A construction of high rate punctured convolutional codes

We consider two methods that construct high rate punctured convolutional codes. First, we present best high rate R = (n-1)/n punctured convolutional codes, for n = 5,6, · · ·, 16, which are found by exhaustive searches. To obtain the best code, we use a regular convolutional code whose weight spectrum is equivalent to that of each punctured convolutional code. We search these regular equivalent codes for the best one. Next, we present a method that searches for good punctured convolutional codes by partial searches. This method searches the codes that are derived from R = 1/2 original codes obtained in the first method. By this method, we obtain some good punctured convolutional codes that have superior performance compared with that of the best codes.