Our paper "Characterizing Distributed Mobile Augmented Reality Applications at the Edge" has been accepted at ACM CoNEXT 2023 with a highly competitive 19.3% acceptance rate! This work provides the first comprehensive characterization of how distributed mobile AR applications perform when deployed at the edge, examining both the unique challenges and opportunities that edge computing brings to AR workloads.
Mobile AR applications have stringent requirements including ultra-low latency (20ms or less), high throughput for processing video frames in real-time, and compute-intensive operations like object recognition and tracking. Our study deployed real AR applications on edge infrastructure across diverse scenarios (gaming, navigation, industrial) and varying network conditions to understand performance characteristics, resource requirements, scalability patterns, and best deployment practices.
We're proud that this paper was awarded all three ACM artifact reproducibility badges (Available, Functional, and Reproduced), demonstrating our commitment to reproducible research and open science.
Artifacts Available
Our findings inform application developers designing edge-aware AR apps, edge operators provisioning resources, platform designers building AR-capable systems, and researchers exploring optimization opportunities. The methodology and insights also apply to other latency-sensitive edge applications like autonomous vehicles, industrial automation, remote surgery, and real-time video analytics.
Conference: ACM CoNEXT 2023
Acceptance Rate: 19.3%
Related: