Critical systems on small platforms need ultra-compact computing

1Defense imaging systems are beginning to deploy advanced sensor technology on small, often unmanned, platforms. Constrained by platform size, systems designers must still transform an unrelenting sensor data stream into information that fits into the bandwidth available for data transmission. To meet this challenge, designers need a new generation of rugged, ultra-compact, and powerful real-time computers.

Sophisticated, sensor-based imaging systems have become critical military assets, providing invaluable intelligence on battlefields around the globe. However, the timely use of that intelligence is impacted by bandwidth limitations in the data links that provide the transmission backbone from a sensor platform to a ground station.

Improvements in data communications will not be sufficient to meet this challenge, as systems will be overwhelmed by the information data streams from new generations of sensors. Hyperspectral Imaging (HSI) and Laser Radar (LADAR) will augment, but not replace, the Electro-Optic Infrared (EO/IR) and Synthetic Aperture Radar (SAR) sensors used today. As data links increase in bandwidth, they will continue to lag behind the breadth and depth of new, sophisticated sensors.

Imaging systems can address the challenge by using computing technology to make better use of existing data-link bandwidth. Processing power co-located on the sensor platform can be used first to turn raw data into images, then for image compression, and, at the most sophisticated level, to execute image exploitation algorithms such as change detection in comparing two images. Each level requires more computing power but enables the data link to be used more efficiently to transmit useful information.

While sensor systems are always hungry for more compute power, another trend makes it difficult to deliver: Sensor platforms are becoming increasingly smaller. Over the past decade, defense forces have found Unmanned Vehicles (UVs) with sophisticated sensors to be extremely valuable intelligence-gathering assets, carrying out long missions without crew fatigue and, most importantly, greatly reducing risks to military personnel. Now that UVs have proven their value, the current trend is to build new generations of smaller vehicles, including airborne, ground-based, or undersea platforms.

To maximize small-platform effectiveness, they must carry the same sophisticated sensors currently deployed in large platforms, not just basic imaging systems such as the video cameras that currently constitute many airborne payloads. Technology organizations are, in fact, making smaller sensors – and now they need ultra-compact computing resources next to the sensor so data can be turned into images that can fit onto the available transmission bandwidth. Packaging technology is also key in these ultra-compact computers.

Requirements for small, deployable computers

The trend toward smaller, unmanned platforms is driven partly by cost, but more significantly by changes in threat. Next-generation platforms will operate in a more lethal battlefield environment brought about by the worldwide proliferation of advanced detection technology. Using multiple, smaller platforms offers a greater likelihood of system survivability than a single, large platform.

While needs vary across a range of implementations, requirements for these next-generation embedded computing systems can be summarized as follows:

On the order of 100 GFLOPS

Systems can be implemented with less than 100 GFLOPS of processing power, but image-exploitation algorithms such as change-detection, geo-registration, or automatic target recognition demand that level of processing or more.

Less than 10 pounds

Case in point: There is a tactical category of smaller unmanned aerial vehicles with a total payload capacity ranging from 60 to 200 lbs. In a general sense, it is reasonable to allocate up to 10 lbs. of that payload capacity to computing, but not much more.

Smaller than ATR-sized

The ATR system for standardized electronics packaging evolved to meet the needs of deployment in manned aircraft. New generations of UVs are not built to fit human dimensions, so it is not surprising that the ATR form factor, even in its half-ATR short form, is simply too big.

Flexible enough to support a range of I/O protocols

An embedded computing system that is processing sensor input must be flexible enough to support multiple types of sensors. Sensor payloads can change from one type to another, or use multiple types within one payload. For example, an EO/IR set of sensors used for a daylight operation may be replaced by a Thermal Night Imaging System (TNIS) for a mission in darkness, while a SAR sensor is used simultaneously for both missions. Supporting this type of sensor flexibility translates into a need for a mission computing system that can support multiple I/O protocols.

Surviving extreme environments

Defense electronics systems must perform in harsh environmental conditions including excessive heat, humidity, poor air quality, high altitude, shock, and vibration. These embedded computers must not overheat, even when temperatures range up to 55 °C and the air is too thin to be used for cooling. At the same time, they must possess the mechanical integrity to withstand high shock and vibration forces. A sample shock specification for rugged applications is 50 g z-axis, 80 g x-y-axis, 11 ms half-sine, with a corresponding vibration specification of 0.04 g2/Hz, based on 20-2,000 Hz, 1 hr/axis.

Small form factors meet the challenge

While the requirements are formidable, advances in processing elements can be utilized to create solutions. Multicore processors and FPGAs can be used to accelerate computing, often in a complementary fashion. New computing enclosures, rugged, sealed, and using liquid-cooled sidewalls, allow these processing elements to be deployed in extremely small spaces.

For example, the recently announced PowerBlock50 from Mercury Computer Systems (Figure 1) is a fully integrated, ultra-compact embedded computer. To meet the first requirement of sensor computing – high performance – it features a modular architecture that allows for flexible configuration of multiple processors, delivering well over 100 GFLOPS of processing power via PowerQUICC III, Xilinx Virtex-4, or Intel processors. A PCI Express switch-fabric backplane is used for high-bandwidth interconnection between modules.

21
Figure 1
(Click graphic to zoom by 2.0x)

This processing power is packed into an ultra-compact and lightweight package. A fully configured PowerBlock50 weighs less than 7 lbs. and measures only 4.1" x 5.3" x 5.8" (105 mm x 134 mm x 148 mm). It can be held comfortably in one hand. State-of-the-art liquid cooling efficiently removes heat at any altitude, and rugged features include locking modules for shock and vibration immunity, EMI isolation, and more.

Future view: Small platforms, small computers

A new generation of small, defense-imaging platforms demands correspondingly small computing systems. The challenge is to meet this demand without compromising performance, flexibility, or survivability. Creative designs using new computing and small form factor packaging technology make it possible.

Thomas Roberts is product marketing manager at Mercury Computer Systems, Chelmsford, MA. He has more than 25 years of experience in systems engineering and technical marketing with IBM, Nixdorf, Compaq, Data General, and Digital Equipment. Thomas has a B.S. in engineering from Cornell and an MBA from the University of Kansas. He can be reached at troberts@mc.com.

Mercury Computer Systems, Inc.
978-967-1291
www.mc.com