Editor's note: GE Intelligent Platforms changed its name to Abaco Systems on 23 Nov. 2015 as a result of the company's acquisition last September by New York-based private equity firm Veritas Capital.
Embedded computing companies in the defense and avionics industries forced to do more with less due the current economic environment are providing greater functionality in a smaller, lighter, and less costly package by leveraging commercial off-the-shelf (COTS) technologies.
"Embedded processors are going the way of mainstream computing," says Mark Snyder, vice president of product management at Alt Software in Phoenix. Military and aerospace systems "are trending toward smaller, cheaper, lower power, and higher performance. They are morphing toward programmable multicore or parallel-floating point architectures, such as programmable GPUs [graphics processing units], and away from standard uniprocessor and dedicated ASIC [application-specific integrated circuit] architectures. The future holds much more of this."
A fundamental shift in which systems designers are replacing discrete systems and components with programmable counterparts is happening throughout virtually all avionics and defense applications. A system's ability to multitask, or fulfill several different roles is key.
Systems designers increasingly are requesting long-life solutions that either reduce or eliminate their dependency on discrete graphics boards and chips, Snyder acknowledges. They want to embrace programmable graphics accelerators for their graphics, video, and parallel-processing capabilities, but they also need to meet traditional military and aerospace requirements for safety certification, reliability, and robust operation.
Military and aerospace firms are turning to companies such as Alt Software to meet myriad GPU needs, such as powering embedded display, situational awareness, and, increasingly, parallel and sensor processing systems. "Our customers are requesting, for instance, capabilities to replace traditional methods of processing pixel-intensive information, such as DSPs and FPGAs, with software-programmable methods using standards, such as Khronos Group's OpenCL [and OpenGL ES 2.0]."
Driving displays
Engineers at Alt Software developed software drivers for the Fujitsu Ruby graphics display controller on Green Hills Software's Integrity real-time operating system (RTOS). "In this case, Alt was selected to bring OpenGL ES 2.0, the standard graphics API [application program interface} used on devices such as the Apple iPhone, to the mil-aero space on a hard real-time OS," Snyder explains. "Green Hills wanted their customers to be able to leverage the vast pool of developers familiar with these new graphics standards, and therefore open up mil-aero platforms to the latest capabilities enjoyed by consumers.
"Fujistu's Ruby graphics subsystem provides a discrete GPU core that offers an ideal bridge for bringing these capabilities into the existing mil-aero processing world," Snyder adds. "Ruby provides full programmability, meaning application developers can design innovative media-accelerated applications that offer new ways to process and display data, all in a low-power form factor that can be easily integrated into existing system architectures."
Many mil-aero systems designers are interested in system-on-a-chip (SoC) architectures or adding GPU cores to their own FPGA or software-based designs, Snyder says. "We will see the advent of more SoC systems, such as Intel Atom- and ARM-based SoCs, making it into mainstream mil-aero systems. Embedded computing will follow the path of our smart phones, iPads, and netbooks, and the mil-aero community will have key challenges to adapt their software and system development mentality to these trends."
Executives at Curtiss-Wright Controls Embedded Computing (CWCEC) in Leesburg, Va., are also seeing increased demand for GPU power. Mil-aero systems designers are "using high-end GPUs from Nvidia [Santa Clara, Calif.] and ATI [a brand of AMD in Sunnyvale, Calif.] for signal-processing applications," says Steve Edwards, chief technology officer at CWCEC. "Especially in radar signals intelligence communities, where they would have used banks of processors or FPGAs, they are investigating GPUs for signal processing." Innovation in this specific area, in fact, is high on CWCEC engineers' research-and-development list for 2010.
Graphics demands
"People are demanding more processor and memory performance in smaller footprints," explains Chip Thurston, technical director at Crystal Group in Hiawatha, Iowa. "Gone are the days of a 450 MHz, low-power XScale CPU and 512 megabytes of memory," he says. "Mobile and embedded applications are now looking to do full-resolution mapping, 3D imagery, and checking e-mail--all without multiple machines."
