Research On Multi-objective Optimization Of Analog Integrated Circuits Based On Adaptive Weighted-sum Method

As the improvement of integrated circuit design and manufacturing,IC has developed to the stage of system on chip.Due to the particularity binary data,the automation of digital integrated circuit has achieved high level.When the designers finish describing the circuits in HDL,then the software tools will synthesize the circuits automatically,so the design scale and efficiency of digital integrated is far ahead in analog integrated circuits.Currently,analog integrated circuits designed by designers manually from front to end,the designers have to cal]the circuit simulation software continuously to check the performance of the circuits and adjust the circuits sizing until the circuits meet the requirements.This method of design is inefficiency,so it is necessary to achieve the automatic design of analog integrated circuits especially when there is less time to develop products.In this paper a new algorithm in multi-objective optimization for analog integrated circuits is proposed.It is adaptive weighted-sum which is based on weighted-sum approach.In traditional weighted-sum multi-objective optimization the algorithm finishes the work by transforming multi-objective functions to single-objective function with weight coefficient.This approach is simple to understand and easy to implement.Also the weight itself reflects the relative importance among the objective functions under consideration.But the traditional weighted-sum approach does not always result in an even distribution of solutions on the Pareto front and cannot find solutions on non-convex parts of the Pareto front although such non-dominated solutions do often exist.Even reducing the step size of weights does not solve this problem.The adaptive weighted-sum method is based on the weighted-sum approach.In this approach,the weights are not predetermined,but they evolve according to the nature of the Pareto front from the problem.Starting from a large weight step size,a coarse representation of the solution is generated and regions where more refinement is needed are identified.The specific regions are then designated as a feasible region for sub-optimization by imposing inequality constraints in the objective space.The typical weighted-sum multi-objective optimization is performed in these regions.When all the regions of the Pareto front reach a pre-specified resolution,the algorithm terminates.This method not only obtained solutions in the concave region but also even distribution.Two optimization examples by adaptive weighted-sum method for analog CMOS integrated circuits are given in this paper and the results indicate the algorithm proposed is valid.