IoT Data Analytics (2026): Architecture, Time-Series, Streaming, ML & Tools

1. Concepts and Reference Architectures

10 entries

What IoT data analytics is and how data flows from devices to decisions. IoT data is high-volume, time-stamped and often noisy; architecture choices follow from that.

01Internet of thingsOverview of IoT devices, connectivity and applications. 02Predictive maintenanceBackground on condition-based and predictive maintenance. 03Anomaly detectionOverview of detecting unusual patterns in data. 04AWS IoT Lens (Well-Architected)AWS guidance on IoT workload design, including data and analytics. 05Azure IoT Reference ArchitectureMicrosoft reference architecture for IoT solutions. 06ISO/IEC 30141: IoT Reference ArchitectureInternational reference architecture for IoT systems. 07Edge Intelligence: Paving the Last Mile of AI with Edge ComputingSurvey of analytics and AI at the edge. 08Designing Data-Intensive ApplicationsKleppmann's reference book on stream and batch data systems. 09Streaming 101: The World Beyond BatchEvent time, windows and streaming concepts. 10Eclipse IoTOpen-source projects for IoT devices, gateways and platforms.

2. Ingestion, Brokers and Device Connectivity

16 entries

Getting device data reliably into a pipeline. Standard protocols and brokers reduce lock-in.

11MQTTLightweight publish-subscribe messaging for devices. 12Eclipse SparkplugMQTT specification for industrial data and device state. 13CoAP (RFC 7252)Constrained Application Protocol for small devices. 14LwM2MLightweight device management and telemetry protocol. 15OPC FoundationOPC UA industrial interoperability. 16EMQXScalable MQTT broker with data integration. 17Eclipse MosquittoLightweight open-source MQTT broker. 18HiveMQEnterprise MQTT platform. 19Eclipse HonoOpen-source device connectivity and telemetry ingestion. 20AWS IoT CoreManaged device connectivity on AWS. 21Azure IoT HubManaged device connectivity on Azure. 22Apache KafkaEvent streaming backbone for device data. 23NATSCloud-native messaging with persistence. 24TelegrafPlugin-driven agent for collecting and shipping metrics. 25Apache NiFiDataflow automation for routing and transforming data. 26Node-REDFlow-based programming for wiring devices and services.

3. Time-Series Databases and Historians

10 entries

Storage tuned for time-stamped sensor data. Retention, downsampling and cardinality drive cost.

27InfluxDBTime-series database and platform. 28TimescaleDBTime-series database on PostgreSQL. 29QuestDBHigh-performance time-series database. 30Apache IoTDBTime-series database designed for IoT and industrial data. 31TDengineTime-series database for IoT with built-in stream processing. 32VictoriaMetricsFast time-series database and monitoring solution. 33PrometheusMetrics monitoring and alerting. 34CrateDBDistributed SQL database for time-series and IoT data. 35ClickHouseColumn-oriented database for analytics. 36Apache ParquetColumnar file format for archives.

4. Stream and Batch Processing

8 entries

Computing over device streams with windows, joins and state. Handle late and out-of-order data explicitly.

37Apache FlinkStateful stream processing with event-time semantics. 38Spark Structured StreamingStreaming queries on Spark. 39Kafka StreamsStream processing library inside Kafka. 40Apache BeamUnified model for batch and streaming pipelines. 41ksqlDBStreaming SQL on Kafka. 42RisingWaveStreaming database with SQL. 43MaterializeStreaming database with incrementally maintained views. 44DebeziumChange data capture for syncing operational data.

5. Analytics, Forecasting and Anomaly Detection

15 entries

Libraries, models and benchmarks for turning sensor streams into predictions and alerts. Evaluate on your own data; benchmark gains often shrink in production.

45sktimeMachine learning framework for time series. 46DartsPython library for forecasting and anomaly detection. 47StatsForecastFast statistical forecasting. 48ProphetForecasting for business and sensor time series. 49tsfreshAutomatic time-series feature extraction. 50PyODPython toolbox for outlier detection. 51MerlionLibrary for time-series intelligence. 52RiverOnline machine learning for streaming data. 53Numenta Anomaly BenchmarkBenchmark for streaming anomaly detection. 54Time Series LibraryDeep learning models and benchmarks for time series. 55TimesFMGoogle's time-series foundation model. 56ChronosLanguage-model approach to time-series forecasting. 57MOMENT: A Family of Open Time-series Foundation ModelsFoundation models for forecasting, classification and anomaly detection. 58PHM SocietyPrognostics and health management community and data challenges. 59MIMOSAOpen standards for asset management and condition-based maintenance.

6. Dashboards, Digital Twins and Platforms

8 entries

Showing results to people and modelling assets. Consistent asset models make analytics reusable.

60GrafanaDashboards for time-series data. 61Apache SupersetOpen-source BI. 62ThingsBoardOpen-source IoT platform with dashboards and rule engine. 63Eclipse DittoDigital twin framework for devices. 64Digital Twins Definition Language (DTDL)Open modelling language for devices and twins. 65FIWAREOpen standards and components for smart cities and IoT. 66Project HaystackTagging conventions for building and equipment data. 67Brick SchemaSemantic model for buildings and their systems.

7. Edge Analytics and Fleet Operations

10 entries

Running analytics near devices and updating them safely. Filter and aggregate at the edge to cut bandwidth and latency.

