposes the best polarity search scheme of MOTLBO algorithm.The optimization schemes proposed in this paper are implemented by C language through VC6.0 compilation,and simulated by MCNC benchmark circuit.The results show that the optimization effect of the proposed schemes is obvious.