A Performance comparision of polar codes with convolutional turbo codes

Abstract

Polar codes introduced recently by Arıkan are the first low-complexity codes achieving symmetric capacity for arbitrary binary-input discrete memoryless channels (B-DMCs). Although being theoretically significant, their practical significance is an issue that has not yet been fully explored. Previous studies have compared polar codes with Reed-Muller codes, where it was found that polar codes can outperform them. In this thesis, to investigate how polar codes perform against state-of-the-art forward error correction (FEC) codes used in practice, we implement a IEEE 802.16 based link-level Worldwide Interoperability for Microwave Access (WiMAX) simulator which incorporates several WiMAX FEC options, and polar codes. IEEE 802.16 standards family define standards for current and next generation broadband wireless access, which will make high data rate multimedia applications in mobile environments a reality. Next generation broadband access standard, pursued by the IEEE 802.16 Task Group m is a work in progress, and requires even more sophisticated error correction schemes so that higher throughput, better QOS, higher mobilities, wider ranges and lower latencies are supported. We perform performance comparison simulations with the convolutional turbo codes (CTC) configurations defined in IEEE 802.16e to see how much of a performance gap exists between polar codes and CTCs. The main findings of the thesis are that, although the polar codes achieve capacity for specific conditions, as expected, for the code lengths and channel conditions we have simulated, the performance of them cannot compete with that of the CTCs with equivalent rates and lengths. It remains a task to see whether polar codes can achieve similar performances with CTCs when used as component codes in other configurations and aid in the advancement of new communication technologies.