Non-Data Aided Rician Parameters Estimation in Temporal Fading Channel With 3 DoFs Gaussian Mixture Model

Rician distribution has been widely utilized to describe wireless fading channel. In the non-stationary temporal fading channel like industrial scenarios, both the specular and scattered components of the multi-path fading channel will be time varying. As a result, the online estimation of Rician pa...

Full description

Saved in:
Bibliographic Details
Published in:IEEE access Vol. 7; pp. 62198 - 62209
Main Authors: Zhang, Qilong, Lu, Guobao, Zhang, Wuxiong, Shen, Fei, Dai, Xuewu, Jiao, Jianbin, Qin, Fei
Format: Journal Article
Language:English
Published: Piscataway IEEE 2019
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Subjects:
Algorithms
Estimation
Fading
Fading channels
Gaussian mixture model
Mathematical model
Maximum likelihood estimation
Modulation
non-data aided
Parameter estimation
Probabilistic models
Reliability
Rician channels
Rician distribution
Two dimensional models
Wireless communication
ISSN:2169-3536, 2169-3536
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Rician distribution has been widely utilized to describe wireless fading channel. In the non-stationary temporal fading channel like industrial scenarios, both the specular and scattered components of the multi-path fading channel will be time varying. As a result, the online estimation of Rician parameters is necessary to provide stable wireless service. The traditional estimation approaches of Rician parameters are designed for channel measurement usage and therefore have to work in the data-aided mode for online estimation with modulated I/Q samples. To solve this problem, some non-data-aided algorithms have been proposed in recent years, but only valid in specific scenarios. In this paper, we formulate the estimation of Rician parameters from modulated I/Q samples as a two-dimensional Gaussian mixture model to provide a general non-data-aided Rician parameter estimation method. By involving a priori information of modulation scheme and the motivation of optimized gradient searching, the independent parameters in the maximum likelihood estimation can be significantly decreased to three, which leads to fast convergence of the modified expectation-maximization algorithm with high accuracy. The combination of these modifications has been finally formulated as a Rician mixture model. The numerical results and field measurements illustrate the feasibility of this methodology.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:2169-3536
2169-3536
DOI:10.1109/ACCESS.2019.2903931