Abstract: Edge computing confronts a massive data surge from edge-native applications like AR/VR, autonomous vehicles, and remote healthcare. The inherent limitations at the edge such as resource heterogeneity, network instability and inelasticity, make it difficult to deploy traditional cloud solutions to manage this data surge. The insight that application awareness is fundamental to handling resources for edge-native applications is the core of my research focus. Towards this, I have introduced four complementary data management techniques: (1) Cargo, a fault tolerant, SLO performance aware storage layer coupled with an edge computing framework (Armada) for heterogeneous edge environments; (2) A novel spatial-awareness based data placement and dynamic storage redundancy strategy for faster data delivery for edge-native applications; (3) ASTRA, a prefetching framework for mobile AR that leverages object associations, AR specific features, spatial/temporal relevance to achieve high cache hit rates; (4) Viveka, a context aware sensing framework for improving energy efficiency and reducing data generation on smart wearables while ensuring minimal impact on application accuracy. My immediate work will focus on extending the Viveka framework for remote health monitoring and exploring strategies to improve data imputation for data sensed from IoT devices. Looking towards the future, I plan to use the techniques identified in the current research to improve energy efficiency for autonomous vehicles, which can be considered a cluster of sensors.

About the Speaker: Nikhil Sreekumar is a Systems Scientist at Qen Labs Inc, USA. His research focuses on energy efficiency and data management, specifically how to use context awareness to manage the data deluge and improve the energy efficiency of smart wearables at the network's edge while ensuring application SLO requirements are met. He earned his PhD in Computer Science from the University of Minnesota, USA, under the guidance of Dr. Abhishek Chandra and Dr. Jon Weissman. His research explored application-aware data management at the edge. Nikhil's work has been published in leading conferences such as ACM SEC, IEEE ICDCS, IEEE IPDPS, IEEE IC2E, IEEE Big Data and has received support from the NSF and the US DoD. His work on AR object prefetching, ASTRA, was awarded the Best Research Paper at IEEE IC2E 2025. Prior to his current role, Nikhil worked as a Software Developer at MathWorks, where he contributed to the HDL and MATLAB Coder infrastructure. His current interests include Energy efficient smart wearables for remote health monitoring, Data management for edge-native applications and HPC sanitization for AI workloads.