DOA Classification and CCPM-PC based GNSS Spoofing Detection and Mitigation Technique
Guanghui Xu, Harbin Engineering University, China & Villanova University, USA; Feng Shen, Harbin Institute of Technology, China; Moeness G. Amin, Villanova University, USA; Chun Wang, Villanova University, USA & Xi’an University of Architecture and Technology, China
The ever-increasing reliance on Global Navigation Satellite System (GNSS) for navigation and guidance has gained growing awareness for the need of protecting the receiver against potential interference including jamming and spoofing. Jamming signal, because of it’s conspicuously higher power and distinctions from the GNSS signal structure, is easy to be detected and mitigated. Spoofing attack is considered more sinister than typical jammers jamming owing to its imitation of the GNSS signals or re-transmitting their replicas. In so doing, the spoofer can commandeer the tracking loops of a victim receiver stealthily. It is not trustworthy on user’s time, position and velocity information once the tracking loop of the GNSS receiver is taken over by the spoofer, which may pose a serious threat to the user’s security. So a lot of spoofing countermeasures are developed in recent years. However, most anti-spoofing techniques focus on the detection of spoofing signal, and they can not provide an efficient way to protect the receiver from the infringing of the spoofer. So in this paper, a spoofing signal’s detection and mitigation technique based on DOA estimation?Cross Correlation Peak Monitoring and Power Calculation (CCPM-PC) and comparison is proposed. Analysis and Design of spoofing detection and mitigation for GNSS receiver under severe multipath environment constitute the main novelty of this paper. Co-array based DOA estimation and elevation angle classification is used to mitigate as much undesired signals as possible. Even if the spoofing signal comes from the space overlapped with the authentic signals, the proposed method can also detect it by power calculation and cross correlation peak monitoring, then mitigation can follow based on subspace processing.
The proposed method can be divided into two successive stages to detect and mitigate the spoofing signal. In the first stage, DOA estimation and elevation angle classification is applied. Since Co-array can provide a larger virtual aperture than the physical array, DOA estimation of much higher number of sources than the number of actual antenna elements will be more efficient for GNSS receiver.
Since the estimated DOA is relative to the antenna array orientation which is attached to the ground-based or airborne platform, we need to transform the DOA estimates into East-North-Up (ENU) coordinate system in order to classify and effectively mitigate the undesirable signals.
After DOA estimation, elevation angle classification is followed according to user’s state. For instance, the reference elevation angle is set as 0 degree for an aircraft. Any signal whose elevation angle below this angle will be treated as undesired signals. Whereas for a terrestrial based carrier, the reference elevation angle will be set as 10 degree or 20 degree (city canyon) according to the specific multipath environment. If some signals elevation angle are detected below the reference elevation angle, they will be mitigated by projecting the received signal into these signals orthogonal subspace. So elevation angle classification can mitigate most kind of multipath and spoofing signals. Therefore DOA estimation is important for spoofing detection and mitigation, but if the spoofing signal comes from the zenith direction like most authentic signals, DOA estimation itself can not work any more. This problem will be resolved at the second stage of the proposed method.
When the spoofer trying to take over the user receiver, it usually represents itself as a weaker multipath signal. This process lasts until the spoofing signal and the authentic signal’s code phase are aligned. The spoofer begins to increase its transmitting power above the authentic signal, so as to move the tracking loop’s correlation peak off its original tracking point which facilitates the receiver takeover mission. Then one can see that during this process there must be a power increase and/or correlation function distortion. And these are two important features one can use to detect the spoofing signal. There are also many other characteristics one can use, for instance, doppler difference between the spoofing signal and authentic signals, spatial angle difference and so on. Current reliable spoofing detection method trends to combine as much of these features as possible. But the detection of spoofing signal is not the final purpose for most users, reliable detection of the spoofing signal followed by effective mitigation algorithm will be more attractive in application, which, however, is paid less attention by the open literature. In the second stage of the proposed method, the feature that a point source spoofer with different PRN codes overlapped in spatial will be used to detect and mitigate the spoofing signal. This kind of spoofing signal usually presents a higher power than authentic signals, which will be utilized in our proposed method. The highest power signal is considered to be a spoofing signal at the beginning, then followed by the verification process that monitoring the cross correlation peaks between the highest power signal and other three high power signals. Since a spoofer usually seeks to mimic at least four satellite signals to take over the receiver, we select the four strongest signal cross-correlations powers to exclude the weaker power multipath. The four highest power signals are extracted from the received signals by using orthogonal projection method. If the highest power signal is a spoofing signal, there must be a correlation peak between the highest power signal and each of the other three signals, otherwise there should be no correlation peaks or no more than one peak in real condition. So if the calculated highest signal’s power is obviously larger than the other three signals, and the highest power GNSS signal is correlated with all of the other three high power GNSS signals, then the spoofing signal is declare detected. And this can be seen as the criteria of the proposed method. The spoofing signal will be mitigated by projecting the received data into its orthogonal subspace if the spoofing signal is detected present in the received data. At last the estimated steering vectors will be used as the weights for desired signals in order to improve their SNR before putting them into their tracking loops.
