In this paper, we propose clustering-based blind detection methods with the aid of systematic convolutional low density generator matrix (SC-LDGM) codes. Inspired by the fact that the received signals naturally… Click to show full abstract
In this paper, we propose clustering-based blind detection methods with the aid of systematic convolutional low density generator matrix (SC-LDGM) codes. Inspired by the fact that the received signals naturally fall into clusters in the block-fading channels, we develop a system constrained Gaussian mixture model (SCGMM) by taking into account the inherent characteristics of communication systems, in which the parameters can be estimated by the expectation-maximization (EM) algorithm. We also present an initialization method for the proposed SCGMM to accelerate convergence. After clustering, a decoding algorithm of SC-LDGM codes is designed to resolve the centroid ambiguity. To further improve the detection performance in the low signal-to-noise (SNR) and deep fading scenarios, we propose an improved label-assisted (ILA) method, which integrates the label-assisted (LA) information into the blind detection algorithm. Numerical results show that the performance of our methods can closely approach the performance with perfect channel state information (CSI). The simulation results also show that the performance can be improved by increasing the encoding memory.
               
Click one of the above tabs to view related content.