“Closing the loop†is a common business buzzword, but have you ever stopped to think about the origin of the phrase? It's rooted in the field of electrical engineering and describes circuit components that are connected and in synch. “Closing the loop†is the inspiration behind a joint research effort to improve how wireless communication devices find and access open spectrum, which is important because devices are competing for connectivity. Cong Shen, assistant professor of electrical and computer engineering at the University of Virginia,Carl Dietrich, research associate professor of electrical and computer engineering at Virginia Tech, andNicholas Polys, director of the visual computing group at Virginia Tech, are co-principal investigators of the project funded by theCommonwealth Cyber Initiative. The initiative is a multimillion-dollar effort by Virginia leaders to create a statewide ecosystem of excellence in cyber-physical systems, and UVA is one of the participating universities. Shen, Dietrich and Polys are developing a closed-loop system for digital and telecommunications networks that will make devices' search for open spectrum more efficient, reducing delays or “hang time†and improving the reliability of data transmission. Two signaling processes are at play. In the first process, “listen-before-talk,†the sending device needs to listen in on the spectrum band it wants to use to ensure the channel is open before it sends a signal. The second process focuses on the acknowledgement or “send receipt†for the data. The sending device continues transmission at set intervals, governed by an automatic repeat request protocol, until it gets a return signal confirming the data was successfully transmitted. Both processes involve a lot of trial and error. Shen and Dietrich posit that connecting these two processes will enable the device to learn from its past tries to successfully access the wireless spectrum more quickly. “The advances in online learning theory and algorithms are what make this ‘closed loop' possible,†Shen said. “A network needs to satisfy a lot of requirements, such as high data rates or ultra-low latency. This complexity poses a key challenge to putting theory and algorithms to work in a hybrid design.†Shen envisions new tools from machine learning to manage system complexity and rethink system design, with a focus on advancing future communication and networking systems. Shen leads theLaboratory for Intelligent Communications and Networkingat UVA Engineering, contributing to the Department of Electrical and Computer Engineering's research strength inmachine learning, signal and image processing and communications. Specific to the Commonwealth Cyber Initiative challenge, Shen is developing novel machine learning methods that enable a device to learn from its past actions and observations--which channel has been used successfully and unsuccessfully—to intelligently select its next action. Combining machine learning, wireless communications, and advanced visualization methods, and going from theory to practice are unique aspects of this work, and the team formed by Shen, Dietrich, and Polys brings complementary and balanced expertise to the project. By early 2022, the researchers expect to develop a prototype with newly designed algorithms. “A successful demonstration is a significant step toward an open-source and multi-dimensional spectrum access system that covers a wide array of bands that are of practical interest,†Dietrich said. Dietrich envisions a multi-dimensional spectrum access system, leveraging Virginia Tech's open-source spectrum access system and a dynamic spectrum access approach developed by Virginia Tech research assistant professorVijay Shah. The team will use Virginia Tech'sCognitive Radio Network testbedto demonstrate the spectrum-sensing algorithms. They also plan to enhance Virginia Tech's open-source spectrum access system, an automatic spectrum management system that supports a centralized approach to frequency assignment that can be used in combination with spectrum sensing by individual spectrum-sharing radios. “When assigning frequencies for communication systems to use for short periods of time, the multi-dimensional spectrum access system will make its decision based on the context, the band, and the specific type of communication traffic such as voice, streaming video, or text message,†Dietrich said. Contextual understanding enables the system to prioritize different types of users, such as military radars and public safety, commercial, or private communication systems. Special events, emergencies and weather alerts, for example, affect different frequency bands in distinct ways. Polys and his team will contribute software engineering and development expertise as well as novel software that lets researchers and others visualize operation of the spectrum access system and spectrum-sharing wireless devices. "Visualization is crucial for understanding our radio spectrum; this collaboration will enable new views and management techniques for this fundamental public asset," Polys said. “This research will build the capacity of Commonwealth research efforts in spectrum engineering,†Shen said. “We look forward to piloting the development of the multidimensional spectrum access system and its companion software to simulate and measure network performance.â€