Frequency spectrum prediction method based on radio frequency machine learning model driving

Abstract

The invention discloses a frequency spectrum prediction method based on radio frequency machine learning model driving. The method comprises the following steps: S1, collecting frequency spectrum data, and preprocessing the collected frequency spectrum data; S2, determining an order for an autoregression model according to an akaike information criterion, and determining a step length M of input data; S3, expanding a linear combination process of the autoregression model into an M-layer network structure, and introducing new trainable parameters into the M-layer network structure to construct M layers of frequency spectrum prediction network models driven by a radio frequency machine learning model; S4, training the spectrum prediction network model by using the training set data; and S5, judging whether training is completed or not, if so, inputting test set data into the trained spectrum prediction network model, outputting a prediction result, and ending the process; and if not, adding one to the number of training iterations, and returning to step S4 until the maximum number of iterations is reached. The network is endowed with interpretability, the prediction performance is improved, and the convergence speed of the network is increased.

Description

technical field [0001] The present invention relates to the field of communication technology, in particular to a spectrum prediction method driven by a radio frequency machine learning model. Background technique [0002] Spectrum prediction is a crucial technology in cognitive radio networks. It infers the use status of future spectrum by learning the regularity between historical spectrum data, which helps to reduce the large amount of time and energy consumed in the spectrum sensing process. Improve the efficiency of spectrum sensing. At the same time, spectrum prediction helps to improve the performance of spectrum management and control, and is an essential part of spectrum decision-making. Existing spectrum prediction methods can be divided into two categories, model-based prediction methods and data-based prediction methods. Model-based methods, such as autoregressive models, are based on the ideal assumption that data at the current point in time can be predicted ...

AI Technical Summary

Problems solved by technology

[0004] These purely data-driven methods all assume that there is sufficient labeled data to complete network training, but this cannot be achieved in actual communication scenarios

The reason is that it is difficult to obtain a large amount of high-quality label data in practice

In addition, this type of algorithm has a slow convergence speed and is difficult to achieve real-time spectrum prediction, resulting in prediction performance that cannot accurately reflect the real spectrum usage status, and cannot meet the real-time requirements of future wireless communication networks

Images