⚡ electrical engineering

DPI Suppression for Non-cooperative Bistatic Radar Based on Direction-Preserving Orthogonality of Signal Eigenvectors

This paper proposes a DPI suppression method for non-cooperative bistatic radar that leverages the direction-preserving orthogonality of signal eigenvectors to eliminate interference components after decoherence, thereby enhancing target DOA estimation accuracy.

Original authors: Yiduo Guo, Jiawei Lu, Yu Zhang

Published 2026-09-14

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Yiduo Guo, Jiawei Lu, Yu Zhang

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the world of radar, there is a fundamental challenge that has long limited how well we can see the world around us. Traditional radar systems work by sending out their own powerful radio waves and listening for the faint echoes that bounce back from objects like airplanes or ships. While effective, these systems are expensive to build and easy for an enemy to locate and destroy. To solve this, engineers developed a smarter approach called passive bistatic radar. Instead of carrying their own transmitter, these systems act like silent listeners, using radio waves already being broadcast by third parties—such as television stations, mobile phone towers, or satellite navigation systems—to illuminate targets. The receiver simply waits for those signals to bounce off a target and return. This method is cheaper, harder to detect, and far more survivable in a conflict. However, this clever trick comes with a massive obstacle: the signal the receiver is trying to hear is often drowned out by the original broadcast itself. The direct path of the radio wave from the transmitter to the receiver is millions of times stronger than the tiny whisper of a reflection from a distant aircraft. This overwhelming noise, known as direct-path interference, acts like a spotlight that blinds the sensor, making it nearly impossible to spot the target unless the interference is somehow removed.

Researchers at the Air Force Engineering University in China have developed a new mathematical technique to solve this blinding problem without needing to know exactly where the interference is coming from. In their study, they focused on the hidden structure of the radio signals arriving at the receiver's antenna array. When a radio signal hits a group of antennas, the data collected can be broken down into fundamental building blocks called eigenvectors. Think of these eigenvectors as the distinct "directions" or patterns that the signal energy takes as it moves through the antenna system. The researchers discovered a specific rule governing these patterns: when a powerful interference signal and a weak target echo come from very different directions, their corresponding eigenvectors naturally separate from each other. They become mathematically orthogonal, meaning they point in completely different directions within the data space, effectively ignoring one another. This separation holds true even when the interference is so strong that it completely overwhelms the target signal, provided the two sources are not located in the same narrow beam of the antenna.

The team tested this idea by simulating three different real-world situations. In the first scenario, they modeled a single strong interference source and a single target. In the second, they added multiple interference sources, such as reflections from buildings or other terrain, alongside one target. In the third, they reversed the setup to include one interference source and multiple targets. In every case, they found that as long as the interference and the target were not arriving from the same angle, the mathematical patterns of the interference remained distinct from the patterns of the target. The interference eigenvectors pointed firmly in the direction of the noise, while the target eigenvectors pointed toward the object of interest, with almost no overlap between the two. This discovery allowed the researchers to create a new algorithm that acts like a filter. Instead of trying to guess where the interference is or using a separate antenna to listen to the direct signal, the algorithm simply identifies the mathematical patterns belonging to the strong interference and removes them from the data. Once these dominant patterns are stripped away, the remaining data reveals the much weaker target echoes clearly.

To verify their method, the researchers ran thousands of computer simulations using a receiving antenna with twenty elements, similar to what might be used in a real passive radar system. They set up scenarios where the interference was sixty decibels stronger than the target signal, a level of intensity that would normally make detection impossible. When the target and interference were separated by a wide enough angle, the new algorithm successfully suppressed the interference and pinpointed the target's direction with high accuracy. The results showed that the system could detect targets even when the signal-to-noise ratio was very low, outperforming several existing methods that rely on knowing the interference's location in advance. However, the study also highlighted a clear limitation: when the target and the interference arrive from the same general direction, the mathematical separation breaks down, and the algorithm cannot distinguish between them. This confirms that the technique relies entirely on the physical separation of the sources.

The significance of this work lies in its simplicity and its independence from prior knowledge. Most current methods for cleaning up radar signals require a reference channel to capture the direct interference signal or need to know the exact location of the jammer to cancel it out. The new approach requires neither. By leveraging the natural mathematical separation of the signal patterns, it can suppress interference using only the data collected by the main receiving array. The researchers demonstrated that this method is robust and effective, offering a practical path forward for passive radar systems to operate in environments filled with strong, distracting signals. While the simulations were conducted under controlled conditions, the results suggest that this eigenvector elimination technique could significantly improve the reliability of passive radar, allowing these silent sensors to see through the noise and track targets that were previously hidden.

