Signal spectrum hole sensing method based on time sequence attention mechanism and LSTM model

Abstract

The invention provides a signal spectrum hole sensing method based on a time sequence attention mechanism and an LSTM model, and relates to the technical field of wireless communication. The method comprises the following steps: 1, receiving spectrum data and binarizing the spectrum data; 2, serializing the spectrum data and constructing a spectrum data set; 3, constructing and training an LSTM model, and inputting the serialized spectrum data set into the model; 4, extracting time sequence characteristics in the spectrum data by using the model, wherein the time sequence characteristics comprise a plurality of hidden state characteristics ht of each group of time ordinal numbers; 5, learning time sequence features by using a time sequence attention mechanism in the model to obtain a prediction vector of the model, and judging a median of the prediction vector to obtain a signal state of the user. According to the method, the attention mechanism is applied to the spectrum hole sensing task, the prediction performance when spectrum signals are distributed complexly is improved, and the accuracy of spectrum hole sensing is remarkably improved.

Description

technical field [0001] The invention relates to the technical field of wireless communication, in particular to a signal spectrum hole sensing method based on a timing attention mechanism and an LSTM model. Background technique [0002] The spectral hole sensing method is one of the important tasks of cognitive radio, which affects all aspects of people's lives, ranging from facilitating communication between people to promoting the development of the country's social economy. However, with the continuous increase of the number of users, new challenges are brought to the wireless communication technology, and the contradiction between user needs and insufficient spectrum resources becomes increasingly prominent. According to the research conducted by the University of California, Berkely, on the measurement of the 30MHz to 6GHZ frequency band, it is found that the current utilization efficiency of the spectrum in the time domain and frequency domain is less than 50%. In ord...

AI Technical Summary

Problems solved by technology

Using the LSTM model based on the time series attention mechanism to fully extract the time series features of the spectrum data, without prior information of the signal, to solve the problem that the traditional spectrum hole perception method has poor perception effect under low signal-to-noise ratio, requires prior information, and is interfered by adjacent frequency bands And other issues

Images