Stealthy Malicious Interference Detection in UAV Formations: A Power-Domain Signaling Game Approach

summary

The gist

Reliable intra-swarm communication is fundamental to cooperative sensing, control, and mission execution in unmanned aerial vehicle (UAV) swarms The proposed framework exploits transmit-power-induced…

In short

The paper proposes a Bayesian security game framework for detecting stealthy jamming in UAV swarms. It uses transmit-power-induced Signal-to-Interference-plus-Noise Ratio (SINR) variations as signals to distinguish malicious jamming from benign interference. This allows the defender to update its belief about the interference type and make adaptive defensive decisions, improving link reliability.

Key concepts

Bayesian Security Game
This is a mathematical model used to describe strategic interactions between two parties: an interfering UAV (jammer) and a defending UAV. The jammer's strategy involves choosing its transmit power to deceive the defender, while the defender uses observed SINR data to infer whether the interference is malicious or benign.
SINR Variations as Signals
The core idea is that a stealthy jammer can disguise its presence by adjusting its transmit power. This power change causes measurable shifts in the SINR observed by the defender. The paper exploits these specific variations to tell the defender whether it is dealing with a legitimate spectrum-sharing user or an actual malicious attacker.
Pure-Strategy Perfect Bayesian Equilibrium (PBE)
A PBE is a stable outcome in this game where both the jammer and defender play optimal strategies given their beliefs. The paper analyzes two structures: separating, where types are distinguishable, and pooling, where different types result in the same observed SINR.
Utility Functions
The utility measures for each UAV reflect its goals. The defender's utility balances the benefit of maintaining a high link rate against the cost of power boosting and potential false alarms. The jammer's utility focuses on maximizing its attack benefit while minimizing jamming power costs.

Terminology used across episodes

This episode discusses

The paper

Stealthy Malicious Interference Detection in UAV Formations: A Power-Domain Signaling Game Approach · Read on arXiv

Jianle Ba, Shaolong Guo, Yuntao Wang, Xin Wu, Zhou Su, Zhendong Li

School of Cyber Science and Engineering, Xi’an Jiaotong University · School of Information and Communications Engineering, Xi’an Jiaotong University

Transcript

Introduction to the show: ident: Security Radio. Generated commentary on the latest security and cryptography papers.

Nadia: I'm Nadia.

Elias: And I'm Elias, here with Priya.

Nadia: We're looking at the paper: "Stealthy Malicious Interference Detection in UAV Formations".

Elias: Here's the gist of the paper. Reliable intra-swarm communication is fundamental to cooperative sensing, control, and mission execution in unmanned aerial vehicle (UAV) swarms The proposed framework exploits transmit-power-induced SINR variations as informative signals to distinguish stealthy jamming from benign spectrum-sharing interference, enabling the defender UAV to update its belief about the interference type and make adaptive defensive decisions This approach is significant because it addresses the challenge where a stealthy jammer can disguise signal-to-interference-plus-noise ratio (SINR) degradation as link fluctuation in benign interference Game Formulation and Modeling The stealthy jamming detection problem is formulated as a Bayesian security game where the interfering UAV strategically manipulates its transmit power to mimic benign interference, while the defender performs belief-based inference from SINR observations The game involves players N = E, D, where E is the interfering UAV and D is the defender UAV The private type of the interfering UAV is modeled as θ ∈ Θ = θL or θJ, where θL denotes a legitimate UAV conducting normal communication and θJ denotes a malicious jammer The defender updates its posterior belief from the observed SINR and decides whether to maintain normal operation or trigger defensive power boosting Utility Functions and Payoffs The utility of the defender UAV D includes the target-link rate benefit, power-boosting cost, false-alarm loss, and missed-detection loss The utility of the malicious UAV UJ is expressed as a function of attack benefit from degrading the target intra-swarm link, incurring jamming-power cost, and suffering a defense-triggered penalty The legitimate UAV's utility UL accounts for its regular communication benefit, power-cost, and defensetriggered penalty Equilibrium Analysis The paper characterizes pure-strategy perfect Bayesian equilibrium (PBE) conditions under separating structures with distinguishable type and pooling structures with indistinguishable observations A candidate (P qL, PrJ) constitutes a pure-strategy PBE if there exists a belief system µ∗ and a defender UAV strategy σ∗D such that the on-path beliefs are Bayes consistent, and the two types satisfy the incentive compatibility (IC) constraints The defender's best response is determined by comparing expected utility gains from triggering a response based on the posterior belief µ(s) Numerical Validation Numerical simulations validate the effectiveness of the proposed scheme in enhancing defender’s utility under stealthy jamming compared to conventional schemes The proposed PBE-based scheme consistently outperforms fixed-threshold, passive no-boosting, and always-on power-boosting baselines across various prior probabilities and distances This result indicates that the proposed belief-driven detection and response mechanism remains effective under distancedependent interference attenuation In this paper, we proposed a Bayesian security game framework for stealthy jamming detection in UAV swarms The proposed framework exploits transmit-power-induced SINR variations as informative signals to distinguish stealthy jamming from benign spectrum-sharing interference, enabling the defender UAV to update its belief about the interference type and make adaptive defensive decisions By jointly considering link-rate protection, defensive power cost, false-alarm loss, and missed-detection loss, the framework provides a cost-aware detection-and-response mechanism that improves the reliability of critical intra-swarm communications while avoiding unnecessary power boosting The proposed scheme improves the expected defender UAV’s utility over conventional schemes REFERENCES 2611.07977N, Reliable intra-swarm communication is fundamental to cooperative sensing, control, and mission execution in unmanned aerial vehicle (UAV) swarms 5 2611.07977N, Since a jamming UAV can adjust its transmit power to disguise signal-to-interference-plus-noise ratio (SINR) degradation as link fluctuation in benign interference, stealthy jammer UAV detection remains a challenge 5 2611.07977N, In this paper, we propose a novel belief-driven stealthy jammer UAV detection scheme from the perspective of Bayesian security game 5 2611.07977N, To capture stealthy interference in shared-spectrum UAV swarms, the interfering UAV is modeled as a hidden benign or jamming type, whose transmit power strategy induces the SINR observed by the defender 5 2611.07977N, We characterize pure-strategy perfect Bayesian equilibrium (PBE) conditions under separating structures with distinguishable type and pooling structures with indistinguishable observations 5 2611.07977N, Numerical results show that the proposed scheme improves the defender’s expected utility compared to the conventional schemes 5 2611.07978N, Consider a UAV swarm where a UAV S transmits information to a UAV D over a critical intra-swarm link 3 2611.07980N, The interfering UAV E is modeled as having a private type θ ∈ Θ = θL, where θL denotes a legitimate UAV conducting normal communication and θJ denotes a malicious jammer 3 2611.07984N, If the interfering UAV is legitimate and selects power PqL, the defender UAV observes an SINR given by s qL = P 0S gSD N0 + λPqL gLD 3 2611.07985N, If the interfering UAV is malicious and selects jamming power PrJ, the received SINR