Anticipated and/or actual results:
In order to verify the effectiveness of the proposed method, two different signal conditions are considered. The first one is based on the platform of Unmanned Aerial Vehicle (UAV). In this case, four authentic signals and two spoofing signals are considered. Where one spoofing signal comes from a terrestrial based transmitter, and another spoofing signal comes from the transmitter fixed on a bloon, the four authentic signals come from the satellites dispersed around the area of zenith. The UAV is flying at an inclined angle.
Matlab simulation results show that after DOA estimation and coordinate system transformation, the spoofing signal comes from terrestrial based spoofer is detected coming from negative elevation angle, and is then removed directly by orthogonal subspace projection. After power calculation and comparison, there is a correlation peak between the highest power signal (spoofing signal comes from the bloon) and each of the other three signals, then the orthogonal projection operation is done to remove the highest power signal. So the spoofing signal comes from zenith can also be found out and mitigated by using the proposed method.
The second simulation condition is based on a terrestrial based vehicle. Firstly, the case with one spoofing signal, one multipath and four authentic signals is considered. The six signals are all with high elevation angles. In this case, the highest power signal is detected correlated with the other three high power signals, which means the highest power signal is a spoofing signal that is ready to be remove. Secondly, the case with two different multipath signals and four authentic signals is considered. In this case, there is no correlation peak between the highest power signal and the rest other three highest power signals when the multipath signal is weak. And there is only one correlation peak when the multipath signal has higher power than the authentic signals, which is rarely happen in real conditions. This indicate that there is no spoofing signal according to the criteria of the proposed method. So the proposed method can detect the spoofing signal successfully based on the estimated signal power as well as the correlation results. By using orthogonal subspace projection to undesired signals, one can mitigate most multipath and spoofing signal successfully.
The proposed spoofing detection and mitigation technique is verified effective in application. Co-arrays applications to GPS receivers provide an efficient way for DOA estimation. By setting a reference elevation angle according to user’s state and transforming the coordinate system from platform coordinate system to ENU frame, most multipath, jamming signals and spoofing signals can be mitigated directly based on the elevation angle classification results. When the spoofing signal comes from zenith direction like most authentic signals, the joint power calculation and comparison, cross-correlation peak monitoring method provide an effective way to detect the spoofing signal. All undesired signals are mitigated by constructing an orthogonal subspace for these signals in the proposed method. Simulation results show that the proposed method works well for both UAV and terrestrial based vehicles regardless of the multipath environment. In addition, by using the estimated steering vectors as the weights for the desired signal, the SNR of all desired signals are improved significantly before entering into their tracking loops.
Significants of the work:
The proposed method provide an effective way to mitigate low elevation angle multipath as well as most jamming and spoofing signals by DOA estimation and elevation angle classification. To the common problem that one is hard to distinguish the spoofing signal from multipath, the proposed method, however, successfully resolve or avoid this problem by doing power estimation followed by selected signals’ cross correlation. So the proposed method can actually mitigate all low elevation angle multipath and multiple point source spoofing signals as long as they are below the reference elevation angle, any spoofing signal comes from the area belongs to authentic signals can also be detected and mitigated successfully. Therefore, the proposed method can not only detect the spoofing signal reliably but also mitigate them successfully.