Nondata-Aided Rician Parameters Estimation With Redundant GMM for Adaptive Modulation in Industrial Fading Channel
Guobao Lu,Xuewu Dai,Wuxiong Zhang,Yang Yang,Fei Qin
DOI: https://doi.org/10.1109/tii.2021.3095253
IF: 12.3
2022-04-01
IEEE Transactions on Industrial Informatics
Abstract:Wireless networks have been widely utilized in industries, where wireless links are challenged by the severe nonstationary Rician fading channel, which requires online link quality estimation to support high-quality wireless services. However, most traditional Rician estimation approaches are designed for channel measurements and work only with nonmodulated symbols. Then, the online Rician estimation usually requires a priori aiding pilots or known modulation order to cancel the modulation interference. This article proposes a nondata-aided method with redundant Gaussian mixture model (GMM). The convergence paradigm of GMM with redundant subcomponents has been analyzed, guided by which the redundant subcomponents can be iteratively discriminated to approach the global optimization. By further adopting the constellation constraint, the probability to identify the redundant subcomponent is significantly increased. As a result, accurate estimation of the Rician parameters can be achieved without additional overhead. Experiments illustrate not only the feasibility but also the near-optimal accuracy.
automation & control systems,computer science, interdisciplinary applications,engineering, industrial