Research On The Turbo Code In The LTE System

The Turbo code is a kind of excellent channel coding schemes. In AWGN channel, its performance can approach the Shannon limit and it is suitable for the situation which has rigorous demand for power. So it is adopted by many communication standards such as the 3G and LTE system. The Turbo code in the LTE standard is mainly focused on in this paper and the key elements such as interleavers and decoding algorithms which can affect its performance are studied. The problems in the coding scheme are pointed out and the improved methods are also proposed which are proved to be valid by simulations.Firstly three different kinds of Turbo codes are introduced and their encoding and decoding structures are also described. After that, the effect of different parameters on the performance of Turbo codes is studied by simulations, such as code length, code rate, constituent codes and iteration times. A conclusion is made out that both the performance and complexity should be considered and therefore a balance is needed when a Turbo code is designed. With the same parameters the performance of PCCC and SCCC is compared. The result shows that SCCC is superior to PCCC in high SNR regions because no error floor is observed in SCCC. But the conclusion is reversed in low SNR regions.Secondly the interleaver design of Turbo codes is studied. The designing criterion is presented and typical interleavers are also introduced. Different from the QPP interleavers whose permutation polynomials are confined to the second order, a new interleaver design based on high order permutation polynomials is proposed and the criterion of selecting the coefficients is given. Compared with QPP interleavers, the new interleaver has lower error floor in high SNR regions. After that, the parallel decoding issue based on the new interleaver is discussed. The results show that the new interleaver is a maximum contention-free interleaver and with proper boundary metrics decoding delay can be sufficiently diminished by parallel decoding with no performance degeneration.Finally the decoding algorithm of Turbo codes is studied. The MAP algorithm and its simplified version are deduced. Because of the complexity of the correction function in the Log-MAP algorithm, a method based on polynomial approximation is proposed to approximate the correction function, in which the logarithmic and exponential algorithms are replaced by addition and multiplication. Simulation results show that in both AWGN and Rayleigh channels the improved Log-MAP algorithm based on the second order polynomial approximation has analogous performance to the Log-MAP algorithm but has lower complexity and decoding delay.