FPGAs fuel military electronic warfare, cognitive radio, and radar signal processing innovation
Demand for increased signal-processing performance continues to be insatiable among military radar, electronic warfare (EW), and communications integrator especially as the latter two move into cognitive applications. In this Q&A with Rodger Hosking, Vice President and Co-founder of Pentek in Upper Saddle River, New Jersey, he discusses how embedded signal-processing can enable cognitive EW and cognitive radio, how FPGAs have revolutionized embedded computing, and how the defense industry is competing for young engineering talent with the behemoths of the commercial world. Edited excerpts follow.
MCHALE REPORT: What types of requirements are driving military radar signal-processing applications?
HOSKING: The military does not have a ‘good enough’ performance point in these applications. They want to be able to detect targets as far away as possible. For radar systems larger Fast Fourier Transforms (FFTs) are needed for more precise Doppler processing, where the radar systems track the speed and direction of targets. Improving resolution at a given range yields more accurate target information, including the size of an aircraft and detection of unique surface structures and reflection characteristics of a jet. This rich set of information in the received signal can help identify a unique target. It’s akin to looking at a fingerprint 10 feet away vs. examining it under a microscope.
Radar waveforms also get more complicated every year as military system designers want better algorithms to glean more information and to improve signal to noise performance so they can extract targets from clutter and noisy environments.
MCHALE REPORT: What factors drive signal-processing requirements in military communications?
HOSKING: The military wants wider bandwidth to handle the latest spread spectrum signals and more channels to handle increased traffic. Key goals are improving classification of radio types, boosting transmission range and noise immunity, and enhancing signal detection and exploitation. To achieve this, they need faster data converters to handle the multi-gigahertz sampling rates to digitize these wideband signals. This, in turn, means more digital signal-processing resources to perform the required algorithms in real time, a task ideally suited to the thousands of DSP blocks in FPGAs, all running in parallel.
Faster embedded system links are also in demand between data converters and FPGAs as well as higher data rates between boards within a chassis and higher data rates between systems and sub-systems. As digitizers and front-end DSP operations move closer to the sensors faster data transmission paths to these distributed acquisition sub-systems are necessary.
MCHALE REPORT: Cognitive is a term being used to describe next-generation radio and EW technology. What capability does the cognition bring and how does embedded signal processing enable it?
HOSKING: If you’re talking about cognitive radio, you’re talking about the ability to understand the surrounding environment, autonomously determine friendly signals from enemy signals, detect potential jamming efforts, and then maneuver transmissions to different frequencies to avoid the jamming attack. Often such operations are also termed “adaptive radio.” Because the enemy is always trying to deny communications and their efforts are becoming more sophisticated, cognitive radios need to have the capability to be a step ahead of the enemy and always improving over time.
Cognitive systems can react more quickly than humans and therefore counter the jamming attacks to restore the communications link with minimal downtime to stay one step ahead of any potential adversaries. A similar approach is being used for cognitive EW, where the cognitive EW system becomes more intelligent and agile in adapting to threats and interference.
To enable cognition you need extremely high-end signal-processing to enable all the computations necessary for fast and effective adaptation. The only place you will get that is from high-end signal-processing systems that leverage FPGAs and commercial general purpose processors.
MCHALE REPORT: How have reduced SWaP requirements affected signal-processing solutions for EW and radar systems?
HOSKING: The increasing need for real-time signal processing horsepower mandates the use of more DSP resources, which are delivered by each new generation of FPGAs. To help manage SWaP issues, silicon geometries shrink the size of each element so that the physical size of the new higher-density FPGAs remains about the same. In addition, the power dissipation per element drops, so the overall power dissipation of the new, higher-performance FPGA is the same or less than the previous generation.
MCHALE REPORT: Pentek has been around for 30 years; well before COTS procurement was introduced in 1994. What role does COTS play in today’s military procurement environment?
HOSKING: Due to recent budget cuts and sequestration, the DoD is no longer funding the development of technology from the ground up, instead placing the research and development spending burden solely on the contractors or through a shared cost effort. This is a smart move, and it has led to more COTS procurement. We would like to see DoD engineers becoming more directly involved with us in system design and product selection, reducing the tendency to out-source those decisions.
