Research On Polarization Code Structure For 5G Mobile Communication Systems

Channel coding is one of the key technologies of mobile communication system, and its ultimate goal is to approach the Shannon limit. Although Turbo code and Low-Density Parity-Check(LDPC) code well approach the Shannon limit, they cannot achieve the Shannon limit theoretically. Polar code is the only approach that can reach the Shannon limit in theory so far. The construction of polar coding is based on channel polarization phenomenon where a part of channel symmetric capacities are near 1, while a part of channel symmetric capacities are near 0. Polar coding selects the channels whose channel symmetric capacities are near 1 as the free bit for transmitting information bit and selects the channels whose channel symmetric capacities are near 0 as the frozen bit. How to utilize an appropriate evaluating parameter to select the best part of channel as the free bit is significant to improve the performance of polar code.This paper utilizes the mean of the probability density function of Log Likehood Ratio(LLR) to evaluate the performance of channel, and provides the construction method of polar code based on this evaluating parameter. First, we select the mean of the probability density function of LLR as the metric. Second, this metric can be calculated by a recursive relation,where an initial parameter(designed signal to noise ratio) should be set. Third, the indices of channels that have bigger value of metric are saved into a vector(index vector of polar codes).Fourth, the encoding module and decoding module of polar codes utilize this index vector to carry out their functions. Fifth, the decoded sequences are used to calculated bit error ratio(BER) and frame error ratio(FER), then the performance of polar codes under different designed signal to noise ratio is compared to find the optimal design value. Last, in order to validate this construction method, we simulate three polar coding approaches(SC,SCL,CRC-SCL) based on our construction method to compare their performance.Simulation results show that our approach has a wider scenario(non BEC channel) and a better performance. In fact, we can find the optimal designed signal to noise ratio and index vector by utilizing this construction method based on real condition of channel.