How to Extend Legacy Systems into the AI & IoT Era

September 3, 2019 Brandon Lewis

With the recent discontinuation of 3rd generation Intel Core processors, how will the industry approach digital transformation?

Organizations in all industries are looking for ways to capitalize on technology megatrends like AI, machine learning and the IoT. More than just increasing efficiency and reducing costs, these enabling technologies allow for the implementation of new applications, services and revenue models.

However, many of the electronic systems in operation today are typically deployed for five years, 10 years, or longer, which makes it difficult for OEMs and system integrators to capitalize on new advancements in technology. In fact, in this time of rapid innovation, the mere presence of legacy electronic systems can accelerate technical and business obsolescence. 

Such is the case in sectors like industrial automation, health care, and test & measurement, where many electronic systems have been designed around integrated architectures. This makes it difficult for engineers to adopt the latest technologies, as upgrading an integrated system with new components often leads to compatibility issues. Therefore, each time a component or subsystem becomes outdated, the entire design becomes increasingly obsolete. 

Currently, this is the situation facing many devices that were designed around the recently discontinued 3rd generation Intel Core processor product line.

Still, the advantages of AI and IoT are too difficult to ignore. Collecting equipment or process data and transmitting it to an AI can immediately produce actionable insights. This is improving business and operational functions in the industries mentioned:

  • In automation they can reduce equipment downtime via predictive maintenance, enhance the yield and quality of manufactured products, optimize supply chains, and increase workplace safety.
  • In health care they can improve the accuracy of diagnoses and quality of patient care, streamline administrative processes, and reduce overall hospitalization costs.  
  • In test & measurement they can increase the accuracy of measurements, decrease the number of test cycles required, free human resources, and reduce expenses. 

Modular, COM Express-based designs allow organizations to capitalize on this potential while maximizing existing engineering investments in both new and legacy systems.

Connected Intelligence in a COM

COM Express addresses the challenge of integrated systems architectures by decoupling compute, networking, and other resources from core system functionality. It achieves this with a two-board architecture based on a compute module and underlying carrier board. The carrier board can be customized with application-specific I/O that allows data and power to be delivered between the COM Express compute module and the rest of the system.

Therefore, a legacy COM Express compute module can be replaced with a newer model containing AI or IoT-centric features. Given the recent discontinuation of popular 3rd generation Intel Core processors, COM Express offers a scalable path forward in the form of 7th generation Intel Core embedded processors.

Seventh generation Core embedded processors are based on the Kaby Lake microarchitecture and provide up to four cores and a maximum clock frequency of 4.2 GHz with a configurable TDP of just 35W. The devices also integrate a range of features geared toward AI and IoT systems, including:

  • Intel Advanced Vector Extensions 2 provides optimized instructions for floating-point computations, as well as 256-bit integer instructions for fused-multiply add (FMA) operations. These types of arithmetic are essential for the efficient, precision execution of AI and machine learning algorithms. Such algorithms can be developed and optimized for the integrated HD graphics units of 7th generation Core processors or companion accelerators like Movidius Vision Processing Units (VPUs) using AI development frameworks like the Intel® OpenVINO toolkit.
  • Support for up to 64 GB of 2133 MHz DDR4 memory, which is essential for AI inferencing because of the wide vector units used by CPUs for advanced arithmetic such as mentioned above.
  • Security technologies like Intel Protection Technology (IPT), Intel AES-NI, Intel Memory Protection Extensions (MPX), and others help defend the boot sequence, sensitive data, and memory regions from network-borne attacks and software vulnerabilities. In an IoT system context, they also accelerate cryptographic workloads to free resources on the main CPU, and offer a unique, unalterable identity that can be used in remote authentication.
  • Intel vPro delivers both virtualization and remote management capabilities that allow embedded systems to function like enterprise servers. Intel Virtualization Technology (Intel VT-x) can help partition the multicore devices so that certain cores hosting enterprise functionality can be updated with new software a la enterprise continuous delivery models. This while remaining isolated from other real-time functions of the system. vPro also includes provisions for remote management so that devices can be monitored, upgraded or repaired securely over the air or wire.

These processors are supported on COM Express Type 2 modules like the ADLINK Technology Express-SL2/KL2.

Figure 1. The ADLINK Technology Express-SL2/KL2 COM Express Type 2 modules provide an upgrade path from the recently discontinued 3rd generation Intel Core processors to more modern 6th and 7th generation devices (Source: ADLINK Technology).

Bridging Legacy and Opportunity with COM Express

With 10-year lifecycle support from ADLINK and 15 years of support on Intel embedded CPUs, the 7th generation Intel Core-based Express-KL2 and 6th generation Intel Core-based Express-SL2 are viable options for migrating legacy equipment into the age of AIoT systems.