MCHALE REPORT: What do you see as the biggest difference between military procurement now and in the 1980s when Pentek was founded?
HOSKING: When we started, we were often dealing directly with the end customers in government organizations, who were designing systems for their own needs. And many of these customers preferred solutions compatible with Intel systems based on Multibus I and Multibus II. We negotiated specifications of the required board level products directly with these engineers and received purchase orders directly from their government procurement office. We supported those customers directly.
We shifted our new designs to VME as it emerged as an open standard that was quickly embraced by our government customers. A few years later, with the [former DoD Secretary Williman Perry’s COTS] directive in the mid-1990s, we saw a shift towards government out-sourcing for new designs. Instead of choosing board-level products, government engineers sent requests for proposals for complete systems-to-systems integrators, who would then select the components. Now, the vast majority of Pentek customer interaction, support, and procurement activity is with prime contractors or smaller integrators.
The benefits may be more cost-effective, competitive solutions, lower recurring costs, and easier insertion of new technology. Perhaps the downside may be less direct interaction between the end-user and the board designer.
MCHALE REPORT: The defense industry no longer drives technology innovation like it once did, as it is now a consumer of commercial electronics technology. That said, has it been more difficult to attract young engineering talent to the defense sector with the “Googles” of the world being a much better draw for young talent?
HOSKING: I recently attended the IEEE Phased Array event and it provides a good example of the challenge faced by the defense industry when it comes to attracting young engineering talent. Many of the younger engineers and engineering students who visit our booth are disappointed that we mostly only sell to military customers. They lament that while the only good steady jobs are in the defense sector either with the government or with defense companies, they are mostly war related and don’t mesh with their worldview. We tell them that radar doesn’t just mean military applications. For example, we sell our products to many different weather radar developers, both here and around the world. Nevertheless, some students seemed prejudiced against the military market and I think that traces back to the culture in American universities today.
Another troubling issue is the lack of American students – male and female – pursuing engineering careers. After attending American universities, many young engineers that we hire originally come from Asia. They are often excellent engineers and we highly value their contributions. Now, there are STEM programs to encourage young people trying to and especially young women to pursue not only engineering careers but in the math and science fields as well. We hope that effort blossoms into more American students pursing these professions.
MCHALE REPORT: What technology has been the biggest game changer for the military signal-processing world in the last 30 years?
HOSKING: It would have to be FPGAs. We’ve been doing the same things in terms of acquiring signals and then performing processing functions for the last three decades. At first we leveraged digital signal processors from Texas Instruments, then the PowerPC from Motorola, and then we shifted to FPGAs. In every case we were able to get what we needed from different technology.
Unlike alternative devices such as CPUs and GPUs, FPGAs offer many unique features that are absolutely critical for military and aerospace systems. Their re-configurability enables designers to arrange and interconnect hardware resources in parallel to match even the toughest real-time requirements. Today’s FPGA products have the ability to reconfigure elements under software control during a mission to adapt to evolving threats.
For example the Kintex UltraScale FPGA from Xilinx raises the digital signal processing (DSP) performance by more than 50 percent with reductions in cost by 39 percent and power dissipation by 18 percent.
They also provide configurable interfaces for wideband sensors like A/Ds and D/As operating at gigahertz sampling rates, as well as gigabit serial system interfaces like PCIe, Serial RapidIO, and Infiniband. Sophisticated memory controllers for DDR4 SDRAMs support read/write data rates as fast as 2.4 GHz. With these diverse, high-performance capabilities, FPGAs deliver complete acquisition, processing, and interfacing sub-systems ideally suited for EW and radar systems that require fast I/O, low latency, and computational intensity.
MCHALE REPORT: What will the biggest game changer for military signal-processing over the next five to 10 years?
HOSKING: The evolution of Systems-on-Chip (SoC). A system controller with high-speed connectivity – like 10 or 40 Gigabit Ethernet – requires a CPU to handle the protocols and tightly integrate operations of the FPGA for real-time processing. Remote signal acquisition modules based on SoCs can be located close to the antenna, perform digitizing and pre-processing tasks, and then deliver the payload data over fast optical links to the main system. They offer a really great combination and one-two punch for signal-processing systems in radar, EW, or even military radio communications.