The Research On Encoding And Decoding Technologies Of Raptor Codes In Lte Downlink

Fountain codes are a novel kind of Forward Error Correction codes which have the advantages of rateless, disorderly and low complexity of decoding, which make them be used widely to the radio, Mobile Communication, Deep Space Communications and so on.This paper studys on encoding and decoding technologies of one kind of Fountain codes-Raptor codes, then compare it with Tubro codes as an channel coding of LTE downlink. The simulation is based on IT++, which is combined Matlab containing many communication modules with C++ possessing high velocity, saving user’s time.Firstly, this paper presents the background and development of fountain codes and then made a brief analysis of the strengths and weaknesses of Turbo codes and Raptor codes. Raptor codes own the natural advantage of parallel decoding architecture, so it is easier to implement than Turbo codes. Secondly, we elaborate the basic principle of Raptor codes,noting that only long block codes can reflect excellent performance, short cycles and low-degree variable nodes in the generator matrix cause the degeneration of performance in the short-block codes.Baed on this,we simulate the reference encoding scheme:PEG algorithm and average degree algorithm which is demonstrated the former can’t enhauce the performance obviously and the latter can reduce the error floor but be restricted by the structure of Tanner map. So, this paper put forward to processing the coding matrix, in this way the log likelihood information can smoothly pass along the edge between the varaibality nodes and check bodes and improve the performance by this method, which is ligntly different from the reference. Then, concerning about decoding, it is approximated by linear fitting of hyperbolic functions to reduce the amount of computation. And we simulate it, demonstrating that the error performance remains well.Finally, we investiget the performance of Turbo codes and Raptor codes in noisy channel and find that new algrithem of Raptor codes of length 5000 outperform Turbo codes, but not well than Turbo codes of length 200 and 2000. When the length gorws, Raptor codes gradually work better than Turbo codes.Then we establish a communication simulation platform of LTE downlink physical layer, including of transmission channel and physical channel, while implementing the HARQ technology. We simulate and compare the performance of Raptor codes and Turbo codes at different block length and code rate.