Small UAS payloads pose SWaP and bandwidth challenges

Story

April 14, 2016

As military UASs continue to evolve and shrink in size – think swarms of tiny drones – their resulting payload footprints pose numerous tight SWaP design space constraints and tradeoffs, together with sensor processing, datalink bandwidth, and security issues as well.

“There’s a trend for ever-smaller [payloads] that’s driven in part by the trend to smaller [UAS platforms],” says Stuart Heptonstall, product manager of graphics and GPGPU Products for Abaco Systems in Huntsville, Alabama. “These smaller [platforms] must be lighter, nimbler, faster, and have longer ranges, which all affects the electronic systems within. Every square inch counts.”

Most applications are demanding increased functionality while maintaining the SWaP envelope, which translates to more performance per slot. “This means leading-edge system-on-chip (SoC) technology on our cards,” Heptonstall notes.

Other systems demand a significant reduction in SWaP, which “can mean smaller board form factors within the [payload], with optimized application-specific functionality,” points out Heptonstall. “The challenge in this scenario? Trading a certain level of flexibility to meet a challenging reduction of SWaP.”

Then there’s the issue of weight: The trend toward smaller [payloads] means “trends for the potential of smaller distributed systems,” Heptonstall continues. “These systems can be placed in smaller spaces within smaller [UAS platforms] and also have the potential for balancing the [UAS’s] weight distribution better. Stripping weight out is always a challenge that directly impacts the design of the cards within.” At the board level, as with size, lowering weight means “providing significant increments in performance per slot in applications where maintaining a SWaP envelope is required,” Heptonstall says. “Conversely, a less-is-more design approach can be deployed in significantly reduced SWaP applications. For these scenarios, certain tradeoffs may be required for wider COTS (commercial off-the-shelf) applications and homing in on an application-specific, weight-optimized design. This is sometimes necessary.” (Figure 1.)

 

Figure 1: The Abaco Systems ICS-8560 video-compression board XMC provides the minimal size and weight required by the constrained environment of a typical UAS.

(Click graphic to zoom by 1.9x)

 

In terms of power, performance is closely linked. “Balancing performance with thermal-power dissipation in small form factors and enclosures is always a primary challenge,” Heptonstall explains. “Endowing the system with industry-leading performance while managing the thermal envelope gives engineers headaches. Again, similar principles apply such as optimizing the board-level design by selecting the right SoCs for the application and optimizing the design and PCB for thermal efficiency. Both board-level design and mechanical-enclosure design must be closely aligned so that the electronics designers are cognizant of the thermal and mechanical challenges, and vice versa.”

Sensor processing for ISR payloads

Reduced SWaP also means packing more performance into shrinking design footprints to meet the insatiable demand for more sensor data in real time.

The U.S. military is “becoming more interested in fusing data and accessing it closer to real time,” says Robin Snider, director of electronic payloads for General Atomics Aeronautical Systems Inc.’s Mission Systems business unit. “You can get information much faster if you do the processing in near-real time onboard the aircraft and then get the data down the link.” (Figure 2.)

 

Figure 2: The Claw integrated sensor payload control and analysis software package is used on unmanned and manned aircraft to cross-cue a variety of sensors and payloads. (Photo courtesy of General Atomics – Aeronautical Systems, Inc.)

(Click graphic to zoom by 1.9x)

 

“Most larger platforms offer multiple modalities, so the ability to merge those apertures with onboard processing to improve situational awareness is highly desirable,” says Peter Thompson, director of business development and technology for Abaco Systems.

Military UAS users “are seeking actionable intelligence from their sensors in real time – whether the sensor is part of a radar, electronic warfare, or ISR (intelligence, surveillance, and reconnaissance) sensor chain,” says Shaun McQuaid, director of product management for Mercury Systems’ Embedded Products Group in Chelmsford, Massachusetts. “This intelligence then needs to be presented cohesively as part of a total situational-awareness package. These capabilities are derived from powerful digital and RF and microwave processing resources that are increasingly being mounted directly onto the platform itself. The issue is that these sensors are becoming so much more capable and more data is flowing; it’s both unadvised and virtually impossible to send all of that data down to an analyst on the ground.”

More processing at the sensor

What are some options to bring the processing closer to the sensor? UAS integrators “are approaching this in one of two ways,” McQuaid says. “One is that the highest processing performance possible should reside on the UAS, but the SWaP required to perform this may be restrictive – it’s platform-dependent. In some cases, we’re seeing our customers stack six or eight 3U modules onto platforms or pods. Alternatively, at a chassis level, it’s entirely possible to gain similar capabilities from just one or two 6U modules using the same SWaP footprint. There are many tradeoffs in risk, reliability, cost, pre-integration effort, and so forth.” (Figure 3.)

