Enum AttackerModel 

pub enum AttackerModel {
    SharedHardware,
    PostQuantumSentinel,
    AdjacentNetwork,
    RemoteNetwork,
    Research,
    Custom {
        threshold_ns: f64,
    },
}

Attacker model determines the minimum effect threshold (θ) for leak detection.

Choose based on your threat model - this is the most important configuration choice. There is no single correct threshold; your choice is a statement about who you’re defending against.

Sources

• Crosby et al. (2009): “Opportunities and Limits of Remote Timing Attacks.” Reports ~100ns LAN accuracy, 15–100μs internet accuracy.
• Van Goethem et al. (2020): “Timeless Timing Attacks” (USENIX Security). Achieved 100ns accuracy over the internet using HTTP/2 request multiplexing.
• Flush+Reload, Prime+Probe literature: Documents cycle-level timing attacks on shared hardware (SGX, cross-VM, containers).

Variants

SharedHardware

Attacker shares physical hardware with the target.

θ = 2 cycles (~0.6ns @ 3GHz)

Use for: SGX enclaves, cross-VM on shared cache, co-located containers, hyperthreading neighbors, shared hosting.

Sources: Flush+Reload, Prime+Probe literature

PostQuantumSentinel

Catch KyberSlash-class timing leaks in post-quantum cryptography.

θ = 10 cycles (~3.3ns @ 3GHz)

Use for: ML-KEM (Kyber), ML-DSA (Dilithium), and other lattice-based cryptography where ~20 cycle leaks have been shown exploitable.

This provides a 2x safety margin: set θ at 10 cycles to reliably catch 20+ cycle leaks.

Sources: KyberSlash (Jancar et al. 2024)

AdjacentNetwork

Attacker on same local network, or using HTTP/2 concurrent requests.

θ = 100ns

Use for: Internal services, microservices, or any HTTP/2 endpoint.

Note: “Timeless Timing Attacks” (USENIX 2020) achieved 100ns over the internet using HTTP/2 request multiplexing. The LAN/WAN distinction is weaker than previously thought.

Sources: Crosby et al. 2009 (LAN), Van Goethem et al. 2020 (HTTP/2)

RemoteNetwork

Attacker over the internet using traditional timing techniques.

θ = 50μs

Use for: Public APIs without HTTP/2, legacy services, high-jitter paths.

Sources: Crosby et al. 2009 (15-100μs range)

Research

Detect any measurable timing difference.

θ → 0 (clamped to timer resolution)

Warning: Will flag tiny, unexploitable differences. Not for CI. Use for: Profiling, debugging, academic analysis, finding any leak.

Custom

Custom threshold in nanoseconds.

Fields

threshold_ns: f64

Threshold in nanoseconds.

Implementations

impl AttackerModel

pub fn to_threshold_ns(&self) -> f64

Convert attacker model to threshold in nanoseconds.

pub fn description(&self) -> &'static str

Get a human-readable description of this attacker model.

Trait Implementations

impl Clone for AttackerModel

fn clone(&self) -> AttackerModel

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for AttackerModel

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter.

impl Default for AttackerModel

fn default() -> AttackerModel

Returns the “default value” for a type.

impl PartialEq for AttackerModel

fn eq(&self, other: &AttackerModel) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Copy for AttackerModel

impl StructuralPartialEq for AttackerModel