Real-Time Telemetry — IoT Data Streaming & Analytics

Real-time telemetry is the circulatory system of an IoT platform. Every sensor reading, every device state change, every anomaly event generated at the edge must travel from device firmware through the network to cloud storage and on to dashboards and alert engines with minimal latency and without data loss. At FSS Technology, we have architected and operated real-time telemetry pipelines processing millions of messages per day from industrial, marine, and smart building device fleets. Our telemetry architectures combine Azure IoT Hub’s device connectivity, stream processing, time-series databases, and WebSocket delivery to dashboards into end-to-end pipelines where sensor events appear on operator screens within seconds of occurring at the physical device.

Telemetry Pipeline Architecture

Ingestion: Azure IoT Hub and Event Hubs

FSS telemetry pipelines use Azure IoT Hub as the device-to-cloud message entry point. IoT Hub authenticates devices via X.509 certificates or SAS tokens, enforces per-device rate limits, and routes telemetry to Azure Event Hubs partitions for scalable stream consumption. IoT Hub’s device twin synchronisation enables cloud-to-device state management alongside telemetry ingestion — the same connection handles both directions. For high-throughput non-IoT data sources, FSS uses Azure Event Hubs directly, with Kafka protocol compatibility enabling producer libraries from Python, Node.js, and .NET without Azure-specific SDKs.

Stream Processing: Azure Stream Analytics and Node.js

Raw device telemetry requires transformation before storage: unit conversion, calibration correction, outlier filtering, and field normalisation. FSS implements stream processing in Azure Stream Analytics for declarative SQL-based transformations — windowed aggregations, pattern detection, multi-stream joins — and Node.js stream processors for complex business logic that exceeds Azure Stream Analytics’s SQL expressiveness. Processed telemetry flows to multiple sinks: a time-series database for historical queries, a real-time cache for dashboard current-state displays, and an alert engine that evaluates threshold conditions.

Time-Series Storage: InfluxDB and TimescaleDB

Time-series databases are optimised for the write-heavy, timestamp-keyed workloads of IoT telemetry that overwhelm relational databases at scale. FSS uses InfluxDB for pure time-series workloads and TimescaleDB (PostgreSQL extension) for applications that combine time-series telemetry with relational data — device metadata, user accounts, alert configurations — in a single database. Both databases provide efficient downsampling, retention policies that automatically expire old high-resolution data, and time-series query functions (moving averages, rate calculations, gap filling) that would require complex SQL on relational databases.

Dashboard Delivery: WebSockets and SSE

React dashboards consume real-time telemetry via WebSocket connections from FSS Node.js backends or Azure SignalR Service. The backend receives new telemetry from the processing pipeline and immediately pushes it to all connected dashboard clients — no polling required. Dashboard charts update in under 500ms from sensor event occurrence, providing operators with current-state situational awareness for monitoring applications. For lower-throughput alerting, Server-Sent Events (SSE) provide a simpler unidirectional push channel with automatic browser reconnection.

Anomaly Detection and Alerting

FSS telemetry platforms integrate configurable alert engines that evaluate incoming sensor data against threshold rules, rate-of-change conditions, and multi-sensor correlation rules. Threshold crossing alerts — temperature above 85°C, pressure below 2 bar — trigger notification workflows: email, SMS, webhook, and in-app notifications to operator dashboards. Statistical anomaly detection using Azure Stream Analytics window functions flags readings that deviate significantly from rolling statistical baselines, detecting gradual sensor drift and unusual operating patterns that fixed-threshold alerts would miss.

FSS Telemetry Expertise

FSS Technology has designed and operated real-time telemetry pipelines since 2017, scaling from 10-device prototypes to production fleets of thousands of devices. Our team has navigated Azure IoT Hub quota management, time-series database query optimisation, and WebSocket scalability challenges that only emerge in production at scale. Contact FSS to discuss your IoT telemetry architecture.