Method for identifying signal modulation mode

Abstract

The invention provides a method of constructing an identification model of a signal modulation mode. The method comprises the steps of: obtaining a training set composed of a plurality of sets of analysis data according to the correlation between the characteristic data of the plurality of data points in the cyclic spectrum of the modulated signal and the signal modulation mode, wherein each set of the analysis data includes the feature data of the plurality of data points in the cyclic spectrum and a corresponding modulation mode; and based on the training set, training the classification model by using the feature data of the plurality of data points in the cyclic spectrum as an input and using the corresponding modulation mode as an output, thereby obtaining the identification model ofthe signal modulation method. The invention provides a scalable signal modulation mode identification model, which is capable of improving the correct recognition rate of the signal modulation mode, and is particularly effective for low signal to noise ratio signals.

Description

technical field [0001] The invention relates to the field of communication technology, in particular to a method for identifying a signal modulation mode. Background technique [0002] Due to the rapid development of communication technology, communication norms and systems are constantly breaking through, especially in non-cooperative communication, see figure 1 In the non-cooperative communication process shown, non-cooperative communication is a communication mode that accesses the cooperative communication system without affecting the normal communication between the cooperative communication sender and the cooperative communication receiver, and is an unauthorized access communication mode. Since the non-cooperative receiver cannot know the relevant information of the sender like the two parties in the cooperative communication, it is necessary to complete signal detection, modulation identification and demodulation in the unlicensed mode. With the application of non-s...

AI Technical Summary

Problems solved by technology

Looking at the various feature parameters used in the pattern recognition method, it is found that there is basically no unified rule to follow for feature extraction, and it is difficult to find a general feature and method for modulation classification. For each classification problem, it must be specific. Analysis, depending on the type of modulation that needs to be classified, looking for specific methods and characteristics, and finding characteristics is very difficult work and requires a lot of effort

[0004] In summary, there are two main problems in the identification of modulation methods in the prior art: it is not suitable for low signal-to-noise ratio environments; the types and quantities of modulations that can be identified are limited, and the scalability is poor

The most important cause of these two problems is artificial feature extraction (such as artificially extracting the number of peaks, etc.), because artificial feature extraction has the following disadvantages: First, artificial feature extraction will discard a lot of useful information, only retaining The artificially observed information makes the abstract representation of the signal incomplete, and generally the artificially extracted features are a value. Under low signal-to-noise ratio, the value will deviate a lot from the normal value, so it is used to identify the modulation mode. The time performance is poor; the second is that feature extraction is difficult, and each classification problem must be analyzed in detail. According to the different modulation types of the required classification, specific methods and features are searched, and the scalability is poor.

Images