Lte System Channel Decoding Research And Multi-core Dsp Implementation

LTE (Long Term Evolution) system is a quasi-4G mobile communication system,which supports a smooth transition from3G to4G. In order to meet the requirements ofthe users with high-speed data rate, LTE provides100Mbps downlink and50Mbpsuplink peak rate with20M bandwidth.TBCC (Tail Biting Convolutional Codes) and Turbo codes are main channelcoding schemes in LTE system. They improve transmission reliability but meanwhilecause a large decoding delay. Therefore the channel decoders must be parallel to reducethe delay.Traditional decoding algorithm for TBCC and Turbo codes are studied in thisthesis, and their BER (Bit Error Rate) performance and computing complexity areanalyzed and compared with. On this basis, parallel decoding algorithms are studied.By taking the characters of TBCC and Viterbi decoding algorithm intoconsideration, a parallel decoding algorithm based on sub-blocks with overlapping isproposed in this thesis. Its computing complexity is lower than existing parallelalgorithms. We simulate BER performance of it in MATLAB and implement it on TImulti-DSP platform, C6474EVM (TMS320C6474Evaluation Module). It shows that forlong code blocks, the decoding latency is inversely propotional to the number ofdecoding cores.For Turbo codes, existing parallel decoding algorithm are compared with and analgorithm based on sub-blocks without overlapping is choosed. We simulate BERperformance of it in MATLAB and implement it on TI multi-DSP platform, C6474EVM.It shows that the decoding latency is inversely propotional to the number of decodingcores.