Physical Layer Security Under Active Adversarial Interference

53 concepts54 papers

Information-theoretic security traditionally relies on a comforting assumption: the adversary is a passive observer. In this classic "wiretap channel" paradigm, an attacker (Eve) silently listens to transmissions, attempting to decode confidential messages without altering the wireless environment. Under this assumption, legitimate nodes can guarantee perfect secrecy by exploiting the statistical differences between the legitimate channel and the eavesdropper's channel—for instance, by injecting artificial noise that degrades only Eve's reception.

However, as wireless networks transition toward highly distributed architectures, this passive adversary assumption increasingly fails. A physical entity capable of listening to a signal is often equally capable of transmitting one. By moving from passive eavesdropping to active interference, an adversary can fundamentally disrupt the physical layer, shattering the assumptions that underpin traditional security proofs.

To build resilient wireless systems, modern frameworks must account for a hybrid threat: an adversary who can simultaneously read from and write to the wireless medium.

The Vulnerability of the Passive Assumption

The core limitation of traditional Physical Layer Security (PLS) is its fragility under active manipulation. As noted by Liang et al. and Poor in Generalized Multiple Access Channels with Confidential Messages, standard capacity-equivocation regions assume that eavesdroppers do not adapt their behavior to enhance interception or disrupt other users. When an adversary actively intervenes, the fundamental trade-offs between transmission rate and secrecy dissolve.

Consider a multi-transmitter system where a helper node injects artificial noise to blind a passive eavesdropper, a scenario analyzed by Zaidi et al. in Multiaccess Channel with Partially Cooperating Encoders and Security Constraints. If the adversary shifts from passive listening to active signal jamming or spoofing, the coordination between the legitimate transmitters is compromised. Similarly, Tang et al. demonstrate that while a helping interferer can protect a wiretap channel from a passive listener, these frameworks break down entirely if the attacker actively manipulates the signals or jams the channel to disrupt the communication dynamics.

Active interference introduces a dual-threat environment. The attacker does not merely minimize the information leakage rate; they simultaneously maximize the legitimate system's error rate or denial-of-service probability.

Passive Paradigm:
Transmitter -------> Receiver (Decodes Message)
     |
     +-------------> Passive Eavesdropper (Listens)

Active/Hybrid Paradigm:
Transmitter -------> Receiver (Signal Jammed/Poisoned)
     ^                      ^
     |                      |
     +-- Active Adversary --+ (Simultaneously Eavesdrops & Jams/Spoofs)

The Arbitrarily Varying Wiretap Channel

To mathematically model an active attacker who can alter their jamming strategy from symbol to symbol, information theorists use the Arbitrarily Varying Wiretap Channel (AVWC). In this model, the channel state is governed not by a benign probability distribution (like Rayleigh fading), but by an adversarial jammer.

In Capacity Results for Arbitrarily Varying Wiretap Channels, Bjelaković et al. and Sommerfeld show that defending against a simultaneous jammer and eavesdropper requires a radical departure from standard coding. Because the channel state changes unpredictably, the deterministic secrecy capacity of an AVWC often drops to exactly zero. This phenomenon is known as the Ahlswede dichotomy: a channel's capacity under active jamming is either equal to its capacity with shared randomness, or it is completely eliminated.

Why does this happen? If an active jammer has sufficient power and channel knowledge, it can perfectly mimic the legitimate transmitter's channel characteristics, as shown by Hosseinigoki et al. and Kosut in Capacity of Gaussian Arbitrarily-Varying Fading Channels. By doing so, the jammer creates a state of symmetrization, making it mathematically impossible for the receiver to distinguish between the true message and adversarial noise.

To overcome this, the system must possess an informational advantage. Nötzel et al. and Wiese demonstrate in The Arbitrarily Varying Wiretap Channel - Secret Randomness, Stability and Super-Activation that providing the transmitter and receiver with a small amount of shared randomness can "super-activate" the channel. This shared secret allows the legitimate nodes to coordinate their codebooks or frequency hops in a way the jammer cannot predict, breaking the jammer's ability to mimic the channel and restoring a non-zero secrecy capacity.

Poisoning the Control Plane via Pilot Spoofing

Active adversaries do not always rely on brute-force energy jamming. In modern multi-antenna systems, such as Massive MIMO, attackers exploit the protocol's control plane through pilot spoofing attacks.

Before a massive antenna array can beamform data toward a user, it must estimate the wireless channel. It does this during a training phase, where the user sends a known "pilot signal." If an active eavesdropper transmits the exact same pilot signal simultaneously, the base station receives a combined signal.

