RBIR-Based Opportunistic Fast Simplified Successive-Cancellation Decoding of Polar Codes

Decoding polar codes through the depth-first binary tree-search based successive-cancellation (SC) algorithm takes a lot of time. Multiple fast simplified successive-cancellation (FSSC) decoding algorithms have been proposed in the literature to reduce the decoding latency and thereby the power consumption, by finding special nodes in the decoding tree, at which instantaneous decoding can be done. We observed that the latency can also be reduced by applying instantaneous decoding at all the nodes of an identified depth level in the decoding tree, as early as the root level at higher signal to noise ratio (SNR) conditions, without the need of traversing till the leaf node. Received bit information rate (RBIR) is a link abstraction metric that can be used to capture the SNR and identify the depth-level for instantaneous decoding. Existing FSSC algorithms rely only on the special nodes, and doesn't opportunistically utilize the time-varying and frequency-selective channel conditions for early depth level decoding. Whereas, relying on only RBIR-based opportunistic early depth level SC decoding (RBIR-OSC) will miss out on the special nodes of FSSC, that can provide significant power saving even in lower SNR conditions. In this paper, we propose an RBIR-based opportunistic FSSC decoding (RBIR-OFSSC) for polar codes, which combines the instantaneous decoding at special nodes of FSSC with the RBIR-OSC. We show that the proposed RBIR-OFSSC decoding reduces latency by 98% compared to conventional SC algorithm, 81% compared to FSSC, and 79% compared to RBIR-OSC, without any performance loss, for the 3GPP defined 5G NR uplink control channel.