Text

theses

This work presents the results of multiple studies focused around spectrum usage efficiency improvements. The specified topic has been approached in several different ways, and considered approaches include coexistence of heterogeneous networks in the same spectrum, improving the efficiency of a network in the available spectrum, as well as dynamic spectrum allocation. The growing shortage in available wireless spectrum has resulted in opening of new frequency bands by the FCC. Some bands that have previously been underutilized are now opened for public usage under the 3-tier spectrum sharing framework. Spectrum access is regulated by a Spectrum Access System (SAS). The three user tiers, in order of priority, are incumbent users with guaranteed spectrum access, Priority Access License (PAL) holders with exclusive access to bands unused by incumbents, and General Authorized Access (GAA) users who access the spectrum opportunistically. The long duration of Priority licenses of 10 years, and the large geographical area the licenses cover, can lead to sub-optimal spectrum utilization. This work presents a market model for spectrum sub-leasing in underutilized areas and spectrum bands with short-term license allocation in smaller, focused geographical regions, which PAL holders can use to maximize profits. The paper published based on this work showed a basic sub-license allocation model along with two improved hypergraph based models which allow sub-license region adjustments or spectrum sharing and coexistence. The improved models resulted in the financial gain increase between 15 and 35\% over the base model, depending on the geographical distribution of the sub-lease bid regions. The increasing need for more spectrum access prompted DARPA (Defense Advanced Research Projects Agency of the US Department of Defense) to start the The Spectrum Collaboration Challenge (SC2) in 2016 to further expand research on spectrum usage efficiency, and mitigate the ever-growing problem of spectrum scarcity. Teams that participated in SC2 designed and developed wireless networks, called Collaborative Intelligent Radio networks (CIRNs), which used Artificial Intelligence and collaboration with other networks to coordinate spectrum usage schedules. To facilitate this collaboration, DARPA has established the CIRN Interaction Language (CIL) -- a language CIRNs can use to communicate with other networks and exchange relevant information to establish common spectrum goals and ways to achieve them. One of CIL's main functionalities was to enable teams to announce their intended spectrum usage and provide information other teams can use to adapt their own channel selection. While potentially a beneficial concept, CIL's effect on ensemble throughput of all networks was never evaluated. The paper published at DySPAN 2019 showed a simplified simulation of the spectrum usage announcement functionality of the CIL, explained the experiments with varying numbers of transmitter/receiver pairs which were run to evaluate CILs gains, and showcased an overall increase in spectrum utilization of 4 to 12\%, depending on the number of present spectrum users. Team SCATTER has been participating in the SC2 since its beginning in 2016. SCATTER's open-source software-defined physical layer (SCATTER PHY) has been developed as a standalone application, in cooperation with imec and the University of Ghent, with the ability to communicate with higher layers of SCATTER's system via ZeroMQ, and used USRP X310 software-defined radio devices to send and receive wireless signals. SCATTER PHY relied on USRP's ability to schedule timed commands, used both physical interfaces of the radio devices, utilized the radio's internal FPGA board to implement custom high-performance filtering blocks in order to increase its spectral efficiency as well as enable reliable usage of neighboring spectrum bands. The thesis describes the design and main features of SCATTER PHY and showcases the experiments performed to verify the achieved benefits. The experiments show the significant reduction of out-of-band transmission leakage achieved by FPGA-based FIR filtering, the achieved throughput of the implemented SCATTER PHY as a function of SNR and the bandwidth of the utilized channel, the attained computation complexity reduction of the improved frequency offset estimation, as well as the achieved packet reception rate. With many cognitive radio based spectrum access methods naturally applying reactive methods for detecting spectrum availability and their spectrum utilization, it is shown how a proactive approach which uses short-term forecasting to predict spectrum occupancy could help avoid the shortcomings of reactive methods, such as reaction delays which are dominant in scenarios with short bursts of a large amount of data traffic. As channel occupancy can be represented as a time series, the chapter affirms the emergence of Long Short-Term Memory (LSTM) neural network architecture as an efficient prediction mechanism compared to several traditional statistics-based predictions, describes the testing framework built for signal generation, data collection, prediction and evaluation, and finds the benefits of LSTM-based spectrum occupancy prediction, while explaining the scenarios in which different methods (such as collaborative methods based on CIL) would be advantageous, and how the two methods could be used in tandem to gain the benefits of both while avoiding all the shortcomings. With many available spectrum access and utilization efficiency approaches, operating on various layers of spectrum access methods, the thesis concludes by reiterating the benefits of each of the specified methods, and explains the gains of using them collectively and how this could shape future research.

http://dissertations.umi.com/gsnb.rutgers:11803

Ph.D.

Includes bibliographical references

doi:10.7282/t3-t7v9-k050

The author owns the copyright to this work.