A Nice Head Start: Q&A with JC Ramirez, ADL Embedded Solutions
February 16th, 2018
By Anne Fisher, Managing Editor at EECatalog
Editor’s Note: “Not all the answers are coming from the hardware-level guys,” JC Ramirez, Director of Engineering at ADL Embedded Solutions, Inc., tells EECatalog. During a quick conversation that Ramirez made time for while preparing to showcase solutions applicable to unmanned or autonomous vehicle mission/payload computing, the Industrial IoT, and more at Embedded World, we spoke about supporting legacy systems, cybersecurity, and processor choice. Edited excerpts follow.
EECatalog: Please tell our readers more about your Edge-Connect architecture and applicability to unmanned systems.
JC Ramirez, ADL: While our Edge-Connect architecture is not unlike other small form factor expansion architectures like Qseven or SMARC COM modules, it is first and foremost a single board computer. That’s the key difference, and it makes possible very small, compact solutions with just the CPU board alone. (See ADLEPC-1500/ADLE3800SEC). Also, the Edge-Connect expansion interfaces are architected in such a way that many of the PCIe x1 lanes are shared (multiplexed) with SATA and USB 3.0 interfaces in addition to being available for custom use. The result is a much smaller expansion bus footprint, allowing for SBCs as small as 75mm x 75mm and full system solutions on the order of 3”x3”x1” (See ADLEPC-1520).
EECatalog: How is the discussion about legacy support different today than it was even a few years ago?
Ramirez, ADL: Up to about two to three years ago the exercise was simply that of figuring out how to adapt the latest interfaces. For example, in PC/104 architecture, the latest interface is the PCIe/104 interface. Legacy buses include the PCI Bus and previous to that the ISA Bus.
When you have a lot of legacy ISA stuff still hanging around, and the customer wants to update to the latest and greatest CPU processors, for companies like ADL, WinSystems, and Diamond Systems—the exercise for the longest time was just, “How do we adapt from one bus to a legacy bus?” If you have the space and the stack height available to you, you can adapt your way back. ADL makes adapter boards that can take you from PCIe/104 back to PCI/104. Other vendors have adapter boards that go from PCI/104 to ISA.
That’s up to two or three years ago. The latest challenges with supporting legacy not only include adapting the buses to the legacy hardware, but also helping customers meet the latest cybersecurity requirements—we’re talking things like FIPS 140-2 from NIST or [Trusted Platform Module] TPM 2.0 from the Trusted Computer Group (TCG)1. The latest cybersecurity is only supported on the latest OSes. So now you somehow have to do something at the OS level to be able to still support the legacy stuff that was designed 10, 15 plus years ago when there wasn’t any inkling that we would reach the kinds of security requirements we have today.
The answers aren’t easy, and not all the answers are coming from the hardware-level guys like ourselves. One solution that end users are using is software emulation, for example, to support those old legacy functions. Many of those old interfaces are slow enough that if your software emulates them with the latest processors you can actually meet the speed requirements of that old hardware, whereas at the time the only way to get that level of speed was with a hardware rather than software solution.
Customers are also using tools like VMWare to simulate the old operating system inside one of the latest operating systems such as Windows 10 or Linux 16.04 and others.
The hardware requirements so that customers or software engineers can do these kinds of things—usually means multicore processors. For PC/104 quad core would be typical. For larger form factors you can even go octal core or higher.
From a cybersecurity perspective—people are still latching onto TPM as a key hardware component of any security solution. It is not clear that we have optimal, discrete chip kinds of solutions for TPM today, so this is where Intel® is beginning to solve this problem nicely for a lot of computing vendors.
Intel has a TPM 2.0 firmware solution, which is part of their Intel PTT [Platform Trust Technology] feature of the chipset. So to the extent that Intel continues to promote the Intel PTT feature of its chipsets, that makes it possible for embedded CPU vendors such as ourselves to enable firmware TPM 2.0 solutions for our customers, so that they can continue to meet their security requirements.
Something to note is that in lieu of discrete TPM kinds of solutions, the storage vendors are also looking to figure out how they can add security and encryption components to their SATA modules.. We have other vendors incorporating a full TPM solution on to their SATA modules. The storage vendors might be a little bit late to this party, but all of this is work-in-progress, and it will be interesting to see where it eventually winds up.
Intel has a nice head start. If you look across at the various vendors, not just ADL but our competitors, and you look at the TPM part of their solutions, you’ll see that a lot of them are touting the firmware TPM solution, and that relates directly back to the Intel PTT function of their chipsets.
EECatalog: As noted here, ADL accommodates a gamut of Intel processors to span a range of small form factor and ruggedization requirements. What is it crucial to consider when choosing among processor options?
Ramirez, ADL: For military and rugged industrial embedded engineers, the critical factors beyond CPU performance and power involve the longevity or long-term availability of the processor, and the design/engineering support services behind it. For this reason, we work exclusively from the Intel Embedded Roadmap whose lifetime availability ranges from seven years to 15 years for the Intel E3800-Series Atom™. Our engineering and BIOS teams provide support for custom BIOS settings and features (logos, etc.) or custom services such as firmware development to enable security features such as Intel PTT firmware TPM, or interfacing to custom encrypted modules, for example. These are areas where Arm and RISC still lag significantly and therefore, are not ideal candidates for these types of applications.
About JC Ramirez
JC Ramirez is VP of Engineering and Product Management at ADL. JC (BSEE, MBA), has a technical background that includes Navy nuclear plant operation, nuclear instrumentation, semiconductor product development and embedded systems engineering.