From IoTWiki
Jump to: navigation, search


[ This article is adapted from IoT Stack: The Need for an Hourglass Architecture ]

In this article we use IoTPlatform as one word rather then the generic IoT Platform

Internet of Things (IoT) encompasses different technologies and standards from the sensor to the cloud. IoT is multi-faceted and inter-disciplinary, encompassing a plethora of networking protocols and technologies, using the latest advances in sensors, wireless connectivity, analytics, AI and with the need of security from the physical world to the cyber domain. Under the challenges of having a complete IoT system working, a natural tendency is to have it vertically integrated, build it grounds up, owning as much as possible, very similar to older generations of IoT/M2M deployments in a vertical or silo-ed approach. This is particularly true in the early days when the core horizontal components are not yet matured and standards still emerging. So everybody builds bottoms up. This also has roots in industrial SCADA world where reliability and control are paramount. It was not unusual for a Honeywell solution for building management to have its own dedicated sensor, networking switches and application. Changing a sensor may have required changes in all part of the solution especially the Application. There was no question of working with a Johnson Control equipment. Sensors and other components were calibrated with dependency among each component. So even shifting a device 10 meters away on a plant floor was not a trivial operation.

Hour Glass Architecture[edit]

‎ IoT Network in ISO layer world

However with the rapid growth of the consumer electronics business especially smartphones the price point of many sensors and the quality increased by quantum jumps. Sensors incorporated Digital Signal processing (DSP) on device and gave smoothed calibrated signal data. This allowed mixing manufacturers and models of sensors. This required decoupling the application from downstream components. The 7 layer ISO model is a great example where a similar decoupling and a layered Hourglass architecture greatly helped adoption. The IoTPlatfrom is the natural evolution to a layered architecture in IoT decoupling the devices and hubs/ gateways from middelware sending data to "on premise" or cloud based application and consumers. The graphic from Smart Home gets another standard illustrates the emergence of a decoupled layer. The biggest surprise is rapid adoption of new protocols like BlueTooth and modernization of older ones like ISA100 standard developed by the International Society of Automation (ISA) officially called "Wireless Systems for Industrial Automation: Process Control and Related Applications".


Machina Research defined the critical need for IoT Platforms in the June 2014 classic paper June 2014 classic report “The critical role of connectivity platforms in M2M and IoT application enablement”


There is great deal of fuzziness on features of an IOT Platform. In A Survey of Cloud IOT platforms] Partha Ray visualizes the features of the IoTPlatform as follows.

Functions of IOTPlatfrom
Device management
Onboarding devices, Receiving data packets, Configuring security profiles , Caliberation. LWM2M
Non IP Capillary network like BLE, ISA, wifi HART, LoRa, Zigbee
Handshake with Hub and gateway and Router and work with edge processing
UDP and IP based protocols (MQTT) and simpler non IP based middelware
Data and API
Semantics of the device and operations. Zigbee or OneM2M profiles.
Anomaly detection, health of devices and hubs and gateways
Bulk on boarding, relocation or transfer of facilities, Zoned operation.
Different device types, models and radio protocols
System Management
Management of entire infrastructure Endpoint, Network infrastructure and servers
Application Development
Multiple touch points Web , Smartphone ,Chatbots, speech (Amazon Alexa) , SMS and WhatsApp
State of network and zones , facilities . Heatmap of activity and busty data

Market Size[edit]

There are over 300+ IoTplatforms and about 20+ listed in The Indian IoT Startup Directory T IoT Applcation Enablement or IoT Platform is seeing considerable activity in terms of acquisitions and new market entrants. After PTC acquired ThingWorx and Axeda, other major software and IT companies have followed. Examples include Amazon acquiring 2lemetry, Autodesk buying SeeControl ,Microsoft acquiring Solair and SAP acquired

According to MachNation – an IoT industry research firm – worldwide IoT application enablement and device management revenue will be USD1.1 billion in 2016 and will grow to USD83.4 billion by 2025 at a compound annual growth rate (CAGR) of 62% over the period.<ref></ref> 

Types of IoTPlatfrom[edit]

The consultancy MachNations defines different type of offerings that are passed as IoTPlatform as of late 2016 <ref></ref>

Types of IoTPlatform by MachNations
Types of IoTPlatfroms by MachNation
Application Enablement Platform (AEP) is a technology-centric offering optimized to deliver a best-of-breed, industry-agnostic, extensible middleware core for building a set of interconnected or independent IoT solutions for customers. An AEP vendor relies on a flexible deployment model; a comprehensive set of device and enterprise-backend connector SDKs and APIs; and a set of well-documented developer resources. AEP vendors assemble a network of application development, system integrator and service provider partners that build custom IoT applications on the platform for customers. Enterprises realize that a well-built IoT application enablement platform (AEP) saves significant development time and money in the creation and operation of an IoT solution. Ayla Networks is an example of an AEP.
A Platform-Enabled Solution (PES) is a solution-centric offering optimized to reduce the time to deliver a fully-enabled, end-to-end, vertical IoT solution to customers. A PES vendor relies on a product-plus-services model; an application development lifecycle managed by in-house and partner resources; and either a rapid application builder or template-based framework. PES vendors use a direct or partner-led sales model and engage a core set of systems integrators to create validated, rapid and secure industry- or application-centric IoT solutions for customers. Philips CityTouch, a connected outdoor lighting offering from Philips, is an example of a PES.
A Connected Device Platform (CDP) is an offering optimized to provide complete carrier-grade lifecycle management of the connectivity element of an IoT device primarily for mobile network operators (MNOs) and mobile virtual network operators (MVNOs). A CDP vendor relies on custom carrier-managed implementations and a deep set of integration points with OSS/BSS for provisioning, billing, alerting and other operational purposes. CDP vendors use a direct sales model to engage with carriers. Jasper Control Center is an example of a CDP.
A Device Management Platform (DMP) is an offering that provides device lifecycle management functionality associated with the deployment and management of IoT assets. Typical IoT assets include IoT gateways, retrofitted and new industrial equipment, and Linux-based Arduino-like devices, Typical DMP functionality includes firmware upgrades, security patching, alerting and reporting about specific metrics associated with IoT assets. is an example of a DMP.
An Analytics Platform (AP) is an offering providing sophisticated data federation, statistical modeling and reporting tools to allow users to draw insights from ingested data. Usually, AP capabilities lay northbound of the data ingestion and processing functionality provided by other IoT platforms like AEPs and PESs. In essence, APs offer advanced analytics capabilities drawing using machine learning and artificial intelligence algorithms. SAP HANA is an example of an AP.

other ways of classifying IoT platforms would be

  • Segment Focus Home , SmartCity or Industrial (manufacturing)
  • Open Source proprietary Open Source is making significant inroads. carrier grade platform use open source components .
  • Scalability and agility tradeoff Enterprise or carrier class platforms are rather cumbersome and less scalable can be very agile
  • Support for non IP IoT Networks


Links to Architecture and Interoperability articles[edit]

Detail example[edit]

Some IoTPlatforms[edit]

Considerations in choosing a IoTPlatform[edit]

Trends and evolution[edit]

This is a evolving component

  • Open Source projects like Linux EdegeX Foundry for gateways are moving functionality for managing south side IoT networks and making it easier for cloud platforms like Amazon AWS or Microsoft Azure to manage devices
  • Companies like Nokia Impact platformare building in integration to thousands of industrial machines