By Dr. Arik D. Brown
Counter unmanned aerial system (C-UAS) mission capabilities protect maneuver forces across the globe. They provide situational awareness and neutralization against unmanned aerial systems (UAS). The proliferation of UAS and their suitability for nefarious intentions has made them a high-priority focus area for both commercial and DoD applications.
The Imminent Dangers of UAS
Off-the-shelf UAS (nano drones) are inexpensive and easily procurable, providing an advanced level of capability for invading protected air spaces that have critical infrastructure and personnel that need to be defended. These nano drones typically include cameras that allow unwanted surveillance of areas of interest.
Larger UAS have increased travel distance and cargo carrying capability, making them more dangerous. For expanded lethality, UAS can easily be retrofitted to carry kinetic weapons and non-kinetic payloads. This can lead to loss of life, damage to structural assets, or denial of service for communications and radar systems.
C-UAS Detection & Countermeasures
As UAS threats have continued to evolve, the DoD has developed a variety of detection and countermeasure systems. There has been a concerted effort to commonize the acquisition and fielding of systems so that warfighters have access to the most effective, expeditionary, and easy-to-use technologies. In support of this, JCO has recently released the recommended “interim C-sUAS list.”
Deployed C-UAS systems such as IM-SHORAD for the U.S. Army and GBAD for the USMC use a multi-domain sensor and payload approach for a complete, integrated C-UAS solution. Typical systems like these employ both a Radar and an EO/IR sensor which work in tandem. The operator can verify radar detections that have been classified as a UAS with the EO/IR sensor imagery. This requires man-in-the-loop verification, which places a heavy burden on the operator.
Potential Improvements with Artificial Intelligence
Classification powered by artificial intelligence (AI) would provide relief to the operator by automating the classification verification. This would be of significant value to the operator, freeing up time to do other mission-critical activities and reducing the decision timeline for threat neutralization. This capability would provide a significant advantage in maintaining battlespace dominance for defeating asymmetrical UAS threats.
Current C-UAS Classification Solutions
The primary payload for an effective C-UAS system is a radar. The radar must be able to:
- Detect, track, and classify all classes of UAS threats
- Render 360º coverage
- Operate in all-weather environments
- Provide clear situational awareness for the end user
Target Differentiation Challenges & Solutions
A challenge for all C-UAS is differentiating between biological targets (birds) and man-made targets (drones). Birds fly at speeds similar to drones and also have a comparable radar cross section (RCS). This creates a source of ambiguity for the radar tracker and can lead to false positive detection. Luckily, there are a number of existing solutions to overcome these challenges.
Micro-Doppler
Some radars have micro-Doppler capability that minimizes the ambiguity presented to the tracker, thereby improving performance. The Doppler signature must be observed over time and then used to discriminate against undesired targets. Ideally, micro-Doppler would enable the radar to be the only sensor required for classification. However, while micro-Doppler is effective, it does not provide a classification confidence level that is 90% or greater by itself.
EO/IR Sensors
To provide a more capable solution, many C-UAS systems employ EO/IR sensors in conjunction with a radar. They are used in a “slew to cue” mode, being cued by the radar toward tracked targets.
Human operators must verify that the radar track has been appropriately classified, making the system effective, but not optimal for the user. An autonomous classification capability that requires little-to-no human intervention (such as AI) would be a tremendous advantage to a C-UAS system.
Advanced Classification / Discrimination Approaches
RADA USA is currently investigating techniques for advanced C-UAS classification/discrimination approaches. They provide the same end result of accurate classification of drones in the presence of birds, but by slightly different mechanisms.
Advanced Classifier Approach #1
The first approach uses data science analytics applied to real-time track data for drone classification without an external sensor. This technique would enable the radar to operate in a C-UAS system without an external EO/IR sensor working in tandem and/or provide an EO/IR sensor with high-confidence drone tracks. Data science analytics can potentially reduce bird tracks substantially and provide an uncluttered view of a radar display.
Advanced Classifier Approach #2
The second approach employs AI technology applied to EO/IR captured video for classification of tracked targets. AI algorithms can be used to classify video imagery and then update the radar track if it is a false positive (e.g. a bird). The track data output presented to the C2 system would then be comprised of primarily high-confidence drone tracks. This minimizes the burden of the operator and aids in increasing the overall decision timeline.
Concepts of Operations for Advanced Classification Systems
The first interaction with the UAV occurs at the max detection range of the RADA USA radar system used. RADA’s RPS-42 and RPS-82 are capable of detecting nano-drones at 6 km and 10 km ranges, respectively.
After detection, the UAV is tracked by the radar. The radar passes the track to the data science analytics algorithm or cues the EO/IR sensor on the UAV and captures the threat with video imagery (EO and/or IR). If the tracked threat is not a UAV (e.g. a bird), then that track is ignored, and the system reports a non-target. If the tracked threat is classified as a UAV, then several options are available:
- Direct warning alert to the operator
- A track report transmitted to the radar and/or Command and Control
- Passing of the tracked target to a kinetic or non-kinetic payload for neutralization
In addition to classification, these techniques can also provide threat intent. This is another benefit to the decision-making timeline of the operator.
Artificial Intelligence & Range
The range at which the AI algorithms can be used is dependent on the EO/IR sensor. Images taken at 1920×1080 resolution with a field of view (FOV) less than 2º can be used for AI classification at ~1-1.5 km. Higher resolution image systems can be used out to and beyond 5 km.
RADA USA Radar Technology
RADA USA has developed and fielded air defense radars for many years, not only for US DoD services (CONUS and OCUNUS), but also in extended use in highly-volatile areas like the Gaza Strip. Their effectiveness for C-UAS missions is unparalleled, providing 24/7 detection capability for the most aggressive threats.
Key attributes for the selection of RADA USA radars for C-UAS have been:
- AESA, software-defined architecture that provides multi-mission capability from a single radar
- S-band operation for all-weather performance
- Commonality with U.S. forces (fielded and in use by USMC, U.S. Army, USAF, USCG, USN, and special agencies)
- Approved for use on the JCO “interim C-sUAS list”
- Extended coverage patterns (360 degree/full hemisphere), making them very capable for top-down, close-range, and high-elevation threats
- Smallest SWAP available
- No auxiliary cooling
- Capable of utilizing most on-board vehicle power systems
- C2 interface open architecture (use in FAAD-C2, MEDUSA, and others)
- Capable of operation “on the move”
- Superior price/performance ratio
RADA USA’s Future Development Efforts
RADA USA’s objective is to perform initial technology development of these advanced classification approaches in the first quarter of 2021. Only one of the approaches will be selected for the initial development effort. The overall development is expected to be broken into 4 phases:
- Proof of Concept
- Algorithm Development
- Algorithm Modification/Enhancement
- Final Test and Demonstration
The initial proof of concept phase is expected to be a 6-8 week effort requiring ~300 hrs.
What’s Next for C-UAS Advanced Classification Approaches?
RADA USA radars are currently being used in theater for C-UAS missions and have a long history of
being used by U.S. DoD customers as the primary sensor in C-UAS systems. The integration of data science analytics and/or AI provides a highly-capable and advanced technology solution for defeating UAS threats. We look forward to the innovative future of C-UAS missions and providing increased safety for warfighters.