Cynthia Hood

Shared spectrum is a building block for 5G. Understanding how allocated spectrum is utilized in time and space is necessary to identify sharing opportunities. Analysis of spectrum measurements remains a largely manual process that requires significant domain knowledge. As a step toward automated analysis of spectrum measurements, this paper describes a case study where unsupervised learning is utilized to analyze wideband spectrum measurements at two different levels.

In response to presidential level directives to make additional spectrum available for commercial use, the United States National Telecommunication and Information Administration (NTIA) identified several government bands for consideration. One of the bands identified with potential for sharing with commercial users was the 4.2-4.4 GHz band. This paper examines the current use of this spectrum as a function of time and location and presents a methodology for evaluation of concepts for sharing between the primary and secondary users. The current incumbent use is aircraft Frequency Modulated Continuous Wave (FMCW) radar altimeters; whereas the proposed commercial secondary users are presumed to be small-cell commercial networks employing LTE technology. Since the performance of the radar altimeter is safety-of-life critical, a deep understanding of co-existence between these systems is necessary to evaluate whether sharing is feasible. This paper presents a method for evaluation of the potential for spectrum sharing between the ground-based LTE systems and commercial radar altimeters. The proposed harmful interference assessment strategy uses MATLAB ^TM and its associated LTE Toolbox along with aircraft location data obtained by monitoring the Automatic Dependent Surveillance Broadcast (ADS-B) signals which are available in real time from the airplanes to perform the assessment.

The constantly growing number of wireless devices and applications makes efficient spectrum utilization critical. Cognitive radio technology offers a solution by employing opportunistic and adaptive selection of transmission parameters and communication strategies. Due to the lack of general simulation tools available for analyzing cognitive radio networks, we propose an agent-based framework for cognitive radio studies. Unlike existing simulation tools, our design facilitates modeling of the physical environment along with the behavior of a network of cognitive radios. The inclusion of real-life spectrum occupancy data collected by the IIT Spectrum Observatory brings the modeled environment very close to reality. Additionally, because of its agent-based nature, our framework enables to study emergent and behavioral aspects of large and heterogeneous cognitive radio networks, an important factor that is being constantly neglected. Preliminary results show that presented solution is promising, but requires more development.