10: at the defender UAV is s rJ = P 0S gSD N0 + PrJ gJD 3 2611.07987N, Since neither the interfering UAV’s type nor its transmit power is directly observable to D, the defender can only rely on the resulting SINR to make its defensive decision 3 2611.07988N, If two different type-power pairs induce the same SINR value, they are observationally indistinguishable to the defender UAV 3 2611.07990N, Let i ∈ 0, 1 denote the defender UAV’s action index, where i = 0 corresponds to a0 and i = 1 corresponds to a1 3 2611.07994N, When the interfering UAV is legitimate and uses power PqL, the resulting target-link rate is R iL(q) = B log2 (1 + P iS gSD N0 + λPqL gLD) 3 2611.07995N, When the interfering UAV is malicious and adopts jamming power PrJ, the corresponding target-link rate is R iJ(r) = B log2 (1 + P iS gSD N0 + PrJ gJD) 3 2611.084N, The utility of defender UAV D includes the target-link rate benefit, power-boosting cost, false-alarm loss, and missed-detection loss 4 2611.085N, If the interfering UAV is legitimate, the defender UAV’s utility is given by UD(ai, θL; q) = ωDR iL(q) − iη∆PS − iCFA 4 2611.087N, If the interfering UAV is malicious, its utility is expressed as UJ(ai, PrJ) = ρ R¯i − R iJ(r) − βJP rJ − iδJ 4 2611.093N, A PBE is an assessment (σ∗E, σ∗D, µ∗), where σ∗E and σ∗D denote the strategies of the interfering UAV and the defender UAV, and µ∗ is the belief system 4 2611.095N, Upon observing an on-path signal s qL, the defender UAV updates its belief about the type of the interfering UAV using Bayes’ rule 4 2611.097N, If s qL!= s rJ, the defender perfectly identifies the type: µ∗(s qL) = 0 and µ∗(s rJ) = 1 4 2611.098N, If s qL = s rJ, the types are indistinguishable, so the posterior equals the prior: µ∗(s) = πJ 4 2611.093N, Let ∆L(q) = UD(a1, θL; q) − UD(a0, θL; q) 4 2611.093N, For a legitimate interfering UAV using power index q, we define ∆L(q) = UD(a1, θL; q) − UD(a0, θL; q) = ωD R iL(q) − R 0L(q) − η∆PS − CFA 4 2611.093N, For a malicious UAV using jamming index r, we define ∆J (r) = UD(a1, θJ; r) − UD(a0, θJ; r) = ωD R iJ(r) − R 0J(r) − η∆PS + CMD 4 2611.093N, Given posterior belief µ(s), the expected utility gain from triggering the response is Γ(s; q, r) =

