Large-Scale Spectrum Occupancy Learning via Tensor Decomposition and LSTM Networks

• Published in: 2020 IEEE International Radar Conference (RADAR) pp. 677 - 682
• Main Authors: Alkhouri, Ismail, Joneidi, Mohsen, Hejazi, Farzam, Rahnavard, Nazanin
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.04.2020
• Subjects: Correlation; LSTM time-series prediction; Radar; Recurrent neural networks; Spectrum occupancy learning; Support vector machines; Tensor CP decomposition; Tensors; Time series analysis; Time-frequency analysis
• ISSN: 2640-7736
• DOI: 10.1109/RADAR42522.2020.9114785

A new paradigm for large-scale spectrum occupancy learning based on long short-term memory (LSTM) recurrent neural networks is proposed. Studies have shown that spectrum usage is a highly correlated time series over multi-dimensions. Therefore, revealing all these correlations using one-dimensional time series is not a trivial task. In this paper, we introduce a new framework for representing the spectrum measurements in a tensor format. Next, a time-series prediction method based on CANDECOMP/PARAFAC (CP) tensor decomposition and LSTM recurrent neural networks is proposed. Our proposed method is computationally efficient and is able to capture different types of correlation within the measured spectrum. Moreover, it is robust to noise and missing entries of sensed spectrum. The superiority of the proposed method is evaluated over a large-scale synthetic dataset in terms of prediction accuracy and computational efficiency.