Long-Term Spectrum Monitoring with Big Data Analysis and Machine Learning for Cloud-Based Radio Access Networks

Abstract

Spectrum monitoring is important for efficient spectrum sharing and resource management in cloud-based radio access networks (C-RAN). In this paper we show how data obtained from long-term spectrum monitoring together with machine learning (ML) operating on big data (BD) can be used in a C-RAN scenario for spectrum management purposes. We propose an approach for spectrum occupancy forecasting which can be used to reduce the delay in making dynamic spectrum allocation decisions and improve the cognitive and management functionalities of cloud-based architectures such as C-RAN. The spectrum occupancy and usage activity in a predefined frequency band is based on the statistical processing of a large amount of collected data and the introduction of a frequency–time resources indicator as a measure of spectrum usage. Furthermore, we apply ML algorithms to predict spectrum usage and compare the predicted with actual measured data. Taking into consideration that the accuracy of the prediction depends on the volume of collected data and the time of prediction on the BD and ML approach, we propose the development of a cloud-based generic processing architecture to solve the “accuracy versus latency” trade-off problem. The proposed architecture is appropriate for deployment in cognitive C-RAN.