68LiteRTOn-device inference runtime. 69Edge ImpulsePlatform for building edge ML for sensors. 70tinyML FoundationCommunity for ML on ultra-low-power devices. 71EdgeX FoundryVendor-neutral edge IoT platform. 72KubeEdgeKubernetes-native edge computing framework. 73K3sLightweight Kubernetes for edge sites. 74Azure IoT EdgeRun containerised analytics on devices. 75AWS IoT GreengrassEdge runtime for local compute and ML. 76MenderOver-the-air update management. 77FlowerFederated learning framework.

8. Semantics, Quality, Security and Regulation

11 entries

Standards for describing sensor data and keeping it trustworthy and lawful. Data access and security rules are tightening for connected products.

78W3C Web of ThingsThing descriptions for interoperable IoT. 79SOSA/SSN: Semantic Sensor Network OntologyW3C vocabulary for sensors and observations. 80OGC SensorThings APIOpen standard for sensor observations. 81OpenLineageOpen standard for data lineage. 82Great ExpectationsData validation. 83NIST IR 8259: IoT Device Cybersecurity ActivitiesBaseline activities for device manufacturers. 84ETSI EN 303 645Cybersecurity baseline for consumer IoT. 85EU Data ActEU rules on access to data from connected products. 86EU Cyber Resilience ActRequirements for products with digital elements. 87GDPR TextFull text of the GDPR. 88MeitY Data Protection FrameworkIndia's data protection resources.

9. Open Datasets for IoT Analytics

6 entries

Public data for prototyping and benchmarking. Check licences before reuse.

89UCI Machine Learning RepositoryClassic datasets including sensor and activity data. 90PhysioNetOpen physiological signal datasets. 91IoT-23 DatasetLabelled IoT network traffic. 92TON_IoT DatasetsTelemetry and network data for IoT security research. 93Pecan Street DataportHousehold energy and appliance data. 94UK-DALEDomestic appliance-level electricity data.

10. Related Pages on This Site

20 entries

Directories for platforms, protocols, databases and analytics. Where each layer of an IoT analytics stack is listed.

95IoT Data Analytics PlatformsVendor platforms and tools for IoT analytics. 96Internet of Things (IoT)Hub of IoT platforms, protocols and sensors. 97Artificial Intelligence of Things (AIoT)AI at the edge and in the cloud for IoT. 98Industrial Internet of Things (IIoT)Industrial IoT architecture, protocols and security. 99Internet of Medical Things (IoMT)Connected medical devices and health data. 100Edge AIOn-device intelligence and runtimes. 101Edge Data Analytics SolutionsAnalytics at the edge. 102Real-Time Data AnalyticsGuide to real-time analytics architecture. 103Streaming Analytics PlatformsKafka, Flink, Kinesis and streaming databases. 104Time-Series DatabasesDatabases for time-stamped data. 105IoT DatabasesDatabases for IoT data. 106Time Series Forecasting ModelsForecasting methods and models. 107Time Series Foundation ModelsPretrained forecasting models. 108Predictive AIForecasting and predictive models. 109IoT PlatformsCloud and edge IoT platforms. 110IoT Device Management PlatformsFleet and firmware management. 111Message Queue SoftwareMessage brokers and queues. 112Publish/Subscribe MessagingBrokers, cloud pub/sub and MQTT. 113Sensor TechnologiesSensing technologies for IoT. 114Digital TwinsDigital twin concepts and resources.

2 entries

Related research directories on this site.

115Edge Data AnalyticsGuide to edge analytics architecture, messaging, stream processing, local storage, inference and fleet operations. 116Operational AnalyticsGuide to operational analytics: HTAP, CDC, reverse ETL, streaming, process analytics and KPIs.

Frequently Asked Questions

What is IoT data analytics?

IoT data analytics is the collection, storage, processing and analysis of data from connected sensors and devices to monitor assets, detect anomalies, predict failures, optimise operations and automate decisions. It combines messaging, time-series storage, stream and batch processing, machine learning and visualisation.

How is IoT data different from other data?

IoT data is usually high-volume and time-stamped, arrives continuously from many devices, can be noisy, late or missing, and often needs processing near the source. Schemas vary by device and firmware, so asset models and metadata matter as much as the readings.

What does a typical IoT analytics pipeline look like?

Devices publish readings over protocols such as MQTT or OPC UA to a broker or gateway, an event log such as Kafka buffers them, stream processors clean and enrich the data, a time-series database stores it, and analytics, machine learning and dashboards use it, with edge nodes filtering or scoring data locally.

Which databases suit IoT data?

Time-series databases such as InfluxDB, TimescaleDB, QuestDB, Apache IoTDB, TDengine and VictoriaMetrics handle high write rates and time-based queries. Many teams add a data lake or lakehouse in Parquet or Iceberg for long-term history and model training.

When should analytics run at the edge?

Run analytics at the edge when latency, bandwidth, privacy or connectivity limits make cloud round trips impractical, for example safety controls, video, remote sites or sensitive data. Send summaries and exceptions to the cloud, and manage models and updates across the fleet.

How do I detect anomalies and forecast from sensor data?

Start with simple baselines such as thresholds, rolling statistics and seasonal decomposition, then try libraries such as sktime, Darts, PyOD or Merlion, and evaluate foundation models such as TimesFM or Chronos. Benchmark on your own data and monitor drift, because published results often do not transfer.

What regulations affect IoT data?

Privacy laws such as GDPR and India's DPDP Act apply to personal data from devices. The EU Data Act sets rules on access to data from connected products, and the EU Cyber Resilience Act and standards such as NIST IR 8259 and ETSI EN 303 645 set security expectations for device makers.

How current is this list?

Links were checked in November 2026. Some sites block automated checks and may need a manual click. This page is a resource list, not engineering advice. For vendor platforms, see the IoT Data Analytics Platforms page.