Sorting of frequency hopping communication transmitter based on NFJMIM feature selection

There are unrelated features and redundancy features in the fine feature set of frequency hopping (FH) communication transmitter, which lead to large computation, low recognition accuracy and poor timeliness in sorting. To solve these problems, a FH communication transmitter sorting method based on normalized fuzzy joint mutual information maximum (NFJMIM) was proposed. Firstly, the fine feature set of FH transmitter is obtained, and the mutual information between fine features and category information is calculated and select the largest feature of mutual information. Secondly, the feature selection feature with the largest normalized fuzzy mutual information is selected in the candidate feature subset. Finally, the improved semi-supervised Collaborative Representation Classifier (CRC) is used to sort FH communication transmitter. Experimental results show that the proposed NFJMIM feature selection algorithm can obtain a reasonable optimal feature set. On the basis of ensuring the correct rate of sorting, the computational complexity is greatly reduced.