Data & IoT Platform

Industrial and commercial IoT deployments are growing faster than the data infrastructure beneath them. A modern smart-factory line can generate upward of 50,000 sensor events per minute; cold-chain fleets crossing multiple jurisdictions produce GPS, temperature, and humidity readings continuously. Without deliberate pipeline design, that volume either overwhelms on-premise brokers or accumulates in cloud storage silos that nobody queries.

The design decisions that determine whether a deployment scales cleanly — ingest protocol selection, edge pre-processing, stream vs. batch boundaries, storage tiering, and activation latency targets — need to be made before a single device is provisioned. Retrofitting them later is expensive and usually requires a re-platform of both firmware and backend.

What is the right MQTT / HTTP ingest pattern for your device fleet?

Protocol choice is not a vendor preference — it is an engineering constraint driven by device power budget, network reliability, and message frequency. MQTT over a persistent TCP session is the right default for constrained devices on cellular or Wi-Fi; HTTPS works for low-frequency telemetry where connection overhead is acceptable; CoAP suits UDP-only networks or NB-IoT links; LoRaWAN covers long-range, low-power assets like footfall counters in retail spaces or field sensors on agricultural land. A mixed fleet — which is the reality for most enterprise deployments — needs a broker layer that normalises all four onto a common schema before any downstream system touches the data.

Common ingest protocols we configure for enterprise fleets:

• MQTT 3.1.1 / 5.0 — persistent session, QoS 0/1/2, ideal for manufacturing PLCs and fleet telematics ECUs
• HTTPS REST / webhook — stateless, firewall-friendly, suited for periodic cold-chain logger uploads
• CoAP over UDP — low overhead for NB-IoT and LTE-M assets with constrained radio duty cycles
• LoRaWAN — kilometre-range coverage for EEC smart-factory perimeter sensors and rural asset tracking

How do you handle edge-to-cloud data synchronisation?

Connectivity interruptions are not edge cases in industrial deployments — they are routine. A fleet truck crossing a mountainous corridor, a packaging line in a steel-structure factory, or a cold-room whose Wi-Fi AP reboots during a power dip will all lose uplink periodically. The edge layer must buffer, deduplicate, and replay events with sequence guarantees. We deploy lightweight edge runtimes — either containerised on industrial PCs or running on purpose-built gateways — that maintain a local write-ahead log and sync to cloud brokers on reconnection without duplicating events or losing ordering within a device session.

The pipeline layers that matter are ingestion, stream processing, storage, and activation — get the boundaries wrong between any two and you pay for it in latency, duplicates, or missed alerts at production scale.

What does PDPA compliance look like for IoT data?

IoT data sits in a compliance grey zone that catches organisations off guard. Retail footfall sensors that capture device MAC addresses, fleet telematics that record driver behaviour, and building management systems that log employee badge proximity all process personal data under Thailand's PDPA and equivalent frameworks across ASEAN. The obligations are the same as for structured enterprise data: lawful basis, data minimisation, retention limits, subject access rights, and an Audit Trail for cross-border transfers to cloud regions outside Thailand. Compliance is not a post-deployment concern — data classification and masking must be built into the Capture Layer and enforced before events reach the stream processor or any downstream storage.

From raw telemetry to operational insight

For enterprise clients that need to unify IoT streams with ERP, WMS, and CRM data, the pipeline feeds directly into our Enterprise Data Platform — a composable data fabric layer that handles semantic unification, access governance, and query federation across sources. Deployments we have shipped include smart-manufacturing quality control for EEC-zone plants, real-time cold-chain exceedance alerting for a regional logistics operator, and fleet telematics dashboards for a Thai petroleum distributor. If you are planning a new IoT rollout or rearchitecting an existing one, speak with our data engineering team to scope the ingest architecture before devices go into production.