1 − µ(s): ∆L(q) + µ(s)∆J (r) 4 2611.093N, Lemma 1: For a candidate (PqL, PrJ), the defender UAV’s best response is σ∗D(a1s) = (1, Γ(s; q, r) ≥ 0, 0<Γ(s; q, r)<0 4 2611.093N, Under ∆L(q) < 0 < ∆J (r), this is equivalent to σ∗D(a1s) = 1⇐⇒µ(s) ≥ µ∗(q, r) = −∆L(q)∆J (r) − ∆L(q) 4 2611.093N, For separating on-path signals, the best response reduces to σ∗D(a1s qL) = 1 iff ∆L(q) ≥ 0, and σ∗D(a1s rJ) = 1 iff ∆J (r) ≥ 0 4 2611.093N, For pooling signals s qL = s rJ = sqr, it reduces to σ∗D(a1sqr) = 1 iff πJ ≥ µ∗(q, r) 4 2611.093N, Proposition 1: A candidate (PqL, PrJ) constitutes a purestrategy PBE if and only if there exists a belief system µ∗ and a defender UAV strategy σ∗D such that i) the on-path beliefs are Bayes consistent 4 2611.093N, ii) the defender UAV is sequentially rational at every information set according to (15); and iii) the two types satisfy the incentive compatibility (IC) constraints 4 2611.093N, If E!= ∅, each element of E induces a pure-strategy PBE 4 2611.078N, Fig.

Nadia: 2 illustrates the equilibrium mechanism of the proposed Bayesian security game 5 2611.079N, Fig.

Elias: 3 compares the expected defender utility under different prior probabilities of a jamming type 5 2611.084N, Fig.

Title and authors: Nadia: Now we're getting into the details of how they model this. The paper calls this problem a Bayesian security game because it sets up the attacker’s strategy and the defender’s response as a strategic interaction >

Elias: That's right. The key is that the interfering UAV chooses its transmit power to influence what the defender actually observes on the target link, which is the SINR >

Priya: I want to ask about how they handle this uncertainty. Since neither type nor their exact power level is directly visible to the defender, how does it work for them to distinguish between a legitimate UAV and a jammer based only on that one number?

Nadia: That's the million-dollar question. The paper suggests that the defender updates its belief about whether it’s seeing a legitimate signal or jamming based on those observed SINR variations >

Elias: They use Bayes' rule to update their posterior belief after observing an on-path signal, which means if they see one set of numbers, they adjust what they think the attacker is doing >

The paper's summary: Nadia: So to summarize what this paper is actually trying to achieve, it proposes a scheme where the defender doesn't just react to a single SINR value but uses that observation to constantly update their idea of whether they are dealing with legitimate interference or stealthy jamming >

Elias: It’s about exploiting those power patterns. They show that if the interfering UAV is legitimate and uses some power, say P q L, it creates one specific SINR value on the target link >

Priya: And if it’s a malicious jammer using jamming power P r J, that creates a different SINR value, s r J = P zero S gSD N0 plus jamming channel gain times its power > <ref:2611.07977#pg1>

Nadia: Right. The whole point is that the two resulting SINR values are different enough to let the defender make a decision >

Elias: They model the situation by showing that both legitimate and malicious types have specific power choices, like low or high jamming power, which result in distinct SINR outcomes >

The paper's improvements: Nadia: The authors suggest a few ways this approach is better than just using a fixed threshold for detection. They look at the trade-offs between things like link rate benefit and the cost of boosting power >

Elias: They introduce utility functions for both sides of the game, so we see how much benefit the defender gets from boosting power versus how much they lose by triggering a false alarm or missing a real attack >

Priya: It’s interesting because they frame it as a cost-aware decision. The defender isn't just looking for any change; they are calculating the expected utility gain from responding based on their current belief, which is captured in that Gamma function >

Conclusion: Nadia: So, to wrap up, the main thing this paper does is move beyond simple detection by using a full security game approach to figure out how to make adaptive decisions about whether or not to boost power >

Elias: It sets up pure-strategy perfect Bayesian equilibrium conditions for both the jammer and the defender under different scenarios, separating types and pooling observations >

Priya: I just want to say that while the math is rigorous, we need to be careful about how much trust we put in the belief system, especially when those two different power settings result in similar SINR values >

Nadia: That's a fair point. But overall, by using transmit-power variations as informative signals instead of just looking at one snapshot, this framework gives the defender a cost-aware mechanism that should improve reliability without just blindly boosting power >

Elias: It’s a solid way to handle stealthy interference in these shared spectrum environments, and it shows how modeling the adversarial behavior and belief-based response can lead to more robust systems >

More episodes

← Home