Technical Summary: DPI Suppression for Non-cooperative Bistatic Radar Based on Direction-Preserving Orthogonality of Signal Eigenvectors

Problem Statement
Non-cooperative bistatic radar (Passive Bistatic Radar, PBR) utilizes third-party illuminators of opportunity (e.g., DVB-T, FM, 5G) to detect targets, offering advantages in cost and survivability. However, a critical challenge hindering reliable target detection and tracking is the intense Direct-Path Interference (DPI). DPI signals, originating directly from the illuminator, possess significantly higher energy than target echoes, often obscuring them. Existing suppression methods, such as spatial adaptive beamforming or time-domain adaptive filtering (e.g., ECA, LS), often rely on assumptions that are difficult to meet in practice: specifically, they may require prior knowledge of interference angles, the establishment of a dedicated Reference Channel (RC) to capture the direct signal, or assume that the surveillance channel contains no direct-path contribution. Furthermore, when DPI and target signals are fully coherent (a common scenario in PBR), standard subspace methods fail to resolve them.

Methodology
The authors propose a novel DPI suppression algorithm based on the "Direction-Preserving Orthogonality" of signal eigenvectors. The methodology proceeds through the following logical steps:

Theoretical Foundation: The paper analyzes the relationship between the eigenvectors of the signal covariance matrix and the steering vectors of the sources. It posits that when the power difference between DPI and target echoes is large, and the sources are not simultaneously located within the radar's main lobe, the eigenvectors corresponding to the DPI and the target exhibit "direction-preserving orthogonality." This means the target eigenvector projects maximally onto the target steering vector while remaining orthogonal to the DPI steering vectors.
Signal Decorrelation: To address the worst-case scenario where DPI and target signals are fully coherent (rank-deficient covariance matrix), the authors employ Toeplitz matrix reconstruction. This process decorrelates the coherent sources, effectively treating them as independent sources with modified power levels, thereby restoring the rank of the signal covariance matrix.
Eigenvector Elimination:
- The algorithm performs eigenvalue decomposition on the reconstructed covariance matrix.
- It identifies the dominant eigenvalues corresponding to the high-power DPIs. The number of DPIs ( $J$ ) is determined by analyzing the gap between adjacent eigenvalues.
- The eigenvectors associated with these dominant eigenvalues (the interference subspace) are eliminated from the signal subspace.
- The remaining eigenvectors, corresponding to the target echoes, are used to construct a new signal subspace.
DOA Estimation: A noise subspace is derived from the new signal subspace. The Direction of Arrival (DOA) of the targets is then estimated using subspace-based algorithms, specifically the Multiple Signal Classification (MUSIC) algorithm, utilizing the constructed noise subspace.

Key Contributions
The paper outlines two primary contributions:

Theoretical Proof of Direction-Preserving Orthogonality: The authors theoretically deduce and prove that under conditions of significant power disparity and angular separation (outside the main lobe), the eigenvectors of DPI and target signals maintain direction-preserving orthogonality. This is demonstrated across three representative scenarios: (a) one DPI and one target, (b) multiple DPIs and one target, and (c) one DPI and multiple targets.
A Novel Suppression Algorithm: Based on this property, the authors propose an eigenvector elimination method. A significant advantage of this approach is that it does not require prior knowledge of the DPI's orientation nor does it necessitate a Reference Channel (RC). This simplifies implementation and reduces computational burden compared to methods requiring RCs or precise angle estimation.

Results
Numerical simulations were conducted using a 20-element Uniform Linear Array (ULA) with broadcast television signals as the illuminator.

Orthogonality Verification: Simulations confirmed that when the target and DPI are not within the same beam, the projection of the target eigenvector onto the DPI steering vector is near zero, while the projection onto its own steering vector is maximal. Conversely, when the target and DPI fall within the same beam, this orthogonality condition is not satisfied, leading to estimation deviations.
Suppression Performance: In scenarios where the target and DPI are in separate beam ranges, the proposed algorithm successfully suppressed the strong DPI (e.g., 60 dB INR) and accurately estimated the target DOA. In contrast, when the target and DPI were in the same beam, the algorithm showed large deviations, consistent with the theoretical limitations.
Comparative Performance: The proposed method was compared against the CLEAN algorithm, the Jamming Jam Method (JJM), and an algorithm from reference [39]. The results indicated that the proposed algorithm achieves a lower SNR resolution threshold and smaller Root Mean Square Error (RMSE) than the CLEAN and JJM algorithms, performing comparably to the algorithm in [39]. It demonstrated high success probabilities even at low SNRs (-20 dB to 10 dB).

Significance and Claims
The paper claims that the proposed method offers a practical solution for DPI suppression in non-cooperative bistatic radar by leveraging the inherent mathematical properties of signal eigenvectors rather than relying on external reference signals or precise interference angle knowledge. The authors assert that the algorithm is straightforward to implement and effective in scenarios where the interference and target are angularly separated. The study concludes that while the method is highly effective under the condition of direction-preserving orthogonality, future research should address scenarios involving multiple interferences and multiple targets where these conditions may be more complex to satisfy.

More like this