Developing New Anti-Drone Radar Technology
Today’s drone threat calls for small and lightweight radars with excellent detection range and coverage. That’s already tricky, as most radars do either one or the other. In the planning of IRIS, a new drone radar system, Robin Radar Systems pledged not to make that compromise. That is a steep task from a technical perspective, without throwing in a small and lightweight form factor and full coverage into the mix. IRIS was designed to be easily lifted, transported and deployed and redeployed as needed, which is pretty unusual for most radar systems. Starting from scratch, that gave Robin Radar’s technical team only two and a half years to transform an almost blank sheet of paper, except for a few high-level specifications, into a working product.
The big question is, how do you make that happen?
Don't Reinvent the Wheel
The first thing to ask was: “Can it be bought off the shelf?” The answer, in this case, was a clear no. And definitely not at a reasonable price, so it was up to us to develop a purpose-built radar to reach upper sensitive balance.
The second question was: Can existing company technology be used? That was where we were able to speed things up. We decided to implement tech from our existing ELVIRA drone radar, which applies micro Doppler, and our MAX bird radar for its antenna and processing capabilities.
Robin Radar’s roots are in bird detection; in fact, the company name, “Robin” is an acronym that stands for: Radar OBservation of Bird INtensity. Detecting drones is actually much less of a challenge. Classifying them as drones, however, is another story.
The amount of birds populating the sky means that discrimination between birds and drones is an absolute must for any radar system worth its salt. This is where micro Doppler comes in—the return signal of any target is frequency shifted in accordance with the relative velocity of the target. That results in the well-known ‘Doppler effect’. This also contains all moving articulations on the targets, such as props. Our processing then decomposes the return signal into its separate Doppler components, in order to detect and distinguish the high velocity spin of propellers. The result is a very robust drone discrimination which cannot be easily counter-measured.
However, detecting drones is one thing; acting on the threat they pose is another. Years of experience have taught that radar is only part of the solution; other sensors such as cameras and RF detection provide additional target classification data. Then there’s jamming, spoofing, or EMP that can be used to accurately divert or eliminate a threat.
For many of these sensors and actors, a full 3D position is highly desirable. ELVIRA isn’t capable of that; it only provides latitude and longitude of the target. IRIS, on the other hand, is.
The receive antenna of IRIS consists of eight channels, each receiving the target return. The direction-of-arrival of the return signal, with respect to the antenna plane, causes small time differences between the signals of each channel. This becomes our well-known phased array technology, used to determine the elevation of each detected target. This technology was already developed for the MAX bird detection radar, so it could be transitioned to IRIS without too much trouble.
In order to measure micro Doppler profiles, a radar needs to spend quite some time on a target, in the order of several tens of milliseconds. Either the radar has to rotate slowly, or the beam must be wide in order to illuminate the target for long enough. The first leads to a slow update rate; the latter to loss of resolution, accuracy and antenna gain.
Dealing with compromises is something engineers are used to, but the sacrifice between antenna gain or update rate was just plain painful. Therefore, it was decided to use two back-to-back antenna arrays, both performing as a full radar system. This obtained a 1 Hz update rate while still rotating at 30 RPM, with a reasonable beam width and a high gain antenna. Packing two transmitters, each transmitting 10 dBW, and two receivers into one stationary radome did prove to be a challenge, though. Not so much because of the physical size, but the required RF isolation between the transmitters and receivers is very demanding. Hours of brainstorming, simulation, and experimenting led to a patented antenna design which hit these set requirements.
Eye on the Prize
After building and severely testing demonstrators, we became convinced that the concept was feasible. From that moment onwards, it became extremely important not to get distracted by new insights and abilities, but focus 100% on the project's predefined goals.
So we asked the question, what do we want? The answer was: We wanted IRIS to be the successor of ELVIRA, providing 3D, mobility and a strongly reduced cone-of-silence. We explicitly kept requirements regarding detection and classification range equal. No compromise.
Even with reuse of antenna technology and having overcome the hurdle of packing the tech into a small radome, many challenges still remained. One of these challenges was the vast amount of data produced by, in total, 16 receive channels. These signals contain the return signal of, for example, a car driving 100 meters in front of the radar as well as the turning of a 12-inch propeller at more than a 1-kilometer distance.
That’s an enormous requirement of the dynamic range—to process implied floatingpoint processing. Only FPGA technology is capable of performing this task in real-time. In IRIS, an Intel FPGA with hard floating-point DSP blocks capable of executing 1.5 TFLOPS is deployed, on which the majority of the signal processing occurs. The hard work of the embedded FPGA leads to a manageable data stream of only several gigabits, which hosts computer executing plot extraction, analysis and tracking.
Two full radar systems, a high-performance FPGA and mechanical direct drive, and all packed into a 60-centimeter IP66 enclosure, generates a lot of heat, so thermal design became the next challenge. For ease of deployment and lighter weight, every effort was made to exclude any heat exchangers. When it came to leading a careful design of DC/DC converters and optimizing the FPGA processing, every watt of heat counted. That’s why, throughout the system, the enclosures and antenna panels were also designed as heat sinks, spreading heat without adding weight to the system.
On the flipside of the product, there was the user interface to be considered. Software needed to be given as much attention to detail as the hardware because it's the face of IRIS, the hub with which users form their main impressions of the radar, so a lot of effort was put into it. It is generally acknowledged that ELVIRA was a little below the ‘state-of-the-art’ bar that was set. Therefore, with IRIS, a concerted effort was made to catch up, beginning with the development of a brand new, pleasingly smooth, web-based application. This interface provides the user with a clear and seamless view of the radar tracks, a background map and control of the radar - from anywhere in the world – as long as there is a network connection to it. Plus, a more consistent look-and-feel reduces the need for training operators.
Validate… and Validate Again
Even after thousands of hours of simulation on RF, electrical, mechanical, digital, and thermal design, there's no guarantee that reality won't throw a spanner in the works.
Physical tests on the finalized product are the real proof in the pudding. In the first place, these tests encompass the key requirements of detection and classification range, and are performed during numerous field trials in all kinds of environments and conditions. In these tests, both false negatives and false positives were assessed, because the latter is also critical to successful deployment.
A far larger part is the testing required for all applicable certifications: CE, Radio Equipment Directive and, in the future STANAG-4370 Mil-Spec. Extensive pre-compliance testing and years of experience with compliance tests helped to get it right on the first try for CE, IP and MIL-STD-810G vibration testing, executed at notified bodies.
Validation can be a layered and lengthy process. This upcoming year, all tests required by STANAG-4370 will be performed, leading to a full ‘Mil-Spec’ product. Pre-compliance testing on the key aspects such as EMC/EMI and environmental temperature point to a successful outcome.
This article was written by Gerben Pakkert, Director R&D and PEC; and Rob van der Meer, System Architect; Robin Radar Systems (The Hague, Netherlands). For more information, visit here .