Communication spectrum prediction method based on convolutional gated recurrent unit network.

Area of Science:

• Wireless Communication
• Signal Processing
• Machine Learning

Background:

• Spectrum scarcity challenges modern wireless systems.
• Dynamic spectrum changes necessitate efficient resource management.
• Accurate spectrum prediction is crucial for system performance.

Purpose of the Study:

• To develop and evaluate a communication spectrum sensing and prediction model.
• To improve the efficiency and accuracy of spectrum resource utilization.
• To address the challenges posed by dynamic spectrum changes.

Main Methods:

• Constructed a communication collaborative spectrum sensing model using channel aliasing dense connection networks.
• Developed a communication spectrum prediction model combining convolutional neural network (CNN) and gated cyclic unit (GRU) networks.
• Utilized deep learning analysis of massive historical communication data.

Main Results:

• The spectrum sensing model achieved a maximum perception accuracy of 0.99.
• The proposed spectrum prediction model reached a high accuracy of 0.95 within 208 seconds.
• Outperformed traditional models like RNN, LSTM, and ConvLSTM in both accuracy and speed.

Conclusions:

• The developed perception and prediction model demonstrates strong performance for wireless communication.
• The model aids in monitoring spectral changes and optimizing spectrum resource usage.
• This research contributes to more efficient utilization of limited spectrum resources.