Most notably, the COM Express Type 2 modules host a VGA, PATA IDE, and PCI-compatible pinout that allows it to interface with a range of systems already well into their deployment lifecycle. They are also natively compatible with operating systems such as Windows 7, Windows Embedded Standard 7, Windows 8.1, and Embedded Linux. So rather than re-integrating a new carrier board, compute module, and software, existing systems based on the popular COM Express Type 2 specification can take advantage of modern processor capabilities by simply dropping in an Express-SL2 or KL2 processor module.

Figure 2. The Express-SL2/KL2’s Type 2 pinout supports legacy interfaces like PCI, PATA IDE, and VGA, allowing existing systems to migrate to modern day performance by simply swapping out compute modules (Source: ADLINK Technology).

For engineering teams embarking on a new system design, the benefits of the COM Express architecture are obvious. The popularity of the standard and the modularity of the architecture provides a straightforward migration path that can sustain and modernize a product for decades. But for legacy systems based on integrated architectures, on the other hand, a retrofit may be in order.

The aforementioned support for legacy I/O and software on platforms like the Express-SL2/KL2 is a possibility in these situations. ADLINK design servers can assist with BIOS modification, hardware migration, and even testing to support retrofit projects. While the NRE investment is higher in the short term, the ability to reuse technology and prolong machinery with modernized processor complex easily offsets those costs.

Engineers looking to add groundbreaking AI performance to Express-SL2/KL2-based designs can utilize the aforementioned Intel OpenVINO toolkit to develop precision AI inferencing algorithms that run on the integrated graphics cores of the host 6th and 7th generation Core processors.

For designs that require even more ML performance, discrete AI accelerators like the Movidius X VPU can be outfitted on a companion carrier board. These extremely low-power hardware accelerators for neural network inferencing workloads integrate specialized vision processing cores that can be used to augment the AI performance of a host processor or provide AI compute capabilities to COM Express designs that don’t include an AI-compatible host.

Once again, with the accelerator housed on the carrier board, engineers are left with more options and more flexibility for the future.

In addition to 6th and 7th generation Core processors, the Express-SL2/KL2 supports Intel Xeon processors and Intel Celeron devices. Along with 0ºC to 60ºC to -40ºC to +85ºC operating temperature range options, this breadth of processor support provides Express-SL2 and -KL2 systems with a sliding scale of performance and power consumption to meet the requirements of most automation, health care, and test systems.

Streamlining Digital Transition

With the ability to upgrade product offerings while keeping previous hardware and software investments in tact, solutions like ADLINK’s Express-SL2/KL2 module can thrust systems into the domain of connected intelligence.

In the health care industry, for example, OEMs are leveraging the devices to upgrade ultrasound, X-ray, and endoscopy machines that benefit from the improved graphics and video encode/decode performance that current-generation Intel processors deliver in 3D imaging applications. In the automation sector, enhanced multicore performance and Gigabit Ethernet interfaces on the Express-SL2/KL2 are empowering Industry 4.0 designs, enabling enterprise and operational applications on the same hardware as well as immediate access to IP networks. Test and measurement systems are utilizing the platform to add more and more performance for increasingly complex simulations.

And, as next-generation compute, memory, and networking technologies are integrated into advanced SoCs, future COM Express technologies will help designers in these industries continue to maximize their IP.

About the Author

Brandon Lewis

Brandon Lewis, Editor-in-Chief of Embedded Computing Design, is responsible for guiding the property's content strategy, editorial direction, and engineering community engagement, which includes IoT Design, Automotive Embedded Systems, the Power Page, Industrial AI & Machine Learning, and other publications. As an experienced technical journalist, editor, and reporter with an aptitude for identifying key technologies, products, and market trends in the embedded technology sector, he enjoys covering topics that range from development kits and tools to cyber security and technology business models. Brandon received a BA in English Literature from Arizona State University, where he graduated cum laude. He can be reached by email at brandon.lewis@opensysmedia.com.

Follow on Twitter Follow on Linkedin Visit Website More Content by Brandon Lewis
Previous White Paper
Think Local: How to Migrate Intelligence from the Cloud to Embedded Devices at the Edge
Think Local: How to Migrate Intelligence from the Cloud to Embedded Devices at the Edge

Adding more cloud compute won’t solve the problem of managing exponentially growing quantities of data for ...

Next Article
Alps Alpine’s Digital Air- and Water-Pressure Sensor is Waterproof

Alps Alpine started mass-producing the HSPPAD143A Waterproof Digital Pressure Sensor, targeting application...

How to Develop Cross-Industry IoT Interoperability

Multi-Part Series