Defining an Open System Architecture Standard for Defense Systems

For many years now, there have been efforts within the US defense community to adopt “Open System Architecture” principles (or OSA) for defense computing platforms. Stretching back to the mid-2000’s there have been numerous efforts by all service branches to define sets of OSA principles or requirements on the embedded computing systems they procure and deploy. While these efforts have resulted in widespread adoption of POSIX functions for common tasks and a gradual shift in the use of common form factors and technologies such as VPX and Ethernet, there are still wide disparities between products from different suppliers, which creates difficulties for integrators. Product variations also make technology insertions or system configuration changes expensive, time consuming, and often locked to a particular vendor.

In 2019 the US defense embedded computing market underwent a transformation. Started in 2017, the Sensor Open Systems Architecture group (or SOSA) is an organization consisting of government, system integrators, and component suppliers who are working to craft an OSA standard for defense sensor systems. If successful, SOSA will standardize hardware implementations and form a guideline for system integration in order to encourage greater reuse and enable rapid future technology insertions.

The Argument for Standardization

Standardization is a natural stage in the lifecycle of any class of product. As a market matures, common implementations and uses begin to emerge. Standards are then established, either formally through a standardization effort or informally by a major player, permitting their implementation to be duplicated by other suppliers. Either way, standardization defines an agreed-upon set of features or implementation details, and results in common usage and general product interchangeability.

The computing industry as a whole has many great examples of standardization that helped put structure to a market and lower development and maintenance costs. Looking at a few examples from the greater embedded computing community, there are two that illustrate the power of standardization. The Mini-ITX (mITX) family of motherboards has been enormously successful. By leveraging and expanding the original commercial ATX form factor and standardizing such things as footprint, power connectivity and profile, a basic feature set, and a more-or-less common I/O aperture, ATX/mITX has enabled a huge market of components such as enclosures and power supplies, and has supported many generations of embedded computing platforms.

Another example is COM Express®. A more specialized form factor, COM Express (or COMe) focused on creating a form factor specifically for the needs of embedded computing platforms. It specifies four standard footprints (although only three are common) and eight types of standard pin connectivity (although only three are common). As a result, COMe defines a platform that allows common carrier board designs to support a wide range of processors allowing integrators the freedom to select the optimal solution for the job, and it enables easy technology insertions as new processors become available.

The common element of these two standards is that they both provide a balance of what is standardized and what remains open for the supplier to offer or innovate. Both rigidly define the board footprint, a minimum feature set, I/O, power, and the integrator’s use case, while leaving things like the choice of processors, application-domain features, and expansion options up to the supplier. Within the parameters defined by the standard, market commonality is established while giving suppliers the room to innovate, differentiate, and form a market.

Defense Industry Standardization Challenges

Defense computing brings a unique set of challenges to the standards process, however. There can be broad variance in the functions and features defense systems must support that can make standards definition quite difficult. Defense system developers must contend with multicomputing (that is, high-performance embedded computing), extremely high bandwidths including orders-of-magnitude differences between systems, varying system architectures, and occasional extremely low latencies (e.g. sub-microsecond).

VPX (as the VITA 46 standard is known), along with its complementary standard, OpenVPX (VITA 65), form a board form factor and backplane connectivity standard that has been widely adopted by the defense embedded computing community as well as a number of other challenging markets. As a standard, and along with a number of other standards documents that address such things as different cooling methods, VPX successfully defines the mechanics of its form factor, its physical dimensions, and the electrical characteristics of its backplane connectivity.

The Kontron VX305C-40G is a 12-core Xeon-D-based single-board computer (SBC) designed to the SOSA Payload/Compute-Intensive slot profile. Its primary I/O features a 40 Gigabit Ethernet (40GBASE-KR4) data plane, 10 Gigabit Ethernet (10GBASE-KR) data and control planes, and an 8-lane PCI Express Generation 3 Expansion plane.

As a result, compliant VPX boards can, from a mechanical standpoint, reliably fit into any compliant slot. The standards also define a set of slot and module profiles, along with a standard way to define board-to-board connectivity in the form of chassis profiles. Where VPX/OpenVPX fell short, however, was in their failure to successfully define “universal” slot profiles. OpenVPX currently contains 30 6U and 63 3U slot profiles, all of which contain “user defined” pins that allow suppliers to define individual pin functionality. Consequently, there is basically no commonality among vendors, which means that systems many times will have a unique backplane designed for connectivity between a specific set of boards. The resulting systems are difficult and expensive to design and integrate, and board-level upgrades are virtually impossible.

The Need for Open Systems Architecture Standards

The goal of the Sensor Open Systems Architecture (SOSA™) is a new, draft standard applying OSA principles to high-performance defense sensor platforms. Focusing on overcoming compatibility and interoperability issues, SOSA has outlined an expansive standard encompassing board-level profiles, signaling, and connectivity rules, as well as more application-level elements such as system design, system and chassis management, security, and chassis connectivity.

While still a work-in-progress, SOSA is already beginning to transform the VPX market. A relatively small set of slot and module profiles have been defined with very concrete rules regarding protocols and connectivity, and these profiles are in the process of being captured in OpenVPX. The primary slot profiles defined in SOSA include:

  • 3U Compute/RF intensive

  • 3U I/O intensive

  • 3U Data/Control Plane switch

  • 3U RF switch

  • 3U Radial Clock

  • 3U Special External I/O

  • 6U SBC

  • 6U Data/Control Plane switch

  • 6U External I/O

There are a number of secondary profiles that include simple variants of the primary profiles to address specialized requirements, as well as legacy profiles to support programs already underway. All have compatible requirements for features such as data plane protocols, power plane utilization, maintenance port implementation, GPIO implementation, and system management.

One interesting thing to note is that, unlike all other OpenVPX slot profiles, the primary SOSA profiles – other than two “External I/O” profiles – have no user defined pins. The External I/O profiles are needed to support connectivity to existing aircraft discrete signals, which can vary widely. This means that vendors are not free to create vendor-specific definitions for backplane pins, which will make system integration and later system upgrades much easier. This does not, of course, prevent suppliers from creating innovative new products with new, sometimes unique features. It will, however, help to break vendor-lock at both the board and integrator level by creating a market of provably interoperable or interchangeable board-level components.

SOSA’s Impact on Defense Designs

The impact of this hardware-level standardization in SOSA is likely to have far-reaching impacts well beyond the sensor platforms targeted in the standard. The OSA principles captured within SOSA for strictly-defined slot profiles, well-defined use cases, fixed definitions for power utilization, and so on, are equally applicable to most VPX-based systems. The power of OSA to enable easy integration and technology updates currently enjoyed by markets such as COMe and mITX will have a profound impact on the VPX market in the years to come.

As the SOSA standard continues to progress to release 1.0 late in 2019, the expectation is that defense programs will increasingly call for compliance and vendors will comply by continuing to expand their compatible product lines. It is anticipated that by the end of 2019 a full infrastructure of boards, power supplies, and backplanes designed in alignment with the SOSA standard will be available, and that integrators will be demonstrating their first conformant systems. Only time will tell if the standard has the impact on the community that everyone is estimates, but the various working groups are laboring hard to create a true VPX-base OSA.

This article was written by Mark Littlefield, Vertical Product Manager, Defense, Kontron (San Diego, CA). For more information, visit here .