Wireless Channel Scenario Classification Based on ISODATA

In recent years, the research of using machine learning for channel scenario classification has received increasing attention. However, when wireless channel is continuously measured and the receiver is moving, the number of channel scenarios is difficult to be decided artificially. A novel scenario classification method based on Iterative Self-organizing Data (ISODATA) algorithm is proposed. The proposed scheme first adaptively adjusts the optimal number of scenarios based on ISODATA, then use the number of scenarios for clustering with the Gaussian Mixture Model (GMM) algorithm, and its superior performance is demonstrated.