Enterprise CIOs have moved beyond the point of evangelizing and selling Internet of Things to internal stakeholders to implementing the right IoT strategy for the organization. They are evaluating a variety of technologies ranging from open source software to turnkey IoT platforms to cloud-based PaaS. With almost every major platform vendor offering an IoT stack, choosing the right platform is becoming a challenge.
The rise of managed cloud platforms made it possible for cloud service providers to offer end-to-end IoT PaaS services. Amazon,
Image Source: businesscloudnews.com
I attempted to identify the key attributes of an IoT platform that align with the requirements of an enterprise. This checklist will help CXOs to choose the right offering for implementing a robust IoT solution. To understand the essential building blocks of an enterprise IoT platform, refer to the article that I published earlier.
1. Comprehensive Device Management
The devices layer is the most important component of an IoT solution. A mature IoT platform comes with comprehensive device management features that let customers on-board existing and new devices with rich metadata. Identical devices that share the same metadata are grouped together. This feature makes it possible to search devices based on their capabilities.
The platform also provides per-device authentication and authorization to enforce enhanced security. It makes it easy to define which devices can connect, send, and receive messages. Devices can be easily blacklisted and whitelisted through declarative policies.
Mature IoT platforms provide remote management capabilities. Through this feature, customers can perform tasks such as firmware upgrades, remote debugging, and maintenance of devices deployed in remote locations.
Finally, the platform makes it easy to leverage an existing inventory of devices and assets maintained in a traditional MRP solution.
2. Support for a Wide Range of Protocols and Standards
M2M communication and industrial automation existed for decades. IoT augments that with data-driven operational insights. Enterprise IoT platforms provide a mechanism to support legacy protocols along with the contemporary protocols to deliver the full spectrum of automation.
IoT platforms should directly or indirectly provide protocol translation. Many existing systems rely on SCADA based RTUs and PLCs for automation. Some of the devices use protocols such as BACNet, Modbus, and Canbus for communication. Through a protocol translation gateway, an IoT platform should support on-boarding legacy devices.
Enterprise IoT platforms should be extensible to support current and emerging protocols. They should support Ethernet, WiFi, PoE, BLE, GSM, LTE, ZigBee, Z-wave, AllJoyn, HomeKit and emerging protocols such as Weave and WiFiHalow.
3. Scalable Message Broker
One of the key scenarios enabled by an enterprise IoT platform is machine-to-machine communication. Legacy devices, existing gateways, intelligent IP-enabled devices must easily exchange messages through the platform. To support this capability, IoT platforms need to implement a scalable, reliable, and secure message broker that coordinates the communication among the devices.
Enterprise IoT platforms support a wide range of protocols for M2M scenarios. They expose HTTP, AMQP, MQTT, CoAP, STOMP, XMPP, and WebSocket protocols to send and receive messages. The choice of protocols makes it possible for disparate devices and gateways to talk to each other.
Message brokers play a critical role in the overall architecture. They are designed for high-scale, low-latency connectivity. Mature IoT platforms don't leave the responsibility of connecting the dots to the customers. They implement closed loop messaging in the form of a publisher / subscriber pattern. Devices can easily publish messages to a centralized endpoint which then distributes the messages to interested parties. Based on MQTT or other protocols, the message brokers provide solid M2M messaging infrastructure out of the box.
Since industrial IoT deals with tens of thousands of devices, the message broker should be able to elastically scale to support the non-linear growth of connected devices in an enterprise.
4. Separate Endpoints for Data Ingestion and Device Connectivity
The devices layer of an industrial IoT solution consists of devices ranging from sensor nodes, actuators, and gateways. Not every device generates telemetry data that needs to be stored and processed. The actuators and switches will act upon the commands they receive via the platform.
IoT platforms should have a clean separation of device management that deals with M2M communication and data ingestion endpoints that acquire data from multiple sensor nodes. Typical M2M endpoints that expose MQTT shouldn't be used for ingesting high-velocity telemetry data. Similarly, it is not a good idea to use ingestion endpoints for sending messages from devices that don't belong to the telemetry datasets. The clean separation of these layers results in efficient M2M communication and data ingestion layers.
5. Support for a Robust, Declarative Rules Engine
The messages and the telemetry data originating from the devices layer need to be constantly monitored to find anomalies and unusual usage patterns. A rules engine analyzes the incoming stream of data and performs an action. For example, when the engine oil level in an automobile falls below a specific threshold, the platform will send an alert to the driver and also a notifies the service station. In this example, the rules engine monitors the data point related to the engine oil for a specific threshold and performs the action of notifying the driver and service station. The combination of the relevant data points, thresholds, and actions form a rule that is maintained by the rules engine.
