Open Source IoT Platform
An IoT platform's real test is time: sensors and controllers deployed to the field can run for a decade, far longer than the average cloud service or its pricing stays put, and hardware you cannot easily recall gets stranded when a vendor goes dark. Running an open source platform yourself changes that horizon. Telemetry lands in a database you own, devices connect to an endpoint you host, and the protocols the fleet speaks stay under your control rather than on a roadmap you do not set. That ownership matters whether you are automating a single house on local hardware or operating thousands of remote units, though those two jobs pull toward very different tools.
Home Assistant
Open source home automation with local control and privacy first
ThingsBoard
IoT platform that connects devices, collects telemetry, processes it with rule chains, and visualizes it on dashboards
Domoticz
Free open source home automation system for monitoring and controlling lights, switches, and sensors
Gladys Assistant
Privacy-first open-source smart home assistant for local automation and device control
Magistrala
IoT platform framework with multi-protocol messaging, device management, and fine-grained access control
Kuzzle
Self-hostable backend for IoT networks and apps with real-time pub/sub, stored data, and built-in search
EdgeX Foundry
Vendor-neutral edge framework that connects IoT devices and sensors to applications through interoperable microservices
ioBroker
Open source IoT platform for home automation, data logging, and system integration on premise
openHAB
Vendor- and technology-agnostic open source home automation software
Our picks
These split into two jobs, a home hub or a fleet backend, so match the pick to your case.
Local-first home automation: Home Assistant Home Assistant runs on hardware you own, from a Raspberry Pi to a dedicated server, putting local control and privacy first across more than 1500 device integrations. Its modular architecture lets you add devices or write custom components, and it is backed by the non-profit Open Home Foundation, which cannot be sold or acquired.
Full-stack device fleet platform: ThingsBoard ThingsBoard takes devices from connection to insight in one place: it provisions hardware, ingests telemetry over MQTT, CoAP, HTTP, LwM2M, and SNMP, and processes it through rule chains that raise alarms. Built-in, custom, and SCADA dashboards visualize the data, multi-tenancy isolates customers, and it self-hosts or runs as ThingsBoard Cloud under Apache 2.0.
Multi-protocol IoT control plane: Magistrala Magistrala gives you one control plane for devices, users, and data instead of stitching together a broker, an auth layer, and a rule engine. It handles MQTT, HTTP, WebSocket, and CoAP as first-class protocols under one identity model, with ABAC and RBAC across multi-tenant domains, plus a rules engine and audit logging.
Vendor-neutral edge framework: EdgeX Foundry EdgeX Foundry sits between physical devices and the cloud as loosely coupled Go microservices: device services talk to hardware over many protocols, core services store and act on readings, and application services export data outward. It communicates over a message bus, closes external ports with security on by default, and is governed under LF Edge.
Deciding which IoT job you are actually solving
Two very different products hide under the IoT platform label, and this list holds both. One is the home and building hub, built to integrate hundreds of consumer protocols on local hardware with privacy as the point; Home Assistant, openHAB, Domoticz, Gladys Assistant, and ioBroker all live here. The other is the fleet platform, built to provision devices, ingest telemetry at scale, and drive rules across many tenants, which is the territory of ThingsBoard, Magistrala, Kuzzle, and EdgeX Foundry. Deciding which you need narrows the field faster than any feature grid, because a great smart-home hub makes a poor multi-tenant backend, and vice versa.
For a fleet platform, the connection model is the operating boundary rather than a checkbox. A system centered on MQTT behaves differently from one built around HTTP APIs, CoAP, or gateway ingestion; ThingsBoard speaks MQTT, CoAP, HTTP, LwM2M, and SNMP, while Magistrala treats MQTT, HTTP, WebSocket, and CoAP as first-class protocols under one authentication model. Look at quality of service, retained messages, command delivery, and how the broker copes when a power cut sends a large share of devices reconnecting at once. For fleets that live in the field for years, per-device identity, credential rotation, and firmware update mechanics matter more than any dashboard.
Then decide where telemetry becomes operational data. Some platforms are strong at ingestion and leave storage and analytics to systems you already run: Kuzzle persists to Elasticsearch with built-in search, and EdgeX Foundry pushes readings outward through application microservices at the edge. Favor a clear data model, documented APIs, and straightforward export over a polished chart builder, and if you need local operation, verify buffering and rule execution during outages. Device registries, credentials, payload schemas, and historical telemetry should never be trapped in an opaque internal format you cannot rebuild elsewhere.
Related categories
Frequently asked questions
What is the difference between a home automation hub and an IoT fleet platform?+
A home hub like Home Assistant or openHAB integrates hundreds of consumer devices and protocols on local hardware, aimed at one household or building with privacy as the goal. A fleet platform like ThingsBoard or Magistrala is built to provision many devices, ingest telemetry at scale, isolate tenants, and run rules and alarms. They overlap on dashboards but solve different problems, and picking the wrong class is the most common early mistake.
Which protocols should an open source IoT platform support?+
Match protocol support to your devices, not to a feature matrix. MQTT is common for telemetry and commands, HTTP suits simple integrations, CoAP appears in constrained environments, and LwM2M shows up in managed cellular fleets. ThingsBoard covers MQTT, CoAP, HTTP, LwM2M, and SNMP, while Magistrala adds WebSocket. Also check quality of service, payload limits, and how the platform handles many devices reconnecting together after an outage.
How do these platforms handle devices that go offline?+
The better platforms assume devices will vanish. Look for queued commands, device shadows or twins, message expiry, local buffering at the edge, and reconciliation when a gateway reconnects. Sleeping sensors need different semantics from always-on machines, and time-sensitive commands should be able to expire rather than execute hours late. For cellular or remote deployments, this offline behavior often matters more than anything on the live dashboard.
What should I expect from OTA firmware update support?+
Treat over-the-air updates as a controlled deployment system, not a file push. Useful features include staged rollout, targeting by model or firmware version, progress reporting, failure capture, and a rollback strategy. Some platforms only track update metadata while your own device agent performs the install, which can be fine, but be clear about which side verifies signatures, resumes interrupted downloads, and survives a power loss mid-update.
How should device provisioning and identity work?+
Aim for per-device identity, not one shared secret across the fleet. Good provisioning supports unique credentials, certificate-based auth where appropriate, rotation, and revocation, plus a way to bind a physical unit to a customer, site, or tenant. Magistrala, for example, governs access with mutual TLS, JWT, and per-device tokens under ABAC and RBAC. Bulk import matters for manufacturing, and clean recovery matters when a device is returned, resold, or compromised.
Can I run one of these fully locally with no cloud?+
Yes, and for the home hubs that is the whole point. Home Assistant, openHAB, Domoticz, and Gladys Assistant run on your own hardware, talk directly to local devices, and keep automation working with no cloud account. Home Assistant is governed by the non-profit Open Home Foundation and runs on anything from a Raspberry Pi to a dedicated server, so your data and control never depend on an outside service.
Where does my device data actually live?+
It depends on the platform's architecture. Telemetry may land in the platform's own database, a time-series store, a message queue, or a downstream analytics system. Kuzzle stores data in Elasticsearch, while EdgeX Foundry is designed to export readings outward to cloud and enterprise systems. Ask what format raw payloads are kept in, how retention is controlled, and whether device registry data and credentials metadata are exportable alongside the telemetry.