When development work began on 6G, AI was identified as one of the critical technologies needed to enable the next generation of wireless communications. This was because AI has the potential to address some of the more complex technical challenges facing the industry. But like many new technologies, making AI work for mobile communications requires close collaboration across government, industry, business, and academia.
To assist businesses aiming to integrate AI into their operations, the Massachusetts state government initiated the AI Jumpstart pilot programme. This initiative provides funding for new computing infrastructure at Boston’s Northeastern University (NU), enabling collaboration between Northeastern’s research facilities, its researchers in various AI disciplines, and industry partners. Notably, Keysight Technologies, an electronic measurement company, has participated in the project from its inception following the funding announcement.
As a partnership between the state economic development agency Massachusetts Technology Collaborative and Northeastern University, the AI Jumpstart programme builds on a 15-year history of investment in university-based research and development projects that have effectively engaged businesses and industries. Through this programme, Keysight Technologies has formed a public-private partnership for collaborative AI research with Northeastern.
“The AI Jumpstart programme is an effective model of collaboration between industry, academia and the Mass Tech/John Adams Innovation Institute,” said Michael B. Silevitch, Northeastern University’s Principal Investigator of AI Jumpstart. “Our collaboration with Keysight provided a platform that demonstrated the viability of the approach.”
Advancing AI learning
Focusing on advancing 5G+ and 6G, Keysight and Northeastern collaborated to identify a specific technical challenge that AI could help solve: adjusting wireless communication systems for coexistence with radar signals.
“Keysight worked with Northeastern researchers and students to define the scope, scenarios, and data sets that will help optimise test techniques for wireless network deployments,” said Josep M. Jornet, Associate Director of NU Institute for the Wireless Internet of Things (WIoT). “The project invoked AI learning from a 5G data set generated in Northeastern University’s RF Colosseum and a representative radar defined through Keysight’s PathWave Software.”
Once the scenarios were designed, the high-priority radar signal created by Keysight was introduced into the RF Colosseum’s real-time LTE commercial spectrum scenario of Rome, Italy.
The result was an I/Q data set, subsequently employed in AI learning. The AI algorithm processed these I/Q samples, analysing throughput, power, and resource allocations in the 5G network. Selected examples of the data samples are presented below:
Tommaso Melodia, Director of NU Institute for the WIoT explains why this research is important: “Keysight and NU share common radio frequency/microwave teaching and testing goals for wireless research and industrial applications. Spectrum sharing is going to be a key technology for 5G and beyond, and AI learning was performed on I/Q data samples to enable spectrum sharing through optimal resource allocation to maximise throughput and minimise power consumption.”
Bringing AI learning to industry
Through the AI-Jumpstart analysis, Keysight is said to have gained a better understanding of the testing needs of wireless and defence developers as spectrum operations advance toward wider bandwidths, higher frequencies, and more complex environments. This understanding will help inform the continued development and refinement of Keysight’s test solution hardware and software, including the integration of AI and machine learning in future test applications.
“The AI learned patterns will assist Keysight in developing improved wideband, real-time capture analysis and closed-loop test applications,” said Roger Nichols, Keysight Technologies’ 6G Programme Manager. “These advancements will be instrumental in evolving our spectral management testing solutions, aiding designers in the swift and precise identification, classification, and prioritisation of signal and waveform characteristics.”
In addition, Nichols pointed out the advantages of early adoption of these AI test advances: it allows commercial and defence developers to conduct preliminary validations of new hardware, software/firmware, and signal processing algorithms. Such early testing is crucial for demonstrating and proving their performance in complex and evolving spectral sharing and coexistence environments.
Enabling student learning
While the primary focus of the AI Jumpstart project is on assisting businesses in deploying AI, the project has also been valuable in defining the training needed to student and industry engineers for this new technology.
“We launched this programme on behalf of Massachusetts to promote the adoption and integration of AI by companies in our state, by connecting them with the researchers at Northeastern,’ said Pat Larkin, Director, Innovation Institute at MassTech. “We’re encouraged by the progress on Keysight’s project, as it shows the direct impact of this research on advancing technology, and equally important, the relationship built with a talent development centre like Northeastern, where students gain real-world expertise working on projects like this.”
“It is highly beneficial to have industry partners involved in mmWave and 6G wireless technology testing. Teaching and research are advancing to optimise and define wireless network standards and performance,” said Carey Rappaport, Northeastern University AI Jumpstart Director. “Keysight’s core knowledge in RF and microwave measurement science, combined with NU’s THz specialists and testbeds, inspire students to pursue relevant classes and careers and support Massachusetts industry engineers with teaching and test techniques.”
For Nichols, this reflects Keysight’s approach to collaboration in advancing technology.
“We are committed to helping the growth of the radio frequency/microwave workforce by gaining a deeper understanding of the effects of software, firmware, and algorithm designs on communication link performance,” he said. “Through this ongoing collaborative research project, we are discovering how to use these technologies while equipping the next generation of engineers with the technical skills needed to make the next leap.”