Turbo Code Decoding Structure Based On Prediction Control

The mobile communication technology has been toward a comprehensive 3G era, which requires communication systems must have a higher communication quality and transmission performance. As an performance excellent channel coding scheme, Turbo code has been widespread concerned by scholars all over the world, up to this day, Turbo code theories have been rapidly developed, particularly in the improvement of iterative structure and decoding algorithm, which is still the focus of the scholars.Turbo code, also known as parallel concatenated convolutional code(PCCC, Parallel Concatenated Convolutional Code). Turbo codes combined the convolutional codes with random interleaver to achieve the idea of random coding and constructing the long length codes by using the short length codes, meanwhile adopting the soft-output decoding algorithm to approximate the maximum likelihood decoding. Since the Turbo code take the very good conditions fit of the Shannon channel coding theory in the random encoding and decoding, it gains a performance which is very close to the Shannon limit performance.The following aspects are included in the main thesis research:1. Two principles of interleaver design are stressed: odd-even protection and symmetrical characteristic, three kinds of symmetrical methods are presented. The odd-even protection and symmetrical characteristic are used in T-Chaotic interleaver (TCOP). Experimental data show that the design of introducing odd-even protection symmetrical characteristic can avoid unequal protection which is brought in the punctured procession, significantly reduce time delays and simplify the complexity of system implementation. Meanwhile, since the generation time delay of the TCOP algorithm is small, just fixing a few variable values such as 2 chaotic initial values, 3 the selected value of symmetric algorithms, the frame can be changed and symmetric algorithm at interleaver can be switched. So it is easy to achieve variable-depth interleaver. 2. The idea of adopting iterative decoding is one of the main reasons why Turbo codes can obtain the excellent performance which is closely to the Shannon limits. In the analysis of Turbo codes decoding structure and transmission characteristics of exterior information, based on the time series and then from the point of view of an exterior decoder component of the transmission of information laws, present an improved iterative decoding structure based on prediction algorithm. Proposed single and double predictive algorithms to the serial decoding and parallel decoding structure respectively. Establish the n-order polynomial linear model of the component decoder, save the values of the former n-iterations exterior information to the linear model, adopting the polynomial fitting curve or the index component algorithm to predict theâ„–(n+1) iterative exterior information. Take the value of predicted exterior information instead of theâ„–(n+1) iterative exterior information. Under the premise of accurately predict the exterior information and the closely bit error rate performance, it simplifies the decoding structure and reduces the number of iterations and iteration delay.3. Establish Turbo codes decoding structure model based on Simulink communication simulation tool, implement simulations in different interleaver lengths, decoding algorithms and numbers of iterations, and analysis the error-correcting performance influence of adopting linear prediction module. Simulation results show that compared with conventional Turbo codes, Turbo decoding model based on linear prediction algorithm can reduce decoding time, meanwhile, there is only a difference of 0 ~ 0.2dB, the lower the SNR is, the better the performance is. Interleaving length and decoding algorithm have the close impact of performance of the traditional algorithm and the decoding algorithm based on linear prediction control. The longer the interleaving length is, the better the error correction performance is, however, the decoding time increases with the length of interleaver length, so once choose an appropriate length; it can achieve the best decoding performance. In the decoding algorithms, Log-MAP algorithm has the best performance, when in the bit error rate of 10-2, Log-MAP algorithm and MAX-Log-MAP algorithm has 0.5dB difference.The unstable predicted value may occur in the process of predicting decoding of the random transmission frame. The compensation algorithm and hardware implementation of Turbo codes based on Linear Prediction Algorithm will be done in the next step.