This thesis proposes a code design method to improve the decoding performance of rate-matched polar codes. Error correction coding is essential for achieving reliable communication, and this study focuses on polar codes, one of the most powerful moder...
This thesis proposes a code design method to improve the decoding performance of rate-matched polar codes. Error correction coding is essential for achieving reliable communication, and this study focuses on polar codes, one of the most powerful modern channel coding schemes. The goal of this study is to construct polar codes by selecting highly reliable split channels for data transmission, a process referred to as code design. However, conventional code design methods exhibit degraded decoding performance for certain code parameter regions and do not sufficiently account for rate-matching, which is crucial for supporting flexible code lengths in practical systems. For instance, one of the well-known code designs, the polarization weight (PW)-based design method offers low complexity and good overall performance, but its reliance on a single design parameter β limits its robustness across diverse scenarios. To address this issue, we propose an improved code design scheme that maintains reliable decoding performance over a wider range of code parameters. Furthermore, we extended to a multiple-input multiple-output (MIMO) transmission scenario. A grouping and layer-mapping strategy is introduced to efficiently distribute polar-coded bits across multiple transmission layers, reflecting both channel quality differences and decoding characteristics. Simulation results demonstrate that the proposed mapping scheme achieves superior decoding performance compared to the conventional layer-mapping method adopted in 5G NR (new radio).