Training Phase (Pilot Spoofing):
Legitimate User ----[ Pilot p ]----> Base Station <----[ Pilot p ]---- Active Eavesdropper
                                           |
                                   (Poisoned Channel Estimate)

Data Phase (Inadvertent Beamforming):
Legitimate User <---[ Weak Signal ]-- Base Station --[ Strong Signal ]--> Active Eavesdropper

As highlighted by Saad et al. in Wireless physical layer security: Part 1, this pilot contamination poisons the base station's Channel State Information (CSI). When the base station subsequently calculates its beamforming weights to send data to the legitimate user, it inadvertently optimizes the transmission beam to steer directly toward the eavesdropper.

The consequences of this attack vector are severe:

  1. Inadvertent Leakage: The secret message is amplified in the direction of the attacker.
  2. Defensive Backfire: In Impact of Artificial Noise on Cellular Networks: A Stochastic Geometry Approach, Wang et al. show that if the base station attempts to emit artificial noise to blind potential eavesdroppers based on poisoned CSI, that noise can accidentally be nullified at the attacker and steered directly toward the legitimate receiver, crippling the system's throughput.

This creates a strategic game for the attacker. As analyzed by Xu et al. in Pilot Spoofing Attack on the Downlink of Cell-Free Massive MIMO: From the Perspective of Adversaries, an intelligent adversary faces a optimization choice: spend their power budget to poison the channel estimation phase (maximizing information leakage) or save it to jam the data transmission phase (maximizing data destruction).

Neutralizing Active Attacks

To defend against an adversary who can both read and write to the channel, researchers have developed mechanisms that force the attacker into a position of disadvantage.

Subspace Hiding

If an active jammer is "reactive"—meaning it listens for a transmission and immediately fires a targeted jamming pulse—traditional spatial filtering fails because the jammer synchronizes perfectly with the data.

To counter this, Marti introduces a technique called MASH (MitigAtion via Subspace Hiding) in Universal MIMO Jammer Mitigation via Secret Temporal Subspace Embeddings. By using a pre-shared secret key, the transmitter embeds the legitimate signal into a hidden temporal subspace. Because the jammer does not know the key, it cannot determine which subspace contains the data. This mathematically forces a highly sophisticated, reactive jammer to behave like a blind, brute-force "barrage jammer," spreading its power across all dimensions and allowing standard spatial processing filters to easily remove the interference.

Key-Dependent Covertness

When a system requires covert communication—where the very existence of the transmission must be hidden—an active adversary presents a severe threat by jamming channels based on observed energy fluctuations to force a detection timeout.

In Covert Communication Over Adversarially Jammed Channels, Zhang et al. prove that a shared secret key is an absolute prerequisite for maintaining reliability under these conditions. The key allows the legitimate nodes to dynamically alter their modulation format and transmission timing, preventing the active eavesdropper from accurately distinguishing between natural channel noise and a covert signal, even while the attacker actively injects jamming power.

Myopic Network Coding

In multi-hop packet networks, an active attacker can selectively read packets from a subset of links and overwrite packets on others. Bitar et al. address this in Network Coding with Myopic Adversaries, establishing the fundamental capacity bounds for networks under "myopic" constraints. By treating the attacker's reading and writing capabilities as a unified information-theoretic bottleneck, the authors show that network error-correction codes can be designed to guarantee data integrity and privacy, provided the network's routing topology forces the attacker to operate with incomplete global knowledge.

By abandoning the passive adversary assumption and modeling the attacker as an active participant, these frameworks shift the security paradigm from a fragile game of hide-and-seek to a robust, information-theoretic defense capable of withstanding intentional environmental disruption.

Key Foundational Work

Generalized Multiple Access Channels with Confidential Messages (2006) — contributed 1 concept elements to this direction.
Quantifying eavesdropping vulnerability in sensor networks (2005) — contributed 1 concept elements to this direction.
History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_passive_adversary_assumption_6d423d22,
  title = {passive adversary assumption},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/passive-adversary-assumption/rdr_6d423d2270e6422ea4b32be512855cdf},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}

This exact version (version 1, frozen text):

@misc{lacuna_passive_adversary_assumption_6d423d22_v1,
  title = {passive adversary assumption},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/passive-adversary-assumption/rdr_6d423d2270e6422ea4b32be512855cdf/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-20},
}

See something wrong? Report a mistake