As slow and patchy internet connections remain a bane for many users, a new development from researchers at UC San Diego's Qualcomm Institute and Jacobs School of Engineering, in collaboration with Texas A&M University, could herald a significant enhancement to the cellular network experience. The project, dubbed "EdgeRIC," is an open-source AI software platform designed to dynamically refine and speed up internet performance over cellular networks by actively managing the distribution of resources based on real-time user demands. According to UC San Diego News Center, the development team boasts this could mark a substantial step forward in network intelligence.
Especially notable is the software's ability to quickly and effectively "take the temperature" of the cellular wireless environment and make immediate adjustments to improve performance. The current challenge for existing radio access networks (RAN, the main gateway for mobile internet, is to adapt quickly enough to rapid changes, which can currently take up to 10 milliseconds. However, EdgeRIC aims to significantly streamline this, reacting and redirecting resources within a swift 1-millisecond timeframe. "This is what EdgeRIC provides: application-aware mobile networks at every millisecond, so everything from our cars and robots which rely on the internet can be smart and safe as well”, Dinesh Bharadia, an affiliate of the Qualcomm Institute and department head of the WCSNG at Jacobs School of Engineering, elucidated, as per UC San Diego News Center.
The platform is not merely a theoretical construct; it was tested at the TAMU Innovation Proving Ground in various simulations involving drones, autonomous vehicles, and other dynamic endpoints. These tests demonstrated that EdgeRIC's microapplications, or μApps, consistently outperformed existing near real-time RIC systems by 5 to 25%, with a 30% improvement in user experience metrics such as video streaming quality. Woo-Hyun Ko, a senior research engineer at Texas A&M University, attested to the robust and optimized nature of the AI algorithms within EdgeRIC, following extensive real-world data collection for their development.
The EdgeRIC team has also thoughtfully incorporated features that allow other researchers to finetune the platform further. A crucial aspect is a built-in function that enables offline software training, enhancing EdgeRIC's capabilities to predict and preemptively address potential disruptions such as lag or dropped video calls. “At the end of the day, our ultimate goal is to satisfy the end user and better understand their needs,” Ushasi Ghosh, a UC San Diego doctoral student leading the study, said, per UC San Diego News Center. The researchers anticipate that continuing to refine the AI algorithms will result in a stronger, more intuitive product—the possibility of which was received with enthusiasm at the recent NSDI 2024 symposium.
