Requirements and issues for developing customized intelligent IoT gateway

As companies evolve their Internet of Things (IoT) strategies, one clear trend taking hold is the need for a customized solution, including the IoT gateway component. In a recent blog on embedded.com, I characterized this trend as an indication that the industry is moving from the relatively simple implementation of leveraging gateways as data aggregators that include cloud-based storage and processing – to the realm of “IoT 2.0” where intelligence is pushed from the cloud to the gateway and edge devices. Further, within this IoT 2.0 realm, security is treated more comprehensively and connectivity is broadened to include more devices across the network up through the business enterprise.

From the perspective of the IoT gateway, IoT 2.0 means the gateway component of the cloud strategy requires comprehensive connectivity, security, and processing capabilities. In essence, the IoT gateway needs to be an intelligent device capable of handling and processing a variety of functions. In a recent report, “The Global Market for IoT & Intelligent Gateways,” VDC Research states that the intelligent gateway segment is the largest and fastest growing segment of the IoT gateway market (Figure 1). Taking a closer look at the intelligent IoT gateway is the focus of this article.


Figure 1: The intelligent gateway is making impressive strides in the IoT world. (Source: VDC/Research 2015)

The trend toward customization
What factors are pushing companies toward customized gateway solutions?

One significant factor is connectivity. The need to extend the IoT reach within the network, whether it’s a factory, hospital, or white goods in the home, means connecting existing legacy, or “brownfield” devices alongside new devices, which typically leads to the need for customized interfaces.

Security is another huge factor. As connectivity proliferates across the enterprise, additional points of attack are potentially created. When intelligent processing is pushed from the cloud to the gateway and edge devices, companies are increasingly concerned about the risk of cyber-attack. These are very real concerns that can disrupt operations, result in theft of information, or worse, cause human injury. Customized gateway solutions help address threat profiles specific to a company’s IoT strategy and implementation.

Finally, specific processing requirements drive customization of the intelligent gateway. These requirements might include network throughput, open source and proprietary IP, soft real-time versus hard real-time performance, power management, graphics, and other factors. These factors also have a significant impact on processor selection.

No question, business pressures are unrelenting. As a result, gateway requirements are becoming more complex and businesses face a significant challenge finding a gateway solution that meets their specific technical needs while at the same time meeting their business objectives. This is what is driving the need for IoT gateway customization.

Customizing connectivity
In most cases, extending the reach means integrating new IoT infrastructure while taking into account existing devices and infrastructure in brownfield deployments. For example, on the factory floor, there are many types of devices such as controllers, sensors, and actuators that include many different types of physical connectivity options including serial Fieldbus, industrial Ethernet, wireless, proprietary, and many others.

Prior to IoT 2.0, the most common way to build such a topology was to integrate a series of off-the-shelf routers, bridges, and gateways that network everything together. Unfortunately, this requires a proliferation of devices, not to mention the need for more infrastructure, many types of spares, and additional points of failure. Clearly, it’s more desirable today to implement an IoT network topology with a minimal number of components. Building a customized gateway is one such approach to accomplishing this goal as shown in Figure 2.


Figure 2. Industrial IoT environments that utilize a customized intelligent gateway are able to simplify the network topology, reduce cost, and improve operational reliability. (Source: Mentor Graphics)

Customizing connectivity often requires both hardware and software. For example, if an industrial automation business needs to integrate new EtherCAT-connected devices at the edge, the business could benefit from integrated EtherCAT hardware and accompanying software stack in the gateway, integrated alongside legacy interfaces.

Next page >>

Addressing security
How much security to build into your IoT network depends on several factors, but it all begins with answering the basic question: “What am I protecting and what is it worth?” As you might expect, the answer can range from very little, to everything. In some cases, standard networking security (SSL and HTTPS), a firewall, encrypted databases, and checksum security might be sufficient. In other situations, businesses might require hardware-enforced security partitioning (much like the ARM TrustZone environment), certificate-enabled authentication, or a security solution from vendors such as Icon Labs which detects security events at the edge and reports them to the enterprise. Security customization requires a comprehensive analysis and workshop to define the security attack surface from a system perspective, and then customize a solution potentially in both hardware and software to sufficiently address the security exposure.

Gateway processing requirements
As VDC notes, the intelligent gateway is basically a powerful application platform.

