THE EFFECT OF TRELLIS TERMINATION ON THE PERFORMANCE OF TURBO CODE

Abstract

This paper introduces a new class of convolutional codes, which is called Turbo Code. Turbo Code was shown to achieve performance in terms of Bit-Error-Rate (BER), which is near Shannon limit. Turbo Code encoder is built using a parallel concatenation of two Recursive Systematic Convolutional (RSC) codes. In this paper, two solutions to the trellis termination problem are presented. The first solution encoder uses terminated Upper RSC encoder and unterminated Lower RSC encoder. On the other side, the second solution encoder uses terminated Upper and Lower RSC encoders. The performance of the two solutions is tested for different circumstances and the results are interesting.

Keywords

• Turbo Code, Iterative Decoding, Feedback Decoding, and SOVA.