Mitigation of Radar Range Deception Jamming Using Random Finite Sets

This paper proposes a radar target tracking framework utilizing random finite sets and multiple hypothesis tracking to effectively mitigate main-beam range deception jamming, specifically range gate pull-off attacks, by modeling jammer-induced biases and false detections to maintain accurate target tracking and enable attack detection without relying on additional signal features.

The Gist

Imagine you are playing a game of "Hide and Seek" with a very smart, high-tech friend. You are the Radar, and your friend is the Target (like a plane or a drone). Your job is to keep your eyes on your friend so you know exactly where they are.

But now, imagine a third player, the Jammer, who wants to trick you. The Jammer doesn't just hide; they use a magic mirror to create a "ghost" version of your friend. This ghost looks exactly like your friend but is standing in a different spot, slowly walking away from you. If you chase the ghost, you lose your real friend. This is called Range Deception Jamming.

This paper presents a new, super-smart way for the Radar to stop being fooled by these ghosts. Here is how they did it, explained simply:

1. The Problem: The "Ghost" in the Room

In the past, radar systems were like a person with a flashlight in a foggy room. If a ghost appeared, the person would just chase the ghost because they couldn't tell the difference between the real person and the fake one.

• The Attack: The Jammer sends back a signal that says, "I am your friend, but I am 100 meters further away than I actually am." Every second, the Jammer moves the ghost a little further away (this is called Range Gate Pull-Off or RGPO).
• The Result: A "naive" radar chases the ghost, loses the real target, and crashes or fails its mission.

2. The Solution: The "Detective" Tracker

The authors created a new system called a Resilient Tracker. Instead of just looking at the signal, this tracker acts like a detective who understands how the trickster works.

They used a mathematical tool called Random Finite Sets (RFS).

• The Analogy: Imagine you are looking at a crowd of people. Some are your friend, some are random strangers (clutter), and some are the "ghosts" created by the trickster.
• The Old Way: The old radar tried to guess which person was the friend, but it got confused by the ghosts.
• The New Way: The new radar assumes that any group of people could contain ghosts. It builds a mental model that says, "If I see a group of people moving in a specific, suspicious pattern (like a ghost slowly walking away), I will treat them as a potential trick."

3. How It Works: Two Superpowers

Power A: The "Adaptive" Detective (The Smart Learner)

This version of the tracker is like a detective who learns on the job.

• The Trick: The Jammer tries to pull the radar's attention away by adding a "bias" (a fake distance).
• The Fix: The tracker doesn't just guess the distance; it calculates the bias in real-time. It asks, "How much is this signal lying?" and then subtracts that lie from the measurement.
• The Result: Even if the Jammer changes how fast they move the ghost, the tracker adjusts its math instantly and keeps the lock on the real target.

Power B: The "Non-Adaptive" Detective (The Prepared Veteran)

This version is for when you already know the Jammer's playbook.

• The Trick: If you know the Jammer usually lies by exactly 50 meters, you can set a rule: "Ignore anything that is exactly 50 meters off."
• The Fix: This is simpler and faster, but it only works if you know the Jammer's specific trick beforehand.

4. Handling Multiple Ghosts

What if the Jammer sends two ghosts at once?

• The Old Way: The radar would get overwhelmed and crash.
• The New Way: The tracker uses a "Multiple Hypothesis" approach. Imagine the tracker has a notepad where it writes down different theories:
  • Theory 1: That person is the real friend.
  • Theory 2: That person is Ghost #1.
  • Theory 3: That person is Ghost #2.
    The tracker runs all these theories at the same time. As new data comes in, it crosses out the wrong theories and keeps the right one. It can even add new "Ghost Detectors" to its notepad if a new ghost appears!

5. The Results: Winning the Game

The authors tested this in a computer simulation with four different scenarios:

1. Straight line: The target flies straight; the Jammer tries to pull it off course.
2. Turning: The target makes sharp, fast turns (like a fighter jet); the Jammer tries to confuse it.
3. One Ghost: Just one fake target.
4. Many Ghosts: Two or three fake targets appearing at the same time.

