Sentinel - Edge Threat Monitoring¶
Cloud Computing SS2026 • Frankfurt University of Applied Sciences • Prof. Dr. Christian Baun
An edge computing monitoring solution that detects people and identifies threats (fire, theft, vandalism) using a Raspberry Pi cluster with AI camera sensor nodes, distributed storage, and real-time alerting.
System Architecture¶
Tasks¶
| Task | Topic | Description |
|---|---|---|
| 1 | Infrastructure | Cluster setup, OS deployment, networking |
| 2 | HPL Benchmark | LINPACK performance measurement (GFLOPS) |
| 3 | MPI Scalability | MPI deployment, OSU benchmarks, scalability analysis |
| 4 | Amdahl & Gustafson | Non-MPI scaling laws with POV-Ray rendering |
| 5 | Monitoring | Prometheus + Grafana monitoring stack |
| 6 | Threat Detection | YOLO model training and object detection |
| 7 | Backend on k3s | Flask API, PostgreSQL, MinIO, Kubernetes |
| 8 | Frontend Dashboard | React SPA, REST APIs, deployment on k3s |
| 9 | Telegram Bot | Notification system with three alert sources |
Hardware¶
| Node | Hardware | IP Address | Role |
|---|---|---|---|
| Master | Raspberry Pi 5 (8 GB) | 192.168.137.10 |
k3s control plane, PostgreSQL, monitoring, registry |
| Sensor | Raspberry Pi 4 (4 GB) + AI Camera | 192.168.137.20 |
YOLO detection, MJPEG stream |
| Workers 1-8 | Raspberry Pi 3 Model B (1 GB) x 8 | .101 - .108 |
k3s agents, backend replicas, MinIO storage |
Quick Links¶
| Link | Description | |
|---|---|---|
http://192.168.137.10/ |
Live Dashboard | |
http://192.168.137.10:3000 |
Grafana Metrics | |
| github.com/mogheess/cloudcomputing | Source Repository | |
| System Overview | Full architecture & test guide |