An analytical method to mitigate the false alarm rates of polar code blind detection in 5G new radio

In 5G new radio (NR), polar codes are adopted for eMBB downlink control channels where the blind detection is employed in user equipment (UE) to identify the correct downlink control information (DCI). However, different from that in the 4G LTE system, the cyclic redundancy check (CRC) in polar decoding plays both error correction and error detection roles. Consequently, the false alarm rates (FAR) may not meet the system requirements (FAR< 1.52 × 10⁻⁵). In this paper, to mitigate the FAR in polar code blind detection, we attach a binary classifier after the polar decoder to further remove the false alarm results and meanwhile retain the correct DCI. This classifier works by tracking the squared Euclidean distance ratio (SEDR) between the received signal and hypothesis. We derive an analytical method to fast compute proper classification threshold that is implementation-friendly in practical use. Combining the well-designed classifier, we show that some very short CRC sequences can even be used to meet the FAR requirements. This consequently reduces the CRC overhead and contributes to the system error performance improvements.