The Outcome:

• The Naive Tracker: Got fooled immediately and lost the target.
• The New Tracker: Kept the target locked on, even when the target was turning sharply and multiple ghosts were appearing. It was accurate to within a few meters, while the old tracker was off by dozens of meters.
• Bonus: It could also tell the operator, "Hey, I think a Jammer is here!" with high accuracy.

Summary

Think of this paper as teaching a radar system how to spot a lie. Instead of blindly believing every signal it receives, it now understands the "behavior" of a liar. It can calculate the lie, subtract it, and see the truth, even when the liar is moving fast or sending multiple fake signals at once. This makes radar systems much safer and more reliable in the chaotic world of electronic warfare.

Technical Summary

Here is a detailed technical summary of the paper "Mitigation of Radar Range Deception Jamming Using Random Finite Sets" by Calatrava et al.

1. Problem Statement

The paper addresses the challenge of Main-Beam Range Deception Jamming, specifically Range Gate Pull-Off (RGPO) attacks, in radar target tracking systems.

• The Threat: In an RGPO attack, a jammer (often co-located with the true target or acting as a false target generator) intercepts the radar signal, adds a time delay, and retransmits it. This creates a "fake" target measurement that appears at a greater distance than the true target. The jammer gradually increases this delay (pull-off) to lure the radar's tracking gate away from the true target.
• The Challenge: Conventional trackers often fail because they cannot distinguish between the true target and the deceptive measurement, especially when the jammer mimics the target's trajectory. Traditional countermeasures often rely on:
  • Signal feature extraction (e.g., amplitude, angle differences), which may be insufficient if the jammer is sophisticated.
  • Heuristic assumptions (e.g., "farther ranges are fake"), which can fail in complex scenarios or against Range Gate Pull-In (RGPI) attacks.
  • Multiple radar systems, which are not always available.
• Goal: Develop a tracking framework that is resilient to RGPO attacks using only position information (no additional signal features), capable of handling data association (DA) uncertainties, false alarms, and multiple simultaneous attacks.

2. Methodology

The authors propose a Multiple Hypothesis Tracking (MHT) framework based on Random Finite Sets (RFS) to model the measurement set, which naturally handles variable numbers of detections (true target, jammer, and clutter).

A. System Model

• State Space: The target state includes 2D position and velocity ($x_k = [p_k, v_k]^T$).
• Measurement Model: The radar receives an unordered set of measurements $Z_k$. This set is modeled as the union of:
  1. Primary Target: A binary RFS (detected or missed).
  2. Secondary Targets (Jammer + Clutter): Modeled as a Poisson RFS.
     • Clutter: Uniformly distributed false alarms.
     • Jammer: Modeled as a secondary target-generated measurement with a specific spatial bias along the Line of Sight (LOS).
• Attack Vector: The deceptive range $R^d_k$ is modeled as $R^t_k + v_{po}(t_k - t_0)$, where $v_{po}$ is the pull-off velocity.

B. Core Algorithm: MHT with RFS

The tracker uses a Gaussian Mixture filter (a closed-form solution for linear Gaussian RFS models) to maintain multiple hypotheses about the target's state and the origin of measurements.

• Clutter Model Integration: The key innovation is incorporating the spatial behavior of RGPO attacks directly into the clutter intensity function. Instead of treating jamming as random noise, the tracker assumes jammer measurements follow a Gaussian distribution centered around the true target but shifted by a bias vector along the LOS.