U.S. Navy officials are among the mil-aero systems designers requiring greater graphics performance. In the modernization of Navy destroyer warships, officials elicited the expertise of engineers at General Micro Systems (GMS) in Rancho Cucamonga, Calif.
"All the hardware used in the modernization of bridge machinery aboard the DDG-1000 [Zumwalt-class destroyer], for example, is from GMS," notes Ben Sharfi, GMS chief executive officer and founder. "It was a refurbishment and remodernization of the embedded electronics, which exclusively are VME-based, 6U, air-cooled, semi-ruggedized systems."
Today's Navy requires lots of functionality in a single box: graphics, performance, and storage, Sharfi says. The system must be powerful enough to handle huge databases, drive several monitors and repeaters, and accommodate RAID controllers. S.I.E. AG formerly Carlo Gavazzi) in Brockton, Mass., provided the chassis for these programs.
Size and weight, common priorities in mil-aero systems, are not terribly important factors when it comes to Navy destroyer modernization. "Adding another 20 pounds to a carrier is not that big of a deal," Sharfi explains. "What is important is graphics functionality and compatibility with legacy equipment -- what they have on the ships already."
GMS was selected in part for its willingness to modify its hardware to fit the Navy's existing infrastructure. "We knew the infrastructure and backbone were not going to change," Sharfi says. "We provided both flexibility and customization. We modernized, but left the physical interface the same as what was already in place. If you look at most hardware makers, they are going with only new stuff."
The demand for increased graphics functionality is not exclusive to the Navy. The U.S. Coast Guard, Department of Homeland Security, and law-enforcement agencies likewise seek high-end graphics and video capabilities.
Curtiss-Wright Controls is fulfilling this need with rugged video display, distribution, and recording technologies that the company gained with its acquisition of Skyquest Systems of Basildon, England, in Dec. 2009.
"More and more, our customers want us not only to provide the box, but also to hook it up to a rugged display," CWCEC's Edwards says. High-end graphics and video are pushing into rugged deployments, providing opportunities to deliver high-definition avionics displays in helicopters, several camera sensors outputting to video recorders for immediate archival, and video distribution and displays solutions.
CWCEC engineers are currently working with police agencies and the Coast Guard to deliver video surveillance technologies. "We are also bringing this technology to [the U.S. Department of Defense] to meet the demand for 360-degree awareness on the battlefield," says Michael MacPherson, director of business development at CWCEC.
Solid-state storage
The increased need for video and graphics content is driving the requirement for greater data storage capacities. All the intelligence data being captured by myriad sensors must be stored, after all.
In response to increased storage demands, the aerospace and defense electronics community is rapidly moving to solid-state disks (SSDs), says Thurston. "As the commercial market has pushed capacities upwards, mil-aero customers have jumped at the chance to move long-term storage of applications and data onto solid-state disks."
Costs are coming down, and mil-aero program managers and end users want SSDs, Sharfi mentions. An important aspect, however, is secure erase capability—an area in which GMS and other firms are innovating.
CWCEC engineers are putting a lot of resources and intelligence around not only information security and data protection, but also the notion of trusted COTS; that is, the ability to protect the technology. "There is a need to protect critical technologies used in unmanned systems as part of foreign military sale," MacPherson cites as an example. "We are adding features and capabilities to enable the protection of data and the technology itself."
Multiple cores
The future will bring embedded and mobile applications having better performing video, as the video controller will likely move onto the CPU (central processing unit), Thurston says. "More system functionality will eventually be brought into the processing unit, which should allow for lower power requirements, more computing capability, and smaller devices.
"Rather than increasing processor clock speed," Thurston adds, "over the past couple years, we have seen an increasing number of cores present in the processor, and new technologies—such as the introduction of additional thread handlers via hyperthreading, and the increase of memory bandwidth by putting the memory controller directly on the processor itself."
