Protocol diversity
Products and sites used different field interfaces, data models, and timing expectations.
SaiAgrotel
Industrial IoT Platform
An industrial automation platform connecting embedded controllers, edge gateways, and cloud services for dependable remote operation.
Embedded CESP32LinuxMQTTModbusEdge Gateway
Case Study
Overview
The platform connected field devices and industrial controllers to gateway and cloud services across variable network conditions.
Its design prioritized protocol adaptation, offline operation, secure fleet management, and maintainable deployment.
- Domain
- Industrial Automation
- Connectivity
- MQTT & Field Protocols
- Focus
- Edge-to-Cloud Reliability
- Lifecycle
- Platform Delivery
Context
Problem Statement
Unify heterogeneous controllers and field protocols behind a consistent device and cloud model.
Maintain safe local operation and trustworthy data delivery when internet connectivity is intermittent.
Leadership
My Role
- Led firmware, hardware, gateway, and cloud engineering teams.
- Defined platform boundaries, interfaces, and delivery priorities.
- Reviewed device-management, telemetry, diagnostics, and update strategies.
- Coordinated field feedback with product and engineering roadmaps.
Platform
Technology Stack
Core technologies and engineering disciplines used across the product.
- Embedded C
- ESP32
- Linux
- MQTT
- Modbus
- Edge Gateway
- Cloud APIs
- OTA
- Telemetry
System Design
Architecture
Controllers retain local responsibility while an edge gateway normalizes protocols, buffers telemetry, manages devices, and connects securely to cloud services.
Architecture overview
- Field Sensors
- Embedded Controllers
- Protocol Adapters
- Linux Gateway
- MQTT Services
- Cloud Applications
Execution
Engineering Challenges
Intermittent networks
The system needed to preserve local operation and reconcile data after extended outages.
Fleet maintenance
Diagnostics and updates had to work across devices with different capabilities and deployment histories.
Trade-offs
Key Decisions
Edge autonomy
Kept essential control and fallback behavior local so cloud availability was not a safety dependency.
Canonical data model
Normalized device data at the gateway instead of exposing protocol-specific details to every cloud service.
Managed update lifecycle
Designed staged deployment, version visibility, and recovery into the fleet-management approach.
Impact
Results
- Created a reusable connection model across multiple industrial solutions.
- Improved resilience through local control and buffered edge operation.
- Reduced cloud integration complexity with normalized device data.
- Made deployed-system diagnostics and version state more visible.
Reflection
Lessons Learned
- Industrial IoT architectures should assume the cloud will sometimes be unavailable.
- A canonical data model is essential when product families span many protocols.
- Field diagnostics need to be designed before fleets reach production scale.