On the efficient search of punctured convolutional codes with simulated annealing algorithm

Punctured convolution codes (PCCs) have a lot of applications in modern communication system. The efficient way to search for best PCCs with longer constraint lengths is desired since the complexity of exhaustive search becomes unacceptable. An efficient search method to find PCCs is proposed and simulated. At first, PCCs' searching problem is turned into an optimization problem through analysis of PCCs' judging criteria, and the inefficiency to use pattern search (PS) for many local optimums is pointed out. The simulated annealing (SA) is adapted to the non-convex optimization problem to find best PCCs with low complexity. Simulation indicates that SA performs very well both in complexity and success ratio, and PCCs with memories varying from 9 to 12 and rates varying from 2/3 to 4/5 searched by SA are presented.