Study On Turbo Code Decoding Structure

Turbo code as a channel coding scheme attractes much attention for its outstanding performance and is widely used in modern digital communication systems. The reason for Turbo code getting a very closely performance to the Shannon limit performance is that it adops the iterative decoding idea. However, large iterative decoding complexity and long time delay of iterative process restrictes its application to the communication systems which require high level of real time performance. Therefor, how to improve decoding performance and reduce time delay is not only the focus study of Turbo code but also the mainly research of this thesis.This thesis takes a research on the iterative decoding structure of Turbo code decoding structure. Through analysising Turbo code decoding structure and transmission characteristics of exterior information, based on the time series, proposes two improved iterative decoding structures based on predictive algorithm. One of them is the parallel doecoding structure of Turbo code based on double predictive control. The other is the parallel decoding structure of Turbo code based on AR module double predictive control.The parallel decoding structure of Turbo code based on double predictive control (DPC-Turbo), namely, adds a linear model between two component decoders, using double predictive exterior information values of the component decoder instead of two iterative decoding procedures. The computational complexity and computing time of the linear prediction control module is less than the traditional parallel iterative decoding algorithm. Exterior information simulations show that comparing to the traditional parallel iterative decoding, time delay can be reduced at least two times by using predictive scheme each time, double using predictive scheme, time delay can be reduced by at least four times. While the BER performance simulation results show that the error-rate performance of the new decoding structure based on DPC-Turbo have a good approximation BER curves to the traditional one. When reasonably choosing the two predictive intervals, it can achieve the coincident BER curve.The parallel decoding structure of Turbo code based on AR double predictive control (AR-Turbo), namely, adds an AR model between two component decoders, using double predictive exterior information values of the component decoder instead of two iterative decoding procedures. The computational complexity and computing time of the AR predictive control module is less than the traditional parallel iterative decoding algorithm. Exterior information simulations show that comparing to the traditional parallel iterative decoding, time delay can be reduced at least three times by using predictive scheme each time, double using predictive scheme, time delay can be reduced by at least six times. While the BER performance simulation results show that the error-rate performances of the new decoding structure based on AR-Turbo have almost the same, sometims even better BER curves to the traditional one. When reasonably choosing the two predictive intervals, it can achieve the coincident BER curve. Namely, at the premise of getting the closely bit error rate performance, the simplified decoding structures this thesis proposed effective reduce iterative delay.