An enterprise IoT platform should expose a declarative mechanism to define the rules. Non-technical users such as business analysts should be able to easily change the thresholds and associated actions without the need to manipulate the running state of the solution. They should be able to visualize the incoming datasets, definition of rules, and a set of actions in a canvas that makes it easy to define and manage rules.
6. Security at Every Layer of the Stack
One of the biggest concerns of CIOs considering IoT is security. Given the risks that come from connected devices, security is paramount to an enterprise IoT solution.
The devices layer should support the highest level of encryption for sending messages and receiving commands. Each device and gateway should be explicitly authenticated by the IoT platform. Proven mechanisms such as PKI, TLS/SSL should be implemented for encrypting the communication.
Unauthorized users accessing an IoT system can cause havoc to an organization. Each user participating in the IoT solution should be authenticated using multi-factor authentication mechanisms. The IoT platform should support authentication and authorization which is tightly integrated with an existing identity platform such as a corporate directory. Role-Based Access Control (RBAC) should be used to perform authorization at a group level.
The IoT platform must support a comprehensive policy-based system that defines the authorization of devices and users. Each policy explicitly defines the permissions of each device, user, role, and applications. For example, only senior decision makers of the organization will have access to the business intelligence reports while floor supervisors can control only the devices deployed on their respective floor.
7. Integration with the Edge Devices
The devices layer consists of disparate devices ranging from sensor nodes, hubs, transparent gateways, opaque gateways, and specialized devices that act as aggregators. Some of these devices are directly connected to the IoT platform while a few are represented through the gateways. Apart from acting as a proxy to the legacy devices, a gateway may double up as an intelligent system with sufficient processing power. It will have the basic compute, network, and storage capabilities to tackle scenarios that demand local processing. These devices are called the edge devices which are typically deployed within the perimeter network.
Enterprise IoT platforms should be able to differentiate simple devices from intelligent gateways. They need to provide seamless integration capabilities for the edge devices to move the heavy-lifting to the specialized platforms while performing actions that demand low-latency locally.
8. Clean Separation of Real-time and Batch Processing of Data
The telemetry data ingested by the sensor nodes needs to be processed by the analytics engine of the IoT platform. Not every data point sent by the sensors needs near real-time analysis. An IoT solution should provide a clean separation between the paths for processing near real-time datasets and datasets that are needed in the longer term.
A subset of the data points generated by the sensors needs to be monitored and processed in real-time. A combination of
Some datasets submitted by the sensors will be valuable in the longer term. They need to be collected, aggregated, processed, and analyzed over a period of time. They contribute to the business intelligence required to make decisions that have a long-term impact. Apache Hadoop, Data Lake, and traditional Data Warehouse are leveraged for delivering the insights from the IoT system.
9. Extensible Data Processing Pipeline
Irrespective of the path that a dataset takes, the processing pipeline should be extensible. An IoT platform should make it possible to extend the data processing capabilities to specialized systems.
With artificial intelligence and machine learning becoming accessible, IoT platforms should support easier integration with them. They should enable invoking external machine learning algorithms within the data processing pipeline. Organizations will immensely benefit from the integration of business intelligence with machine learning. They open up new doors to scenarios such as predictive maintenance of expensive assets, which would result in saving multi-million dollars.
Enterprise IoT platforms should support invoking 3rd party APIs and services within the data processing pipeline. For example, passing the latitude and longitude data originating from a connected car should be transparently mapped to a location within the processing pipeline. This capability makes it possible to enhance the system by seamlessly integrating with the emerging technologies.
10. Integration with Existing Line-of-Business Applications
Enterprises are known for their ability to deal with heterogeneous systems. They have significant investments in line-of-business systems, corporate directory services, messaging and collaboration, enterprise resource planning, customer relationship management, and material resource planning systems among other assets. In a majority of the scenarios, the devices participating in an IoT solution are already managed through the MRP and ERP systems.
Enterprise IoT platforms should be extensible enough to leverage existing systems while augmenting their capabilities. They should seamlessly integrate with corporate directory services for user management, MRP for asset management, and ERP for supply chain management and asset tracking.
For many CIOs, IoT strategy is only an extension of their existing enterprise application integration system and business intelligence system. IoT platform vendors should provide extensibility to support integration with these systems.
