Filippo Cugini

As part of our work on the Allegro project, we investigated the effects of various sequencing strategies by framing our multi-objective optimization challenge as a lexicographic optimization problem. 📊 In the analysis, we explored all permutations of three key...

As part of the Allegro Project, we explored how infrastructure layers are utilized under different optimization goals — with some insightful findings. 📊 Layer Utilization (Figure 32): Latency Optimization heavily favors the near-edge due to its proximity and...

As part of our work with Allegro, we evaluated performance across two hierarchical cloud-edge topologies: 📊 Topologies Analyzed: Basic: 19 nodes Extended: 53 nodesBoth consist of 3 layers: near-edge, far-edge, and cloud, with the cloud layer provisioned to handle all...

In the context of the ALLEGRO Project , we developed a powerful Multi-Agent Rollout Mechanism to further optimize microservice placement across distributed edge/cloud infrastructures. 🌍⚙️ 🧠 Why Rollout?Standard heuristics can be fast but suboptimal. Exact methods? Too...

As part of the #ALLEGRO project , we developed an intelligent Greedy Resource Allocation Mechanism for efficient microservice placement across edge/cloud infrastructure. 🌐⚙️ 🔍 What does it do?Our heuristic approach balances performance and resource efficiency by...

As digital infrastructure evolves, traditional monolithic applications are rapidly reaching their limits. With the rise of 5G/6G, virtualization, and optical networks, application design must adapt to meet increasing Quality of Service (QoS) demands. Enter the...

In our latest work on the Allegro project, we applied lexicographic optimization to tackle complex combinatorial problems involving conflicting objectives like cost, latency, and availability. Rather than optimizing all objectives at once, lexicographic optimization...

As part of our ongoing work in intelligent distributed storage, we evaluated the performance of our Deep Reinforcement Learning (DRL) mechanism during the inference stage, comparing it against an optimal solver across various optimization objectives. 🧠 Figure 26...

Before deploying our DRL-powered distributed storage allocation system, we invested in an intensive training phase to ensure our agent’s efficiency and stability. 🎯 By carefully tuning the hyperparameters α (learning rate), β (latency weight), and γ (discount factor),...

The ALLEGRO project introduces the MLFO, a dedicated orchestrator for managing distributed AI/ML pipelines across the network. Unlike traditional intent-based systems focused on single network entities, MLFO enables a global, scalable, and reconfigurable AI/ML...