The comparisons of performance metrics be-tween cloud-aided and edge-only models provide a demonstration of advantages offered by a Cloud-Aided framework, particularly in regards to RNN performance which changed from 0% autonomy to 90% autonomy, outperforming convolutional neural networks when the domain shifted.
This paper benchmarks the performance trade-offs among fully onboard, cloud-based, and split-computing architectures for lightweight VLMs using SmolVLM-256M as a representative lightweight VLM and shows that no deployment strategy is universally optimal.
Zoha Azimi, Reza Farahani, S. Dustdar et al.· 0 citations
A risk-adaptive edge-cloud architecture in which onboard traffic assessment determines when cloud reasoning is requested is presented, in which onboard traffic assessment served as a practical trigger for selective VLM inference in these experiments.
Meng Ma, Shu-Yang Li, Nai-Gang Wang et al.· 0 citations
UAVCoNet is presented, a lightweight deep learning architecture optimized for real-time traffic occupancy estimation, congestion heatmap generation, and bottleneck analysis and can be integrated with intelligent transportation system for rapid assessment of traffic state and subsequently for real time traffic managemen...
Traffic management in emerging megacities requires real-time analytics over thousands of CCTV video streams under latency, bandwidth, compute and energy constraints. We present TrafficFab, an autonomic edge--cloud testbed for AI-driven traffic management, designed to validate a representative slice of a megacity deploy...
Mayank Arya, Pranjal Naman, Priyanshu Pansari et al.· 0 citations
This work presents GreenPipe, an automated profiling-training-validation pipeline that builds multi-resource regression models from external power meter measurements and attributes power to containers proportionally, exposing performance-energy trade-offs across workload configurations.
Meng-Xue Wang, Pei-Ni Liu, Amirhosein Taherkordi et al.· 0 citations
This paper proposes a spatiotemporal hypergraph-driven framework integrating high-order topological feature extraction with dynamic resource modeling, and introduces dynamic hypergraph sequences to naturally encompass local conflict domains, mitigating the topological blind spots and “over-smoothing” issues inherent in...
Kun Ding, Xi-Wen Qiu, Nian-Feng Weng et al.· Journal of Cloud Computing· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
