Research Of Performance Estimation Of Channel Coding

Channel coding is always an important and hot topic in wireless communications area. From block code to convolutional code and then Turbo and LDPC, channel code is becoming more and more approximate to random coding, and the performance of channel coding is close to channel capacity in Shannon theorem. However, since the increase of encode complexity, maximum likelihood (ML) decoding whose decode complexity increase exponentially when code length increases is becoming almost impossible. Accordingly, lots of advanced performance estimation arithmetic of channel coding, including performance bound and important sampling method, gradually appear. However, low estimation precision and huge computation complexity cannot please scientists. Hence, how to estimate the error rate of channel code efficiently and then measure the performance of channel coding has been the topic that scientists and researches focus.Firstly, in-depth study of the channel coding weight distribution, further analysis by the distance characteristics of the channel coding error rate. In particular, block codes, convolutional codes, Turbo codes and LDPC codes channel coding theory of weight distribution are analyzed and compared. Clarify the impact of low weight distribution channel coding error rate. Further, with the large distribution of said channel coding to draw weight distribution, which is estimated by curve fitting a probability distribution.This paper then analyzes estimation of channel coding error rate from the perspective of performance bound. First, a detailed analysis of the principles of union bound appears in this paper. Proposed the concept of group-wise error probability on the basis of the pair-wise error probability in union bound, and further propose an algorithm to improve union bound. In order to get more tight performance bound, this paper proposes a new performance bound:intersection bound which is inspired by set theory. A proof is provided by this paper to demonstrate that intersection bound is a strict upper bound. Moreover, union bound and intersection bound are compared through simulation, indicating that intersection bound is more accurate than union bound.Another contribution of this paper is the analysis of decision region, which could be applied in performance analysis of channel coding. As a key parameter in decision region, radius of decision region could be used to estimate word error rate after decode. This paper proposes a new method which relies on code word set to get accurate radius of decision region. In order to reduce computation complexity, low weight code word set which is used to replace complete code word set is taken into consideration, indicating the new method is feasible.