Using artificial intelligence to hunt for slow zones in overcrowded networks
February 9, 2018
We’re all connected — and not just in a yogic sense.
By 2022, there will be 29 billion connected devices across the globe, according to a forecast from the June 2017 Ericsson mobility report. All of these devices will want a piece of the radio spectrum — the cluster of frequencies used by television, radio, and wireless signals — which will overcrowd radio bands.
Researchers like Lingjia Liu and Yang (Cindy) Yi, associate and assistant professors, respectively, in the Bradley Department of Electrical and Computer Engineering, are approaching the spectrum scarcity problem from various angles.
With techniques to tap unoccupied channels and improve spectrum efficiency Liu, Yi, and their collaborators are exploring new ways to meet the skyrocketing demand.
Minimizing interference and allocating spectrum
In the first, a $353,819 National Science Foundation (NSF) project titled Spatial Spectrum Sensing-Based Device-to-Device Networks, Liu seeks to manage wireless interference, which can mean dropped calls or poor connections, and to find the best way to allocate spectrum access in heavily used networks.
Liu is incorporating a technique called spatial spectrum sensing for device-to-device communication as a method to improve the overall efficiency of the spectrum. Spatial spectrum sensing allows devices to identify and use local empty spaces and periods of low traffic within the network bands. Short-range and local communication strategies like this help optimize spectrum and energy use.
Liu plans to develop a framework for analysis and design based on detection theory and stochastic geometry. After determining the best design and techniques to use, Liu’s team will evaluate the effects of this strategy on overall network performance.
“Direct device-to-device communications between user devices that offload cellular network traffic has the potential to be an integral part of the solution to address this mobile data challenge,” Liu said.
Tapping temporary availability for short-term communication
Many researchers have been exploring meeting wireless demand by increasing the availability of the current infrastructure. The most popular methods involve either sharing previously restricted bands of the radio spectrum using cognitive radio networks or enhancing wireless networks using expanded bandwidth, massive multiple-input multiple-output systems, and densified heterogeneous networks.
However, both of these methods have limitations and impact spectrum and energy efficiency, according to Liu. “Current hardware platforms present formidable challenges for supporting high-computational complexity and low-power consumption,” Liu said.
Liu and co-principal investigator Yi are leading a team developing a third option inspired, in part, by the human brain. They are working with collaborators from the University of Rhode Island on a new $700,000 NSF project, called Enabling Spectrum and Energy-Efficient Dynamic Spectrum Access Wireless Networks using Neuromorphic Computing. The team is designing a network architecture that allows devices to dynamically search for spectrum bands that are temporarily not at full capacity in their area, and use them for short-range communications.
Because dynamic spectrum access is so computationally complex, the proposed network’s computing devices will mimic the neurobiological architecture of the human brain — one of the most efficient and sophisticated systems in the known universe.
This new wireless network design shifts the focus from a centralized base-station-controlled approach to a more decentralized system, said Liu. In the new model, individual users will play stronger roles in spectrum access, changing the network topology by using smart computing devices.
“In this way, we will enable our nation's next-generation wireless network in an intelligent, spectrum-efficient, and energy-efficient dynamic spectrum environment,” Liu said.
Written by Kelly Izlar