CS&E Research Team Wins Best Paper Award at the IEEE IC2E Conference
Department of Computer Science & Engineering (CS&E) PhD student Nikhil Sreekumar and Professors Abhishek Chandra and Jon Weissman received the Best Paper Award at the IEEE International Conference on Cloud Engineering (IC2E) for their paper titled, “ASTRA: Association, Spatial proximity and Temporal Relevance based Adaptive prefetching for Edge AR.”
Sreekumar, Chandra, and Weissman’s paper focused on augmented reality (AR) applications. The virtual objects or images that appear in a user's AR environment are either stored locally on the headset or retrieved from the cloud. This process results in a lot of time and delays when data needs to be downloaded or fetched over the network. As a result, users may experience delays before the AR objects are displayed, which can disrupt the interaction flow and lead to a less engaging user experience.
“We are trying to tackle this challenge by using Edge Computing Paradigm (ECP),” Sreekumar said. “This puts data closer to the user rather than putting it onto the cloud. Within the ECP, we added extra optimizations that do prefetching based on the user's interest, behavior, and location. This allowed us to improve the latency associated with the objects rendering onto the AR headset, or even the mobile device on which the AR system is running.”
Augmented reality is a newly emerging application. When looking at existing works in this field, there are some limitations. For example, many AR applications feature small virtual objects, which limits the interactivity of AR environments.
“Augmented reality applications are now being used in many different domains,” Chandra said. “A lot of new devices are coming out that help you get richer information about your environment and about things that you’re interacting with. We found a lot of problems in existing approaches and existing state-of-the-art devices. That is what got us interested in trying to solve some of these problems more systematically.”
The next phase of the project is to implement the ECP platform in a real-world setting to see how the system performs in different scenarios. The work currently is only in a simulation-based environment.
“The specific technique that we used in mobile environments specifically works with users who are interacting with the environment and the context around them,” Chandra said. “A lot of it is dependent on their location, time of the day, and their inherent preferences. These insights are applicable in many other application domains. We are looking at extending this work to other areas, which include mobile computing in the healthcare domain.”
The 2025 international conference was hosted in Rennes, France, and provided a comprehensive forum where researchers can exchange practical experiences related to cloud engineering. ASTRA got accepted as the best paper for its novel techniques for prefetching augmented reality objects. This unique technique has never been done before and bridges together some of the ideas from different communities, data mining, and systems.