Crystal Group's latest rugged embedded computer, the TCM2 (Tactical Computing Module 2), takes advantage of a dual-core 2.53 GHz Intel Core 2 Duo CPU. The small-footprint (11x12.75x3-inch), high-performance embedded computer is designed to withstand aggressive temperatures (-40 to +65 degrees Celsius) and 7.18 GRMS of random vibration, as well as run on conventional 120-volt vehicle power. It also offers up to 8 gigabytes of RAM and two 2.5-inch hard drives or SSDs, expandable to as many as eight additional 2.5-inch drives with a slim expansion base.
The hottest multicore news of late is the 2010 Intel Core processor family, including the Intel Core i7, born of the company's 32-nanometer production process. Following Intel's release of the series during the 2010 Consumer Electronics Show, technology firms serving the mil-aero community -- including Adlink Technology in San Jose, Calif.; CWCEC; Extreme Engineering Solutions in Middleton, Wis.; GE Intelligent Platforms in Charlottesville, Va.; and Kontron in Poway, Calif. -- introduced embedded innovations that take advantage of the 32-nanometer processors.
Kontron announced the integration of the new Intel Core i7 processor across numerous platforms found in military design, including AdvancedMC, CompactPCI, COM Express, and VPX, describes Thomas Sparrvik, vice chairman of Kontron. "Using the 32-nanometer manufacturing process for exceptional performance per watt and lower power consumption and heat dissipation, Kontron's new Intel Core i7-based platforms utilize a more efficient two-chip solution that provides better signal integrity and minimizes board space, enabling higher performance in smaller, power-constrained portable designs."
This technology also delivers enhanced integrated graphics capabilities and data flow performance via the integrated Intel QM57 Express chipset and advanced display interfaces, Sparrvik adds. It is, therefore, well suited to visually demanding and compute- and graphics-intensive military applications.
"Applications can now support multiple graphical and multimedia functions," says Sparrvik, outlining further benefits of Intel Core i7 platforms. "Military system design flexibility is increased with an integrated ECC memory controller to match high data integrity requirements. Design flexibility is further improved with additional I/O (input/output) and PCI Express configuration options, and long embedded system life requirements are supported with an extended seven-year lifecycle. Military designers can capitalize on the intelligent, feature-rich Intel Core i7 processor architecture and also satisfy challenging multiple requirements in terms of performance, power, graphics, memory, software compatibility, security, and upgrade path migration." Virtually all these requirements apply to computers embedded within combat vehicles, for instance.
One from many
Virtualization is enabling the replacement of several, preexisting systems with a single embedded computer.
"Win-T removed four servers and replaced them with one GMS box. Why that works is simple: virtualization," Sharfi explains. Many mil-aero organizations are running old software -- legacy mapping, diagnostic, communications, and other -- on Pentium M processors, he continues. "Now, with multicore, we have more than enough horsepower to virtualize multiple machines in one machine and have plenty of bandwidth. With virtualization, multiple systems can be removed and a single, small, inexpensive, high-performance embedded system brought in. It is the focus of the U.S. Army, for example. You'll see that trend continue and many companies do well on this platform."
Virtualization not only frees up space, but also lessens a platform's weight -- the top identifier for the U.S. Air Force in its modernization efforts. For every thousand pounds of weight removed, hundreds of thousands of dollars in fuel are saved, Sharfi says. "Weight is king in aircraft. Thousands of pounds of weight can be cut from an aircraft simply by going outside of the traditional way of doing things -- such as racks of electronics being replaced by embedded computers that are virtualized." When 12 images run on one system with virtualization, the platform benefits not just from the removal of 1300 pounds of weight, but also from reduced power consumption and cooling needs over the lifetime.
CWCEC officials are "looking at LRU [line replaceable unit] consolidation," says MacPherson. "As modernization programs put more capabilities in platforms, space is not available to add more LRUs. [Military systems designers] want to replace multiple LRUs with one LRU incorporating greater functionality than all the removed systems and taking up half the space. Two enablers are: an increase in processing performance and a consolidation of I/O. Now only two interfaces share information; it used to be 5 to 10."
Adding value
Just as it is now common to replace several systems with a single solution in mil-aero environments, systems designers are trending toward acquiring complete systems rather than components from several sources. Most also now seek "value-added" services.
