As part of our work with Allegro, we evaluated performance across two hierarchical cloud-edge topologies:
📊 Topologies Analyzed:
- Basic: 19 nodes
- Extended: 53 nodes
Both consist of 3 layers: near-edge, far-edge, and cloud, with the cloud layer provisioned to handle all workloads.
🧠 Evaluation Methods:
We compared our custom heuristic approach and multi-agent Rollout algorithm with MATLAB’s built-in MILP solver. Due to high computational cost with MILP on larger setups, the basic topology was used for fair comparison.
⚙️ Scaling for Realism:
The extended topology mirrors the node attributes of the basic but scales up to reflect realistic deployments:
- 40 near-edge nodes
- 10 far-edge nodes
- 3 cloud locations
This evaluation lays the groundwork for scalable, efficient orchestration in distributed AI systems at the edge. 🌐📶
#AI #EdgeComputing #CloudComputing #MultiAgentSystems #Optimization #PerformanceEvaluation #AllegroProject #Research #DistributedSystems
