Genetic Algorithm for Microwave Computer-Aided Design: The State of the Art

The design of microwave integrated circuits is complex and has traditionally been done by highly experienced designers. Although several electronic design automation (EDA) tools allow for addressing some complexities of these circuits, they do not always make it possible to optimize the performance objectives of the circuit sought by the designer. The genetic algorithm (GA), a multiobjective optimization evolutionary algorithm, has been used in the design of analog components and circuits for a few decades. The algorithm is robust and efficient and surpasses classic optimization techniques based on numerical methods in many applications. Limitations of the GA include the need to predefine a circuit topology that can achieve the desired objectives and the considerable computing resources required when the algorithm is to perform circuit synthesis. Like digital design, the trend in analog design is towards more automation, which reduces the design complexity, cycle and cost and improves optimization capabilities. This survey showed that the new generation of EDA tools will be based on machine learning and multiple optimization techniques, including evolutionary algorithms such as the GA, a direction taken by several mainstream EDA suppliers.