Improved approximate check polytope projection algorithm of ADMM penalized decoding for LDPC codes in IoTs

The alternating direction method of multipliers (ADMM) penalized decoding method is a popular technology for low-density parity-check (LDPC) codes in Internet of Things (IoT) communications. In this method, Euclidean projection is a key factor affecting computational complexity. Existing projection algorithms require either sorting operation or iterative execution, resulting in greater computational complexity. By binarizing the indicator vector, this paper proposes an improved approximate check polytope projection algorithm (I-APPA) to improve projection efficiency. The simulation results indicate that I-APPA can reduce calculation steps and improve projection efficiency. Therefore, the proposed I-APPA algorithm is convenient for hardware implementation in IoT devices.