Learning-Based Spectrum Occupancy Prediction Exploiting Multi-Dimensional Correlations
Explore the concept of multi-dimensional spectrum occupancy prediction in wireless networks. Discover methods leveraging time, frequency, and space correlations for accurate spectrum prediction and efficient spectrum utilization. Validate the feasibility through experimental measurements. Address the critical need for efficient spectrum usage in the evolving landscape of wireless communication technologies.
Download Presentation
Please find below an Image/Link to download the presentation.
The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author. If you encounter any issues during the download, it is possible that the publisher has removed the file from their server.
You are allowed to download the files provided on this website for personal or commercial use, subject to the condition that they are used lawfully. All files are the property of their respective owners.
The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author.
E N D
Presentation Transcript
October 2020 doc.: IEEE 802.11-20/1709r2 Learning-based spectrum occupancy prediction exploiting multi-dimensional correlations Date: 2020-11-02 Authors: Name Mehmet Ali Aygul Affiliations VESTEL, IMU Address Istanbul Medipol University Istanbul Medipol University Istanbul Medipol University Vestel Electronics Corp. email mehmet.aygul@std.medipol.edu.tr mahmoud.nazzal@ieee.org Mahmoud Nazzal VESTEL, IMU huseyinarslan@medipol.edu.tr H seyin Arslan VESTEL,IMU basak.ozbakis@vestel.com.tr Ba ak zbak VESTEL Submission Slide 1 Mehmet Ali Aygul, IMU; Vestel
October 2020 doc.: IEEE 802.11-20/1709r2 Outline Abstract Spectrum occupancy prediction A motivation to using multi-dimensional correlations Time and frequency Space Methods for exploiting multi-dimensional correlations The tensor-based method Composite 2D-LSTMs-based methods Experiments and results Summary Submission Mehmet Ali Aygul, IMU; Vestel Slide 2
October 2020 doc.: IEEE 802.11-20/1709r2 Abstract Several sensing use cases have been discussed recently. Accurate spectrum prediction and efficient utilization of the spectrum is critical to support them. In this contribution, the multi-dimensional spectrum occupancy prediction concept is discussed, and some examples in the literature of these methods are given. The feasibility of the methods is validated through experimental measurements. Submission Slide 3 Mehmet Ali Aygul, IMU; Vestel
October 2020 doc.: IEEE 802.11-20/1709r2 Spectrum occupancy prediction Wireless networks and information traffic have grown exponentially over the last decade. The radio spectrum is a limited resource, and it should be used efficiently. ? = n n H0: H1: there is no PU a PU is present Hx Hx+n n Problem: finding the spectrum occupancy Submission Mehmet Ali Aygul, IMU; Vestel Slide 4
October 2020 doc.: IEEE 802.11-20/1709r2 A motivation to using multi-dimensional correlations Time and frequency correlation (b) (a) High correlation in the neighboring frequency bands of each operator A strong correlation across both time and frequency for each operator individually Submission Slide 5 Mehmet Ali Ayg l, M ; Vestel
October 2020 doc.: IEEE 802.11-20/1709r2 A motivation to using multi-dimensional correlations Space correlation Submission Mehmet Ali Aygul, IMU; Vestel Slide 6
October 2020 doc.: IEEE 802.11-20/1709r2 Methods to using multi-dimensional correlations The tensor-based method Motivated by the existence of a multi-dimensional correlation, current literature uses tensor-based methods for spectrum occupancy prediction. Such methods have high processing time, and they assume that 3D data can be provided at any time. Submission Mehmet Ali Aygul, IMU; Vestel Slide 7
October 2020 doc.: IEEE 802.11-20/1709r2 Methods to using multi-dimensional correlations Composite 2D-LSTMs-based method: Submission Slide 8 Mehmet Ali Ayg l, M ; Vestel
October 2020 doc.: IEEE 802.11-20/1709r2 Experiments and results A real dataset is used for simulations. Compared ARM, BIF, 1D-LSTM using only time correlation, 2D-LSTM using time and frequency correlations, ConvLSTM using multi-dimensions as a tensor, and the composite 2D-LSTMs-based method. Validations in terms of precision ( ), recall ( ), and F1-score performance metrics The precision metric quantifies what percentage of positive results are actually positive. The recall provides information on what percentage of true positives are identified correctly as positive. The F1-score gives an overall measure with the harmonic average of precision and recall for the accuracy of a classifier model. ? ? ? ? + ? ? ? x , F1 score = 2 ? = ? + ?,? = ? + ? Submission Slide 9 Mehmet Ali Aygul, IMU; Vestel
October 2020 doc.: IEEE 802.11-20/1709r2 Experiments and results Accuracy comparisons Measure Measure Method Method F1 score 0.9440 0.9495 0.9690 0.9726 0.9762 F1 score 0.8783 0.9232 0.9312 0.9338 0.9388 ? ? ? ? ARM BIF 1D-LSTM 2D-LSTM ConvLSTM Composite 2D-LSTMs 0.9210 0.9264 0.9602 0.9720 0.9760 0.9681 0.9738 0.9780 0.9732 0.9763 ARM BIF 1D-LSTM 2D-LSTM ConvLSTM Composite 2D-LSTMs 0.8863 0.9336 0.9465 0.9462 0.9479 0.8704 0.9130 0.9165 0.9216 0.9298 0.9727 0.9742 0.9735 0.9476 0.9233 0.9353 Submission Slide 10 Mehmet Ali Aygul, IMU; Vestel
October 2020 doc.: IEEE 802.11-20/1709r2 Experiments and results Complexity comparisons Execution Time (s) Training Execution Time (s) Training Method Method Testing Testing The tensor- based method The tensor- based method 610.3 2.9 608.7 2.7 Composite 2D-LSTMs Composite 2D-LSTMs 58.9 0.8 57.8 0.7 Submission Slide 11 Mehmet Ali Aygul, IMU; Vestel
October 2020 doc.: IEEE 802.11-20/1709r2 Summary Multi-dimensional spectrum occupancy prediction is discussed to use spectrum efficiently. A composite 2D-LSTMs-based method is shown as a sub-optimal and more realistic method. Relevant experimental measurement is shown to prove the feasibility of this method. Submission Slide 12 Mehmet Ali Aygul, IMU; Vestel
