Phase noise aware optimization of Rotary Traveling Wave Oscillator with neuro-genetic nondominated sorting algorithm

The design of Rotary Traveling Wave Oscillator (RTWO) requires the optimization of phase noise, power and frequency tuning range. RTWO design is a multi-parameter multi-objective problem which requires simultaneous optimization of performance measures. In this paper we apply neural network modeling and genetic algorithm to obtain optimal values of design parameters for low power consumption and phase noise. We then propose how the algorithm can be modified for phase noise aware optimization at the expense of power consumption. Neural network is used for estimation of power consumption and phase noise given transmission line width and spacing. The genetic algorithm is based on the popular nondominated sorting multi-objective optimization algorithm. Optimization is performed on a 5 GHz RTWO.