Can artificial intelligence be trained to improve wireless network performance, classify emerging devices on a network, or conduct real-time machine learning at the network edge? Those are just a few of the concepts being explored in new research programs funded by Intel and the National Science Foundation with a specific focus on future wireless systems, the latest in a series of joint research funding efforts.
Most of the 15 projects that are part of the Machine Learning for Wireless Networking Systems (MLWiNS) funding focus on the use of deep learning or neural networks, and range from explorations of AI and the role it could play in networks’ physical layer, to spectrum awareness, to how to train neural networks in a wireless environment.
The program description for the funding said that researchers are “expected to identify realistic problems that can be best solved by ML and to address fundamental questions about expected improvements from using ML over model-based methods.”
“5G and Beyond networks need to support throughput, density and latency requirements that are orders of magnitudes higher than what current wireless networks can support, and they also need to be secure and energy-efficient,” said Margaret Martonosi, assistant director for computer and information science and engineering at NSF. “The MLWiNS program was designed to stimulate novel machine learning research that can help meet these requirements – the awards announced today seek to apply innovative machine learning techniques to future wireless network designs to enable such advances and capabilities.”
The funding program description said that award sizes would range from $300,000 to $1.5 million for periods up to three years. MLWiNS is aimed at accelerating “fundamental, broad-based research on wireless-specific machine learning (ML) techniques, towards a new wireless system and architecture design, which can dynamically access shared spectrum, efficiently operate with limited radio and network resources, and scale to address the diverse and stringent quality-of-service requirements of future wireless applications” and that it also wants to target “reliable distributed ML by addressing the challenge of computation over wireless edge networks to enable ML for wireless and future applications.”
Among the funded programs:
-At Rice University, researchers will work to train large-scale centralized neural networks by separating them into a set of independent sub-networks that can be trained on different devices at the edge. This can reduce training time and complexity, while limiting the impact on model accuracy.
-Research teams from the Massachusetts Institute of Technology and Virginia Polytechnic Institute and State University will “explore the use of deep neural networks to address physical layer problems of a wireless network” and work to “develop new algorithms that can better address non-linear distortions and relax simplifying assumptions on the noise and impairments encountered in wireless networks.”
-At the University of California Irvine, researchers will work to develop ML methodologies to provide “reliable distributed computing in drone-infrastructure systems” with a “layer of intelligence located in individual drones” plus methods to compress data streams to be transferred for remote analysis.
-University of Notre Dame researcher will dive into the question of “sensor quality vs. quantity” when it comes to spectrum measurement and sensing.
The full list of funded projects is available here (pdf).
