Connect
Industrial edge computing and gateway integration
The edge is where industrial data becomes usable: where a register map becomes a named tag, where a thousand samples become one meaningful value, and where a dropped network link stops being data loss. We design, build and deploy the layer that sits between your machines and everything above them.
The problem
Sending everything to the cloud is the wrong default
Raw industrial data is high-volume, noisy and mostly uninteresting. Shipping all of it upstream costs bandwidth, costs storage, adds latency to anything that needs to react, and creates a hard dependency on a link that will eventually fail. The decisions that need to be fast — an alarm, an interlock, a local display — should not depend on a round trip to a data centre.
- 01
The dashboard goes blank whenever the plant's internet connection drops
- 02
Data gaps every time there is a power event or an ISP outage
- 03
Cloud costs scaling with sample rate rather than with value
- 04
Alarms arriving late because they are evaluated upstream
- 05
Every new machine brand requires a new point-to-point integration
What we build
Concrete deliverables
Nouns, not adjectives. This is what actually gets handed over.
- 01
Gateway architecture and hardware selection
DIN-rail industrial gateways or edge PCs specified for the environment — temperature range, DIN mounting, redundant power, watchdog, no moving parts.
- 02
Protocol conversion
Modbus RTU/TCP, OPC UA, Ethernet/IP, S7 and serial devices on the field side; MQTT, Sparkplug B, HTTPS or a database write on the north side.
- 03
Store-and-forward buffering
Local persistence sized to your worst realistic outage, with ordered back-fill and de-duplication on reconnect.
- 04
Edge processing and rules
Aggregation windows, deadbands, unit conversion, derived tags, state machines for run/idle/fault detection, and local alarm evaluation.
- 05
Local dashboards and HMI
Where operators need a screen that works with or without a network, we serve it from the edge device itself.
- 06
Remote management
Secure remote access, configuration versioning, over-the-air updates and health monitoring for the gateway fleet.
How it works
The technical path, step by step
Where the engineering decisions actually get made.
- 01
Field side
The gateway polls or subscribes to machines over their native protocols, on the OT network, in read-only mode.
- 02
Normalise
Registers and tags are mapped to a consistent model: named, typed, scaled, unit-tagged, and timestamped at the edge rather than on arrival.
- 03
Reduce
Deadbands, sample-rate policy per tag and aggregation windows cut volume by one to two orders of magnitude without losing anything anyone looks at.
- 04
Decide locally
State detection and alarm logic run on the device, so a fault is recognised in milliseconds regardless of upstream availability.
- 05
Publish and buffer
Data is published north over MQTT or HTTPS. On failure it queues to local storage and back-fills in order when the link returns.
Protocols & technologies
Specifics, because vague answers cost you money later
Chosen per site according to what is installed, not according to what we would prefer to work with.
| Technology | Where it is used | Engineering note |
|---|---|---|
| MQTT 3.1.1 / 5 | Primary north-bound transport | QoS 1 with persistent sessions; TLS with client certificates. |
| Sparkplug B | Structured MQTT payloads | Birth/death certificates give you positive knowledge that a device is alive, not just an absence of data. |
| OPC UA (client + server) | Both consuming and republishing | The gateway can act as an OPC UA server to expose a unified model to SCADA/MES. |
| Modbus TCP / RTU master | Field-side polling | Configurable poll groups per device so a slow device cannot stall a fast one. |
| S7 / Ethernet-IP / Profinet | Vendor-native reads | Chosen per site based on what the installed base actually supports. |
| REST / webhooks | Integration with business systems | For low-frequency, transactional data such as work orders and batch records. |
| Node-RED / custom runtime | Edge logic | Node-RED where the customer's team will maintain it; a compiled service where determinism matters. |
Brownfield
One gateway, many machine generations
The gateway exists precisely so that mixed-vendor, mixed-vintage plants can be treated as one system. A single edge device routinely handles a 1990s Modbus RTU serial link, a mid-2000s Ethernet PLC and a current-generation OPC UA controller simultaneously, and presents them upstream as one consistent model.
Works with
- Serial RS-232 / RS-485 devices via converters
- Any Modbus TCP or RTU capable device
- Siemens, Rockwell, Mitsubishi, Delta, Schneider, Omron controllers
- Energy meters from Schneider, Selec, L&T, Secure, Elmeasure
- Existing SCADA historians as a data source
Use cases
What people actually ask us for
Each one is a situation followed by the outcome it produces — not a feature list.
Unreliable plant connectivity
NowRural or industrial-estate sites with frequent link loss.
AfterLocal buffering means outages become a gap in transmission, not a gap in data.
High-frequency condition monitoring
NowVibration data at kHz rates that cannot be streamed.
AfterFFT and feature extraction at the edge; only the features go upstream.
Latency-sensitive alarms
NowA condition that needs a response in under a second.
AfterRules evaluated on the gateway, with the notification sent locally and logged upstream.
Protocol islands
NowSix machine brands, six incompatible integration paths.
AfterOne gateway model, one tag standard, one upstream interface.
Air-gapped or restricted networks
NowSecurity policy forbids OT devices reaching the internet.
AfterEdge collects and serves locally; a controlled, one-way path publishes a filtered subset if needed.
Implementation
How a project runs
Eight stages, each producing something you own. Full detail on the process page.
- 01
Discover
1–2 conversations
- 02
Audit
1–3 days on site
- 03
Design
1–2 weeks
- 04
Engineer
2–8 weeks per phase
- 05
Integrate
1–2 weeks
- 06
Deploy
Your shutdown window
- 07
Monitor
First 4–6 weeks
- 08
Optimise
Ongoing, if you want it
Mixed-vendor machine monitoring rig
Three PLC brands and one machine with no controller at all, read through one gateway and normalised into a single tag model — the situation almost every Indian plant is actually in.
Read the write-upQuestions
Edge Computing & Gateways — common questions
What hardware do you use for edge gateways?
It depends on the environment and the workload. For straightforward protocol conversion, a DIN-rail industrial gateway is sufficient. Where edge analytics, local dashboards or higher-frequency processing are needed, we specify a fanless industrial PC with adequate storage. We select for temperature range, DIN mounting, power tolerance and long-term availability rather than for cost alone.
Can the gateway sit on the OT network safely?
Yes, with proper segmentation. The standard arrangement is dual-NIC: one interface on the OT network for read-only machine access, one on a separate network for northbound publishing, with no routing between them. This follows the Purdue model and is normally the arrangement your IT and safety teams will require.
What happens during a power failure?
The buffer is persisted to non-volatile storage, so a power cut does not lose queued data. The gateway restarts unattended and resumes. Where the site has frequent power events, we specify a small UPS or a device with a supercapacitor-backed shutdown.
Do you use Node-RED?
Where it fits. It is excellent when the customer's own engineers will maintain and extend the flows, and it shortens delivery significantly. Where determinism, performance or long-term maintainability matter more, we write a compiled service instead. We tell you which we are proposing and why.
How do you update gateways once they are deployed?
Configuration is version-controlled and deployed remotely; firmware and application updates are staged and rolled out per site with rollback. Each gateway reports its own health, version and connection status, so a fleet of twenty is manageable by one person.
Bring us your protocol problem
Mixed vendors, serial equipment, an unreliable link, or a security policy that complicates everything. Describe the constraint and we will propose an edge design.