 

Figure 3: (Mercury) The rugged OpenVPX Ensemble LDS6526 processing blade from Mercury Systems integrates the Intel Xeon processor D-1500 system-on-chip (SoC) for intensive signal processing applications such as ISR payloads for UAS platforms.

(Click graphic to zoom by 1.9x)

 

Mercury Systems is also seeing interest in direct digitization – taking RF directly from the sensor and digitizing behind the sensor itself to avoid tuning and down-conversion, says McQuaid. “UASs require wide-spectrum bandwidth and the ability to digitize it on the fly. With this, you have an extremely broadband digital data stream emerging directly out of the sensor. This is made possible with FPGA processing that’s attached directly to the sensor and with efficient backend processing.”

As a result, [we see] a push toward “high-density, server-type architectures, which are designed to duplicate the server architectures you’d find in Amazon’s or Google’s data centers,” McQuaid continues. “They use the same Intel processors and similar general-processing architecture. Our customers are placing these processing capabilities right behind their sensors to perform server-class, big-data analytics. At the end of the day, what we’re really doing with these high-end sensors is solving a big-data problem – and it has to be solved on the platform because we don’t have the data link bandwidth or time to deal with it in a traditional ground-based scenario.”

Bandwidth challenges

Innovation in embedded computing and signal processing have enabled reduced SWaP and sensor fusion, but the downlinks for getting actionable intelligence off the payload are still stubbornly bottlenecked. To solve this, embedded computing companies and system integrators are driving the processor closer to the sensor to parse the vast amounts of sensor data being generated before it is downlinked.

Lack of bandwidth in downlinks is always an issue, but it’s one the industry has lived with for decades and has sort of come to grips with, notes Snider. “You’d think there’d be a continual struggle to get more bandwidth, and there is at some level, but the reality is that we can do almost everything with relatively small data link pipes,” he adds. “But there are new kinds of capabilities, specialized sensors with lots of pixels and data that small data links do limit, so the way they’re being handled now is by recording the data and post-processing it. In these cases, wider-bandwidth data links have value.”

“Video is a significant challenge,” says Abaco Systems’ Thompson. “As with all sensor-derived data, the trend is for resolutions to increase to provide more detailed information; that’s certainly true for video, where high definition is rapidly becoming the norm. The trick here is compression – JPEG2000 or more, increasingly H.264 (AVC/Advanced Video Coding), and in the future H.265 (HEVC/High Efficiency Video Coding).”

Securing UAS payload data

Security is a challenge for UASs because it’s extremely difficult to secure something on the tether of an RF link. “We’re working to improve security because threats within this realm are continuing to worsen,” says General Atomics’ Snider.

All UAS applications out there today are dealing with cybersecurity or system-integrity concerns, and security must be tightened because of the possibility of nation-states or other malicious actors taking over control of drones – or even worse, swarms of drones – by gaining access to their data or communications.

“Many nation-states would love to play that game, so we need to be very careful about what we’re exposing,” Mercury Systems’ McQuaid notes. “Secure processing is definitely the path to the future and soon may not be optional.”

Future of UASs

Looking forward, smaller and cheaper UASs are the way to go, according to Abaco Systems’ Thompson. “Expect to see more networking and intelligence on these smaller platforms to enable cooperative swarming,” he says. “Sensors will be bigger and better – with more dense focal-plane arrays and new modes such as through-the-wall imaging of the inside of buildings. This, of course, will drive data rates and the required compute power ever higher.”

Not surprisingly, SWaP will continue to play a central role. “Watch for a continued focus on SWaP reductions and greater moves toward leveraging the best commercial processing and software technology – especially from the data-center domain – and applying it to unmanned vehicles,” McQuaid says. “As a result, the tactical, mobile cloud-based architecture will gain further traction. By using a pool of computer resources that are made available, numerous unmanned platforms can tap into it to offload some of their sensor processing. Discrete platforms don’t have to handle the entire processing burden alone anymore … this is a form of swarm architecture and it’s starting to make its presence felt.”

Applications for UASs are expected to continue to grow. “We’re seeing new types of missions, such as using UAVs to monitor sonobuoy fields, which hasn’t been done before,” Snider points out. Another area to watch is the “radical trend toward vertical takeoffs and landings. “This is an area being actively explored for a whole new generation of UASs. There’s also a push for autonomy; initially, there was resistance to it, but now we’re seeing greater interest in having at least some level of autonomy on the aircraft. Eventually they’ll need to become more or less fully autonomous.”

 

Sidebar 1

(Click graphic to zoom)