"Embedded board vendors that offer higher levels of integration in the form of functional subsystems with fully-developed FPGA code, software drivers, and diagnostics will become more attractive to system integrators as a way of minimizing their risk and costs in delivering the final system," says Rodger Hosking, vice president of Pentek Inc. in Upper Saddle River, N.J.
"Customers are working on increasing their value proposition, and asking us to increase ours as well," says CWCEC's Edwards. "As a result, we are seeing more integrated systems sales -- for example, a complete subsystem with chassis and power supply, integrated solutions that stop short of the application."
CWCEC's built-in test (BIT) is an example of a "value add," says Edwards. "When the card powers up, it does a self-check and reports any failures, such as in communications, memory, or between nodes on a board. It really stops at the backplane. It is systems-level, not just modular-level, BIT with some interboard, intraboard, and backplane connectivity testing to ensure that everything is working together and functional."
Interest in integration
It is not enough to ensure that a component or system is working; rather, it must be capable of working in concert with various other electronics -- and do so every time, without fail.
"Customers continue to demand embedded solutions with higher levels of subsystems level integration that actually work together, and as advertised," says Douglas Patterson, vice president of worldwide sales and marketing at Aitech Defense Systems Inc. in Chatsworth, Calif. "In addition, they're looking for flexible levels of customization services and value-added, 24/7 technical support services when something they think ought to work, doesn't."
The basic physics of customer applications have not changed, says Patterson. "We are still getting requests for full MIL-SPEC [-55 to +85 degrees C] products combined with smaller size, lighter weight, and lower power. This encompasses integrated electronics control subsystems housed in lightweight, environmentally sealed, ambient/passive air or cold-plate [conduction-cooled] rugged enclosures." Aitech is expanding its lower-power, high-performance board and subsystem-level products to address this need, most recently with its NightHawk RCU series.
The NightHawk RCU is a rugged, compact Intel Atom-based, self-contained control unit that weighs 4.5 pounds, has a slim profile, and delivers natural convection cooling. The solution is suited to use as a data concentrator unit (DCU) and remote interface unit (RIU) in various military platforms, such as manned, unmanned, ground, and airborne vehicles. The NightHawk RCU can also provide Condition Based Maintenance (CBM) functionality for military tracked and wheeled vehicle applications, reducing the overhead costs of preventative vehicle maintenance.
The newly released NightHawk, based on the low-power Intel Atom processor operating at 1.6 GHz, provides as much as 2 gigabytes of DDR2 SDRAM as well as between 4 and 8 gigabytes of SSD memory with an optional expansion to as much as 250 gigabytes for extended and remote data collection and storage applications. Its I/O interfaces include two Gigabit Ethernet ports, six USB 2.0 ports, and four multifunction RS232 serial ports, dual graphics/video ports, keyboard/mouse and stereo audio in/output ports, and an I/O set specifically tailored for embedded military applications.
Network connectivity
Network-centric warfare and the Global Information Grid (GIG) have matured, and military programs are starting to require network-ready systems. "The network is part of the systems that are deployed," CWCEC's MacPherson explains. "Now that we understand netcentric warfare, we need to make sure all systems on a battlefield can connect and all products are able to form a network on the battlefield."
To be effective, a netcentric battlefield requires considerable, readily available bandwidth. "Military design is evolving dramatically based on demands for increased bandwidth and faster, more sophisticated signal processing," Sparrvik says. "The military is focused on using highly reliable technology to its fullest potential, as demonstrated in ongoing initiatives such as WIN-T and the U.S. Navy's Consolidated Afloat Networks and Enterprise Servers (CANES) program.
"WIN-T is fundamental in defining and advancing the military's secure network communications, with incremental implementation bringing greater levels of networking capabilities to various deployed units and ground command operations," Sparrvik adds. "Full network mobility and more robust connectivity enable greater network access than ever before. CANES, for example, is consolidating and reducing the Navy's afloat information systems networks. Technologies integrated into a CANES program application reduce the size and cost of its technology infrastructure, improve the existing path of legacy applications, and advance command, control, communications, computers, surveillance, and reconnaissance (C4ISR) capabilities in the process."