One significant embedded trend is the use of advanced multicore SoCs. Solutions such as the recently-announced Mentor Graphics Intelligent Gateway System Design Kit (SysDK) is based on an NXP i.MX6 (ARM Cortex A9-based) quad processor. This processor is optimized for graphics performance and for use with high-performance displays, digital signage, and related graphics applications. Some customers might desire a gateway based on the NXP LS1021A (ARM Cortex A7-based) due to its optimization for networking throughput. In other cases, and depending on the requirements, customers might want to use a more complex heterogeneous SoC, such as the TI Sitara AM572X which includes 2x Cortex A15 + 2x Cortex M4 + DSP. So when we talk about an intelligent IoT gateway, customization should happen in both the hardware and software. If a customer needs a lot of processing power for their intelligent processing, a quad-core processor with multiple OSs or one OS running in SMP mode might be the answer. Or for more rudimentary processing, perhaps a single OS and a single-core processor might be enough.

With complex SoCs, the concept of a “secure converged gateway” has entered the discussion. What is meant by secure converged gateway? In reference to secure , it’s how security concerns continue to rise as intelligence is driven out of the cloud and towards the edge – and as more private and potentially public devices connect to the gateway. The idea of converged means that in many industries, convergence of functionality is a means toward reduction of size, weight, power, and cost, often referred to as SWaP-C.

Customers are now talking about converging security, proprietary capabilities, and open source environments into ever more powerful intelligent gateways based on increasingly complex SoCs. This is no longer a discussion about off-the-shelf gateways, but about customization and building a gateway that fully utilizes the underlying hardware and meets the differentiated needs of the business.

There are clear desires to converge proprietary applications alongside open source Linux-based applications for both performance reasons and for intellectual property protection. In some cases, companies are leveraging the heterogeneous cores for power management. One example of this is the TI AM572x, where the power-hungry Cortex A15 cores are put to sleep until they are needed, while the M4 cores perform all of the necessary functions.

Such complex designs also require a tooling solution that enables developers to design, develop, and debug heterogeneous architectures from an overall system perspective. Platforms such as the Mentor Graphics Multicore Framework and heterogeneous solutions are designed specifically to address these system design complexities as shown in Figure 3.


Figure 3: Heterogeneous SoCs and heterogeneous OS solutions enable powerful and customized “secure converged” gateway solutions. (Source: Mentor Graphics)

One final factor driving the need for gateway customization
Aside from the factors already mentioned, one final – and extremely important factor – is cost reduction. When a customer purchases an off-the-shelf gateway, costs will be incurred from a couple perspectives. First, there is a high likelihood that a significant portion of the hardware in the gateway will go unused. Thus, the customer is paying for unnecessary hardware, which can add up when volumes are considered. Additionally, if the needed functionality is not present, more hardware will have to be purchased and integrated at a later time (e.g., adapter for 802.15.4 mesh connectivity) increasing cost and adding new layers to the system.

Today, businesses could design and build a gateway on their own, or approach a contract manufacturer. The time and cost of defining requirements, vetting the design, obtaining initial samples, testing and validating functionality, and iterating through the redesign-test-validation cycle multiple times can consume quarters, if not years of time. And one must also consider the cost and complexity related to the operating system(s) tools, security, connectivity, power management, graphics, booting mechanism, Inter-Processor Communications (IPC), and the list goes on.

Summary
As businesses endeavor on their IoT 2.0 paths, they have increasingly complex requirements to consider, but they also have options with tradeoffs. There are three general paths for businesses to evolve their IoT gateway strategies:

  • Continue to integrate and leverage off-the-shelf hardware/software solutions
  • Leverage off-the-shelf hardware, and invest in the customized gateway software
  • Customize the hardware and the software

In each case, a business must consider the following questions:

  • What is the cost of the hardware? Am I paying for hardware I don’t need?
  • What are my options if I need customized hardware support?
  • Do I have the ability to customize the software, to make optimal usage of my hardware in both the short-term and long-term?

We are entering the era of the customized intelligent IoT gateway. As companies seek to improve their business results and differentiate themselves from the competition, customized solutions are clearly required. These solutions must take into account brownfield installations, comprehensive connectivity, end-to-end security, and specific requirements as intelligence and processing are being pushed from the cloud toward the edge. Clearly, the IoT supply chain is moving in this direction. Solutions such as Mentor Graphics's recently announced Intelligent IoT Gateway SysDK are designed to address customization needs for emerging intelligent IoT gateways.

Warren Kurisu is the Director of Product Management in the Mentor Graphics Embedded Systems Division, overseeing the embedded runtime platform business for the Nucleus RTOS, Mentor Embedded Linux, virtualization and multicore technologies, safety certified runtimes, graphics and development tools. Warren has spent nearly 30 years in the embedded industry, both as an embedded developer and as a business executive, working broadly in industries including aerospace, networking, industrial, medical, automotive, and consumer.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.