A Least Square Method for Parameter Estimation of RSC Sub-Codes of Turbo Codes

A new algorithm is developed for parameter estimation of the recursive systematic convolutional sub-codes of turbo codes. It is based on a least square cost function of the encoder coefficients. A simple iterative process assisted by a try-again scheme is designed for the optimization of the cost function. Simulation results show that, compared with the existing method, the new algorithm significantly improves the performance while the complexity is quite reasonable.