RF detection and tracking can have limitations. A solid understanding of these limitations will greatly aid in the development and deployment of AeroShield into a larger system. Some RF weaknesses include:
• Noise and interference from sources using the same frequency bands. See Noise and Interference.
• Weak signal strength for distance drones, and the inability to identify a signal as uniquely a drone, rather than other similar transmission sources.
Reflections and Multipath
Multipath is caused by RF reflections and results in two instances of the same signal to show up at the receiver at different times. Reflections can be caused by:
• Large metal surfaces
• Buildings
• Mountains and hills
• Bodies of water
Usually there will be a direct path of the RF signal to the receiver, which is the signal that we are looking for. When a signal is reflected to the receiver from one of these objects, we see double. Depending on the relative paths, we either see a spread out signal, or two distinct signals. The weaker of the two will be in the noise, and the stronger signal will be detected.
Multipath is a primary cause for uncertainty and scatter in TDOA measurements. If wildly varying location estimates are found, multipath is the likely culprit. The best solution is with antenna placement, since it is often impossible to remove the reflective surfaces.
Noise and Interference
The drone video signal is most often found in the 2.4 GHz or 5.8 GHz ISM bands. These bands are heavily used and may contain a lot of competing signals. Good RF detection requires a certain threshold transmission power over the noise floor. If the noise floor is high, then detection is problematic. Noise can cause two related problems.
1. Due to noise discrimination, we may not be able to see a drone signal over the noise floor.
2. Noise may trigger the detection algorithms causing positive detection alerts.
Minimize the Effects of Noise
AeroShield minimizes the effects of noise by providing user defined tools to:
• Create a mask to detect a drone signal on the power spectrum and only look at signals that violate that mask. A signal mast have a certain width to be of interest. This is user selectable, but typically 6 to 8 MHz. Drone video will normally be either 10 or 20 MHz wide. Because sub-channels in the ISM bands overlap, band edges may not be distinct, and we may not see the full width of a signal. So we set the threshold width lower than the expected signal width.
• Set the suspected signal power level of the mask violation to be a certain level above the mask.
Transient Signals
The signal must persist for a number of measurements. Many transient signals come and go in these bands, but the drone signal should be fairly stable during the course of the flight. (Note: RF measurements use a short duration max hold on the spectrum power level. This provides visibility to low duty-cycle drone signals while discriminating against truly transient signals).
The user controls for noise by adjusting the settings appropriate for a particular deployment environment:
• Mask Depth: A deeper mask collects RF signal over a longer time to capture transient signals and avoid positive detections.
• Mask Violation Width: Set this wide enough to discriminate against narrow signals, but narrow enough to see true drone signals, even with potential channel band overlap.
• Transient Signal Filtering: This controls how long a signal must remain present before we begin investigating it as a potential drone. If you have a fairly clean environment, low filtering will help detect signals as quick as possible. If the RF environment is noisier, then a longer setting (High) will help illuminate triggering on a lot of signals.
Wi-FI Traffic
Drone video signals occupy the same bands as Wi-Fi, both 2.4 GHz and 5.8 GHz ISM bands. These bands are heavily used and may contain a lot of competing signals. WiFi causes additional problems as they submit:
• Interference
• Modulated signal sources traceable using TDOA
We cannot distinguish between a Wi-Fi signal and a true drone signal using the RF power spectrum. AeroShield attempts to distinguish in that it normally requires a detected source to move in order to create a detailed tracking event. A stationary drone and a standing person using Wi-Fi on their phone will look the same to AeroShield.
Ignore List
In order to minimize the number of positives, AeroShield utilizes an Ignore List for signals that it has detected but has determined are not drones. Signals that go onto the Ignore List remain there for several minutes. If the signal remains present for a period of time, it will be ignored by AeroShield until its entry on the Ignore List expires. The expiration time is user selectable. The default value for the Ignore List is 2 minutes.
It is not advisable to set the expiration time to a very long duration value to stop detections. If a Wi-Fi signal turns on, it will likely look for an available channel in the ISM band. Once that signal leaves, the frequency is again available, and is a likely candidate for a drone controller to select. If a signal persists for over 2 minutes, it is worth taking a few seconds to reevaluate it. It is generally better to evaluate a number of potential signals, even though they prove , than to be too stringent and miss true drone signals. Further, this is why RF detection is one part of a larger drone detection system. Other components need to be brought to bear on a suspect signal to determine if it is real or something to ignore.
System Deployment Requirements
System requirements are critical to a successful deployment. AeroShield has stringent deployment considerations in order to ensure adequate timing and data transport needs are meet. AeroShield requires at least a quad core processor and ample RAM to simultaneously acquire and process large amounts of data from three RSMs. A slow PC will cause lower spatial resolution as the tracking points get further apart in time. At some point, a slow PC will not be able to keep up with the data needs of the localization algorithms.
Of equal importance is the network connection to the RSMs. AeroShield acquires RF data from the three probes, aligns them in time, and uses the speed-of-light to calculate the distance difference to each RSM. This allows a triangulation of the source location. The network connection must be high-speed and low latency in order to acquire data that actually overlaps in time from each probe.
Built into AeroShield is a Latency test that determines if the network connections are adequate for AeroShield to operate properly. This is available in the AeroShield UI, as well as through both the .NET APIU and the TCP/IP API.
There is a small benchmark program (BenchMark.exe) provided in the application installation directory. Running this program takes about 10 to 15 seconds. BenchMark.exe exercises the PC in ways similar to AeroShield, and produces and score when complete. A computer that scores below 300 is not adequate and will fail with AeroShield. A score above 600 is desirable.