Quantization of CRC-Aided Successive Cancellation List Decoder for Polar Codes

In order to implement a low-complexity CRC-aided successive cancellation list decoder for polar codes on digital chips, we study the quantization technologies for the initial log likelihood ratios and the path metric distance of the decoder. In order to reduce the block error rate performance loss using the uniform quantization method, we analyze various optimization techniques such as quantization bits and quantization precision for the initial log likelihood ratios and the path metric distance. We provide quantization simulation for realistic code length N=2 10 and code rate R= 0.5. Simulation results show that 6 bits quantization on interval [-31, +31] for the initial log likelihood ratios and 8 bits quantization on interval [-255, +255] for the path metric distance can not only bring the block error rate performance loss less than 0.1 dB, but also decrease the hardware implementation complexity significantly.