Effects of Puncturing Patterns on Punctured Convolutional Codes

Abstract

Punctured convolutional codes are known to have low complexity compared to their non-punctured counterpart, while retaining a good performance. Analyzing the performance of punctured convolutional code can be simplified by using non-punctured equivalent code. In this paper new punctured convolutional codes with rates of 3/8, 3/7 and 3/6 are proposed, and their performances are studied by first constructing non-punctured equivalent codes. Simulations results show that different puncturing patterns will affect the code performances. Further investigations show that puncturing adjacent bits is to be avoided as it tends to degrade the code performance, as indicated by a decrease of the free distance by 9% and 33% below average for code rates 3/7 and 3/6 respectively. On the contrary, dispersed punctured bits will yield good code performance as indicated in the increase of the free distance by 27% and 32.45% above average for code rates 3/7 and 3/6 respectively.