Anomaly Detection in Shared Spectrum

Tschimben, Stefan

Graduate Thesis Or Dissertation

Anomaly Detection in Shared Spectrum Public Deposited

Analytics

Citations

Citeable URL: https://scholar.colorado.edu/concern/graduate_thesis_or_dissertations/vh53ww97s

Abstract

Demand for wireless communication devices has been growing relentlessly since the advent of mobile communication. Even though spectral efficiency and throughput keep increasing, consumer demand continues to seemingly outpace that growth. Spectrum sharing is becoming a more attractive solution to solving various capacity constraints as the resulting perceived spectrum scarcity can mostly be attributed to inefficient spectrum management. The observed shift from exclusivelylicensed spectrum to sharing between unlike users has the potential to allow wireless communications services to continue growing at its current pace. However, increasingly complex sharing arrangements come with an increased risk of interference. This makes it necessary to address such events in a timely manner. At the same time, research into using machine learning for solving issues such as signal classification, decision-making processes, and anomaly detection in wireless communication has been growing. To support high quality machine learning research in anomaly detection for wireless communications, high quality IQ data are necessary. To this end, this research develops a data collection procedure that is able to produce high quality IQ data usable for machine learning as well as other wireless communications research. The collected high quality data is used to train three autoencoders for anomaly detection in shared spectrum: a Long ShortTerm Memory (LSTM), Variational, and a Deep Autoencoder. These three algorithms are used to successfully identify anomalies in the time and frequency domain of recorded IQ data in the form of unauthorized LTE transmissions on top of Wi-Fi communication.

Creator