Overview
GreenGuard is a real-time monitoring and analytics platform built for agricultural machinery manufacturers. Connected devices — processing machines installed at farms and facilities — transmit telemetry data every five seconds to the platform, which processes, stores, and visualises this data for operators and administrators.
The platform provides live equipment monitoring, historical analytics, configurable threshold alerts, and service request workflows — giving manufacturers and operators complete visibility into machine health and performance.
The Problem
Agricultural machinery operates in remote locations, often without on-site technical staff. Equipment failures that could be prevented with early warning instead cause costly unplanned downtime during critical harvest periods. Manufacturers had no visibility into how their machines performed in the field after sale.
The technical challenge was significant: devices configured by the manufacturer transmit data continuously, and the platform needs to ingest, process, and serve this data in real time to multiple concurrent operators — without introducing delays or data loss.
Solution
The system was designed around a high-frequency data ingestion pipeline with a clear separation between the write path (device telemetry) and the read path (operator dashboards and analytics). Real-time data is served to mobile and web clients via WebSockets, while the historical database is optimised for time-series queries with appropriate indexing strategies.
Alert thresholds are configurable per machine type and per parameter, allowing operators to receive notifications when values deviate from acceptable ranges — enabling proactive maintenance rather than reactive repair.
Architecture
- Device Ingestion: HTTP/MQTT endpoint receiving telemetry from manufacturer-configured IoT devices every 5 seconds — validated, processed, and persisted in real time
- Real-time Layer: WebSocket broadcasting to connected clients for live dashboard updates without polling
- Time-series Storage: MySQL with composite indexes optimised for device + time range queries across large historical datasets
- Analytics Engine: Aggregation queries for hourly, daily, and monthly summaries — pre-computed during off-peak periods for dashboard performance
- Alert System: Configurable threshold rules evaluated on each data point — triggers push notifications to operator mobile devices via Firebase
- Service Workflow: Operators can raise service requests from alert events, with assignment, status tracking, and resolution recording
- Mobile App: Flutter application for real-time monitoring, alert management, and service request handling in the field
- Admin Dashboard: Vue.js web interface for fleet-level monitoring, historical analytics, and alert configuration
- Infrastructure: Laravel backend deployed on Linux with Docker, configured for high write throughput
Technologies
Key Challenges
- High-frequency data ingestion: Handling continuous writes from multiple devices at 5-second intervals without write bottlenecks or data loss, requiring careful database configuration and write batching strategies
- Real-time delivery: Broadcasting live telemetry to multiple connected clients simultaneously without degrading ingestion performance — required careful separation of ingestion and broadcast paths
- Historical query performance: Queries across weeks or months of 5-second data points on a traditional RDBMS required composite indexing, query optimisation, and pre-aggregation strategies
- Alert reliability: Threshold evaluation had to be performed on every data point — missing an alert event has real operational consequences. Designed with explicit retry logic and audit logging
- Mobile offline handling: Field operators are often in areas with poor connectivity — the mobile app required offline alert queuing and sync-on-reconnect behaviour
Results
- Real-time telemetry ingestion and live dashboard for all connected machines
- Configurable threshold alerting with push notifications to field operators
- Historical analytics with drill-down from fleet level to individual device parameters
- Service request workflow from alert event to resolution with full audit trail
- Mobile application for field teams with offline capability
- Equipment monitoring enabling predictive maintenance rather than reactive repair