Skip to content

Chapter 2: Use Case Architecture

Three Core Use Cases with Multi-Source Validation

This chapter details the architecture for three Edge AI use cases deployed at Mumbai and Chennai hub sites. Each use case demonstrates Edge AI + Observability Fusion with real-time multi-source validation.

Use Case 1: Intelligent Physical Security

2.1.1 Security Functions & AI Models

Six security functions: perimeter intrusion, loitering detection, tailgating, license plate recognition, crowd density, and access control correlation.

2.1.2 Camera Deployment & Network Architecture

Complete camera deployment design (135 cameras per site), network infrastructure, bandwidth calculations, and hardware specifications (UCS XE9305 + NVIDIA L4).

2.1.3 Observability Integration Timeline

500ms end-to-end detection timeline with millisecond precision: frame capture, GPU inference, multi-source validation (ISE, Splunk, TE, AppD), and automated action.

Use Case 2: Building Automation

2.2 Smart Building Optimization

HVAC and lighting optimization using occupancy detection, BMS API integration, and AgenticOps WF-009 automated workflows. Target: 15-20% energy savings.

Use Case 3: Safety & Compliance

2.3 Safety & Compliance Monitoring

PPE detection (hard hat, safety vest), fire/smoke detection using thermal cameras, and slip/fall detection with pose estimation.

Cross-Use Case Architecture

2.4 Infrastructure Sharing & GPU Utilization

How all three use cases share the same Edge AI platform (UCS XE9305 + NVIDIA L4), GPU utilization analysis (80-95%), and multi-pipeline processing.

Architecture Highlights

  • AI Models: YOLO v8n, DeepSORT, custom CNNs (PPE detection), OpenPose (fall detection)
  • GPU Platform: NVIDIA L4 24GB (120 TOPS INT8, 72W TDP, 20ms inference time)
  • Network: 960 Mbps per site, VLAN 150, SGT-70, PoE+ 3,000W budget
  • Integration: Real-time API calls to ISE, Splunk, ThousandEyes, AppDynamics, BMS
  • Performance: <500ms end-to-end latency, <5% false positive rate

Previous: Chapter 1: Executive Summary | Next: Chapter 3: Platform Architecture