Hardware Software Partitioning Using Four Levels Hybrid Algorithm Technique
Zaini Abdul Halim, B. Surendra Babu, Mustamin Mustaffa · 2020
In embedded system design, system module is implemented either on SOC hardware or software. Limitation of hardware is cost whereas the limitation of software is response time. Hardware software partitioning is a concept that is used to choose either the system module should be implemented in hardware or software. Genetic Algorithm (GA) and Particle Swamp Optimization (PSO) are widely used techniques in hardware software partitioning. Hybrid of these techniques may give a better solution in order to meet objectives which are cost reduction and faster time response. In this paper, four techniques are used which are GA, GA-PSO, GA-PSO-PSO and GA-PSO-PSO-PSO. These techniques are developed using MATLAB. Analysis are made based on the response time to execute the algorithm and minimum cost that are required to implement the system module. Results show that the GA-PSO-PSO-PSO technique gives the minimum cost with minimum number of iterations. The improvement is around 6.5% compared to GA algorithm.