Improved Construction of Short Polar Codes for URLLC With Low-Power IoT Devices

To address the challenges of reliable and efficient short-packet communication in the Internet of Things (IoT), especially under ultrareliable and low-latency communication (URLLC) constraints, this work optimizes the design of short polar codes tailored for low-power IoT devices. To improve the reliability of short polar codes under successive cancellation list (SCL) decoders, we introduce a novel heuristic optimization algorithm guided by a unified metric. This algorithm carefully balances the tradeoff between the number of minimum-weight codewords (also known as error coefficient) and the reliability of selected information subchannels. Through a guided and deliberate disruption of the partial order property of polar codes, our algorithm reduces the error coefficient to enhance maximum likelihood (ML) decoding performance, while managing the impact on subchannel reliability. Numerical results demonstrate a consistent and significant improvement over the baseline RM-Polar and Gaussian approximation (GA) constructions across various code parameters. Furthermore, our approach features low offline design complexity, achieving state-of-the-art or highly competitive performance against other advanced schemes, particularly at low code rates. This makes our method highly suitable for URLLC, as the resulting optimized codes can be deployed on existing 5G hardware with zero additional on-device decoding complexity, while the achievable coding gain directly translates into transmission energy savings.