A Low-Complexity SNR Estimation Algorithm Based on Frozen Bits of Polar Codes

Polar codes attract a lot of attention due to their provably capacity-achieving property with low encoding and decoding complexity. Frozen bits, the special structure of polar codes, are helpful for signal-to-noise ratio (SNR) estimation. A novel SNR estimation algorithm based on frozen bits is proposed in this letter. In the proposed algorithm, the mapping from SNR to frozen bit error rate is theoretically analyzed when code length is infinite and approximately calculated when code length is finite. Then, SNR can be estimated according to the mapping. Simulation results show that the proposed algorithm has better estimation performance with a relatively lower complexity.