C. Proposed Strategies

The paper presents two distinct approaches:

1. Adaptive Strategy (Bias Estimation):

   • State Augmentation: The state vector is augmented to include the jammer-induced bias ($b_k$) as a dynamic state variable: $\tilde{x}_k = [p_k, v_k, b_k]^T$.
   • Mechanism: The tracker dynamically estimates the bias in real-time. If the bias estimate converges (uncertainty drops below a threshold), the tracker identifies the measurement as a jammer and corrects the target position.
   • Multi-Pulse Extension: To handle simultaneous RGPO attacks, the clutter model is extended to a Gaussian Mixture with a time-varying number of components ($C_k$).
     • Dynamic Management: Components are managed via "Vigilant," "Active," and "Dormant" states.
     • Vigilant: Ready to detect a new attack.
     • Active: Confidently tracking an ongoing attack (low uncertainty).
     • Dormant: Inactive due to high uncertainty (saved for later).
     • This allows the system to spawn new mixture components for new attacks and merge/remove them as attacks start or stop.

2. Non-Adaptive Strategy (Prior-Based):

   • Used when accurate priors on the jamming range/bias are available.
   • Does not augment the state vector. Instead, it fixes the bias parameter and inflates the covariance matrix along the LOS direction to account for the uncertainty of the attack range.
   • Effective when the attacker's behavior is predictable but less robust than the adaptive version against unknown pull-off velocities.

D. Jamming Detection

A probabilistic detection metric is derived to determine if the tracker is currently under attack. It calculates the probability that at least one measurement in the current set is jammer-generated (rather than a false alarm) based on the likelihood ratios of the mixture components.

3. Key Contributions

1. Feature-Independent Resilience: A tracking solution that relies solely on motion state and spatial characteristics, eliminating the need for complex signal feature extraction (e.g., amplitude or angle discrimination).
2. Adaptive Bias Estimation: A novel method to dynamically estimate and correct jammer-induced biases, significantly improving tracking accuracy over static models.
3. Dynamic Mixture Management: A mechanism to dynamically manage the number of Gaussian components in the RFS clutter model, enabling the mitigation of simultaneous RGPO attacks with different trajectories.
4. Probabilistic Detection: A derived expression to detect the presence of deception jamming in real-time without relying on heuristics.

4. Experimental Results

The authors evaluated the framework using 1,000 Monte Carlo runs across four scenarios (straight-line and maneuvering targets with single or multiple simultaneous attacks).

• Tracking Accuracy (RMSE):
  • Both adaptive and non-adaptive strategies achieved near-optimal performance, closely matching a "clairvoyant" benchmark (which knows the true target state and ignores DA uncertainty).
  • Compared to a "naive" tracker (which only handles false alarms), the proposed methods improved position accuracy by up to 20 meters.
  • The adaptive tracker outperformed the non-adaptive one after an initial adaptation period, particularly when the jammer's pull-off velocity deviated from priors.
• Jamming Detection:
  • The adaptive strategy demonstrated superior detection capabilities. It maintained high probability of detection even as the bias increased.
  • The non-adaptive strategy failed to detect attacks when the induced bias exceeded the assumed prior bounds (e.g., dropping detection probability to near zero when bias > 100m in Scenario 2).
• Multiple Attacks:
  • The dynamic component management successfully tracked scenarios with up to three simultaneous RGPO attacks, correctly identifying the number of active jamming sources and maintaining track continuity.
• Loss of Efficiency (LoE):
  • In the absence of jamming, the proposed strategies showed negligible degradation compared to the Posterior Cramér-Rao Bound (PCRB), proving they do not sacrifice performance when no attack is present.

5. Significance

This work represents a significant advancement in Electronic Warfare (EW) and radar signal processing:

• Robustness: It provides a mathematically rigorous framework to counter sophisticated DRFM-based deception attacks without requiring hardware changes or multi-static radar setups.
• Scalability: The dynamic management of mixture components allows the system to scale to complex, multi-jammer environments, a common future threat scenario.
• Practicality: By relying on motion dynamics rather than specific signal features, the solution is applicable to a wide range of radar systems and is resilient to jamming techniques that mimic target signal characteristics.
• Real-time Viability: The use of closed-form Gaussian mixture solutions ensures the algorithm remains computationally tractable for real-time implementation, despite the complexity of handling multiple hypotheses.