Programs such as these require advanced technologies that not only deliver performance and low power improvements in smaller footprints, but also meet critical system requirements for high reliability, Sparrvik explains. "High-reliability demands in mil-aero applications will mean that embedded computing platforms will need to leverage technology advancements along with satisfying more stringent needs for thermal management, mean time between failures (MTBF), and enhanced ruggedization. This translates into additional opportunities for embedded computing solutions as well as greater demand for embedded systems."
Data transfer
Pentek's Hosking notes a definite shift in new systems toward serial fabric-based system architectures, using PCI Express and Serial RapidIO, to improve board-to-board data transfer rates because of higher signal bandwidths, more powerful FPGAs and processors, and faster peripherals.
Pentek's digital signal processing IP cores for FPGAs are among the highest-performance designs in the industry, according to Hosking. "For example," he says, "Pentek's Model 7151 with four 200 MHz, 16-bit A/D converters features an FPGA IP core that delivers 256 software radio digital down converters (DDCs) in a single PMC/XMC module. As the industry's highest-density DDC, the 7151 is ideal for surveillance of hundreds of communication signals in a small form factor, including unmanned land, maritime, and airborne vehicles."
"The latest FPGA technology and serial fabric interconnects will dominate most designs," Hosking says. "These technologies will increasingly be deployed in PC server systems and in VXS and VPX platforms. Government concerns about the longevity and maintainability of new technology and architectures for embedded systems have mandated the need for industry-wide standards. One notable response has been the OpenVPX initiative and its relatively rapid transition to VITA 65 for ratification and eventual ANSI [American National Standards Institute] approval."
Standards and scalability
"For mil-aero, meeting SWaP [size, weight, and power] requirements continue to drive new designs toward smaller embedded computing form factors, such as 3U VPX, 3U CompactPCI, MicroTCA, and Computer-On-Modules (COMs)," Sparrvik says. "Satisfying program lifecycle requirements, reducing time-to-market, and minimizing the engineering resources needed for a particular application are also very important for mil-aero contractors to remain competitive. Because there are many embedded computing platform options and each brings its own advantages, designers of mil-aero applications are finding it crucial to work with a proven COTS manufacturer to find the optimal platform for their specific design.
"Mil-aero system developers are demanding the ability to scale solutions and add features without dramatically changing the form factor to meet their development and time-to-market goals," Sparrvik adds.
One of the big trends in embedded computing for mil-aero applications in the past few years is the move from VME to VPX, and from VPX to OpenVPX, says CWCEC's Edwards. "It has opened a range of applications, not only in terms of the size of systems, but the things you can do with the systems now that you have that much communications bandwidth between the systems. We've seen a number of our customers moving in that direction, especially in a majority of ground-based systems."
VME is still strong in deployed systems, in which the technology is upgraded by the whole chassis is not replaced; however, new platforms take advantage of VPX and OpenVPX, explains Edwards.
CWCEC engineers have delivered the company's first system to be built with VPX components, in fact. Company personnel delivered the radar processing subsystems to Northrop Grumman in Baltimore, Md., for use in the U.S. Marine Corps' Ground/Air Task Oriented Radar [G/ATOR] program and in accordance with a $4.3 million contract.
The High Mobility Multipurpose Wheeled Vehicle [HMMWV]-mounted G/ATOR uses active electronically scanned array technology to provide aircraft detection and tracking, cruise-missile detection and tracking, ground-weapon location, and air-traffic control. Its modular architecture is designed to deliver operational flexibility and the ability to incorporate new processing platforms and technologies as they become available. The rugged, air-flow-through, radar processing subsystem from CWCEC employs open architecture-based standards and software to provide a modular, scalable solution.
"We sold the whole subsystem to Northrop Grumman for the radar, which sits on the back of a mobile combat vehicle," Edwards explains. "We had to custom design a card to handle their proprietary I/O; other cards were COTS. We designed the chassis, power supplies, and a novel cooling approach, as well as qualified systems for their application to sit on--marking a number of firsts for us."
