Study On IPSO And Its Application On Hardware Circuit's Optimal Design

In the hardware circuits,the design of SOC is mainly based on the configuration of IP core and executing methods.Due to the varieties of IP and the conflict of IP parameter,the SOC design space is very complex.An important SOC design work is the configuring of all cores' parameters,such that the architecture is tuned for the application,i.e.,the software running on the SOC architecture,and for the power,size and performance constraints of SOC.One main task of the SOC system-level synthesis is design space exploration.The essential of the task is finding the optimal set of solutions to a multiobjective optimization problem.Due to the complexity of the parameter design problem in SOC,this thesis proposes a thought of implementing improved Improved Particle Swarm Optimization(IPSO) which is based on crowding distance and Dynamic Weighted Aggregation into the parameter design problem of SOC.IPSO is modified by storing nondominated solutions externally and selecting a nondominated solution from external archive randomly using as the global bestThe proposed algorithm introduces the Pareto dominance relationship and crowding distance of fitness to preserve population diversity, and incorporates DWA procedure to close to the best solution of every objective.Compared with ordinary MOPSOand NSGA-II,this thesis demonstrates IPSO based on crowding distanceå'ŒDynamic Weighted Aggregation gets certain superiority in solving multi-objects optimization problems.In order to implement PSO into the real application fields,we make some adjustments for our algorithm:distrete method for particle's speed vector;coding method for particle swarm;selection plan for the leaders;elimination of unrational configuration.The application of fitness crowding distance truncation enables the parameter configuration keep a certain diversity in object space.Because of this,the improved IPSO implemented in parameter optimization algorithm has the ability to quickly find the best parameter configuration in SOC design space.This thesis uses the m script in Matlab to construct the mathematic model of IPSO. The IPSO is written in C++ language,integrated with highly parameterized SOC design platform developed by California University.This thesis selects three standards of measuring the performance of the multi-objects optimization problems,compares the parameter depended searching strategy,multi-objects GA searching strategy and improved PSO searching strategy in following three instances:1)image,use bitmap for memory copy;2)key,reverse the color of a bitmap;3)matrix,transpose a 10×10 integer matrix.By comparing the accuracy and performance of the simulation results of the three methods,we use the data to quantificationally prove that IPSO performs better than other related algorithms and use Pareto distribution in object space to qualitatively prove that IPSO has the ability to solve multi-object optimization problem.Simulation results demonstrates that the SOC structured parameter configuration optimization method adopted in this thesis has good discrete and non-dominated attributes in planar space composed of power consuming and running time.This method greatly shortens the search time and the accuracy and search performance are much better than parameter related methods and GA.