Spectrum sensing method based on small sample training neural network

Abstract

The invention provides a spectrum sensing method based on a small sample training neural network. The method comprises the following steps: pre-training a neural network detector; pre-deploying a detector in an actual operation environment of a system; calculating fine adjustment values and test loss values of the detector in different sensing environments through iteration of a small amount of data and a small amount of gradient, then calculating gradient of initial parameter corresponding to the loss value, conducting gradient updating on the initial parameter so that the initial parameter of the detector can rapidly adapt to change of the environment; performing online adjustment based on initial parameters of a pre-trained neural network detector; and inputting the sampled received signal into the adjusted neural network detector to predict the probability of whether the frequency spectrum is occupied at the moment. According to the spectrum sensing method, the detection performance similar to that of an existing neural network detector based on a large number of samples and gradient iteration can be achieved, and the calculation amount and the sample number required by detector adjustment can be effectively reduced.

Description

technical field [0001] The invention belongs to the technical field of wireless communication networks, and relates to a spectrum sensing method based on small sample training neural network. Background technique [0002] Spectrum sensing, that is, judging whether the current spectrum is occupied by receiving signals, has received more and more attention. Because of its wide range of applications, spectrum sensing technology is widely used in many communication scenarios. For example, cognitive radio technology and anti-jamming communication technology. [0003] Research on spectrum sensing technology generally assumes certain prior information, such as known communication characteristics or channel characteristics of the sending end. However, such prior information is often difficult to obtain in practical applications, which leads to poor practicability of this type of method. With the rapid development of machine learning technology, existing research gradually focuses...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide a spectrum sensing method based on small-sample training neural network to solve the technical problems in the prior art that require a large number of samples and a large number of calculations in the neural network-based spectrum sensing method

Images