Research On The Efficient And Low Complexity Algorithm Of Turbo Codes

Turbo codes proposed by C.Berrou in 1993 have received considerable publicity for its superior performance. The code is close to Shannon's theoretical limit only a difference of 0.5dB in the bit error rate of 10^-5, which has been known as the focus of the coding theory in recent years and is widely used in the 3G communication systems. The recent research has focused on decoding algorithm and interleaver and so on. Turbo codes decoding algorithm can be divided into two types. One is MAP algorithm and the other is VA algorithm. The former is an optimal decoding algorithm and Log-MAP algorithm is the logarithm domains form of MAP algorithm, which is the center of this paper.This paper studies in Turbo code decoding algorithm and spline interpolation function is proposed to calculate the correction function of Log-MAP algorithm and a number of piecewise linear is used to replace the Jacobian formula in combination with the scale factor to further optimize the extrinsic information between the decoder so as to improve the decoding performance. The former proposes the interpolation function to calculate the correction function, which solves the problem of larger computational complexity of the correction function and eliminates the exponent and logarithm computations in the decoding. The latter studies of using a different number of piecewise linear formula to replace the Jacobian formula and makes the analysis of errors and performance. We also study different scaling factors which impacts on the decoding performance and acquire the best scaling factor values which make Max-Log-MAP algorithm better performance. Simulation results show that two methods make the Turbo decoding performance equivalent to Log-MAP algorithm and computing complexity and decoding time both are reduced significantly.According to the design requirements of the subject, this paper presents two kinds of improved scheme. Based on these two programs, the effective low-complexity Turbo decoding algorithms are achieved. Last part of the paper raises an issue for further study of the prospects.