Anti-Jamming MAC Protocol

• Anti-jamming MAC protocols are communication strategies that employ randomized, multiplicative backoff mechanisms to adaptively counter intentional jamming and interference in wireless networks.
• They utilize slotted time, local state synchronization, and interference models like SINR to achieve constant throughput despite adaptive and reactive adversarial conditions.
• Empirical evaluations of protocols such as AntiJam, Jade, and SADE reveal convergence within 10^4–10^5 slots and steady-state throughput in the 20–40% range, underscoring their practical resilience.

An anti-jamming MAC protocol is a medium access control mechanism designed to maintain robust, efficient communication under active adversarial interference—specifically, when an attacker purposefully corrupts or disrupts the wireless medium. The resilience of anti-jamming MAC protocols is evaluated against highly capable adversaries: adaptive, reactive, and bursty jammers. Three principal lines of work, including the protocols AntiJam (Richa et al., 2010), Jade (Richa et al., 2010), and SADE (Ogierman et al., 2013), establish rigorous performance guarantees under various interference, network, and adversary models.

1. Network and Adversary Models

Anti-jamming MAC protocols are typically analyzed in slotted or synchronized time, with each slot subdivided into a transmission or listening/sensing action. The canonical network models include the single-hop setting, the unit disk graph (UDG) for multi-hop topologies, and the Signal-to-Interference-plus-Noise Ratio (SINR) physical interference model (Ogierman et al., 2013). Each node is assumed to be always backlogged.

The adversary models are parameterized as follows:

• (T, 1–ε)-bounded adversary: In any window of w ≥ T slots/rounds, at most (1–ε)·w can be jammed per node. This ensures each node enjoys at least an ε fraction of non-jammed slots.
• Adaptive and reactive jamming: The adversary has access to the full protocol history (including internal states and random choices) and uses carrier sensing to time its jamming with maximal disruption.
• SINR adversarial setting: The adversary injects additive noise at each node, subject to a per-window budget, so that the SINR at any receiver is degraded but constrained by a (B, T)-energy-limitation.
• Uniform versus k-uniform adversaries: A uniform adversary applies identical jamming/noise across all nodes; k-uniform divides nodes into k groups and jams each group all-or-nothing per slot.

Nodes have no ability to distinguish between natural collisions and adversarial interference unless specifically augmented (e.g., via RSSI-based detection as in Jam-X (Boano et al., 2012)). Nodes are assumed to operate without centralized coordination, often with only loose synchronization.

2. Algorithmic Core: Multiplicative Adaptation

The central mechanism in anti-jamming MACs—across AntiJam, Jade, and SADE—is a distributed, randomized backoff governed by local multiplicative-increase/decrease logic on each node’s transmission probability (p_v) and a window/threshold variable (T_v).

State variables per node (v):

• p_v ∈ (0, ĥp]: Send probability, upper-bounded to prevent instability.
• T_v: Window/threshold for adaptation (slot counter for backoff adjustment).
• c_v: Local counter, reset after T_v slots.

Per-slot adaptation (AntiJam/Jade/SADE):

1. With probability p_v, node v transmits a packet (possibly carrying its local state for synchronization).
2. Otherwise, node v senses the channel:
   • Idle detected: p_v ← min{(1+γ)·p_v, ĥp} (multiplicative increase), T_v ← T_v–1.
   • Successful decode: p_v ← p_v/(1+γ) (multiplicative decrease), T_v ← max{T_v–1, 1}.
   • Busy/collision/jammed: p_v and T_v unchanged (immediate).
3. At c_v > T_v: reset c_v; if no idles in last T_v slots, p_v ← p_v/(1+γ), T_v ← T_v + 2.

The cap on p_v (e.g., ĥp ≤ 1/24) enforces bounded contention. The adaptation constant γ is chosen as Θ(1/(log T+log log n)) in AntiJam and Jade, and Θ(1/(log T+log log n)) for SADE, to optimize convergence and mixing time.

On every successful transmission, AntiJam enforces full state synchronization by embedding (p_v, c_v, T_v) in the packet, ensuring all honest nodes evolve in lockstep—a property critical for fairness and throughput stability (Richa et al., 2010).

3. Theoretical Guarantees: Throughput, Optimality, and Convergence

Anti-jamming MAC protocols are analyzed via the competitive throughput metric: a MAC protocol is c-competitive if, over any sufficiently large interval, it achieves at least a c-fraction of the non-jammed (“free”) slots as successful transmissions.

Key results include:

Protocol	Adversary Type	Topology	Throughput Lower Bound	Convergence Bound
AntiJam (Richa et al., 2010)	(T, 1–ε)-bounded, adaptive, reactive	Single-hop	c = Θ(1)	O((1/ε)·log N·max{T, (1/(ε γ²))·log³N})
Jade (Richa et al., 2010)	(T, 1–ε)-bounded, adaptive	UDG multi-hop, degree ≥ 2/ε	c = Θ(1)	Ω((T log n)/ε + (log n)^4/(γ²ε))
SADE (Ogierman et al., 2013)	((1–ε)θ,T)-bounded, adaptive SINR noise	Arbitrary Euclidean, SINR model	c = 2^{–O((1/ε)^{2/(α–2)})}	Ω((T log N)/ε + (log N)^4/(γ ε)^2)

The bound c=Θ(1) for AntiJam and Jade is independent of the number of nodes and network size, depending only on ε and for SADE, also on the path-loss exponent α. Notably, these protocols are essentially optimal: for B > carrier-sense threshold θ in the adversarial SINR model, no MAC protocol can achieve positive throughput even in dense networks [(Ogierman et al., 2013), Theorem 2].

After mixing, these protocols maintain the global sum of sending probabilities p_t = ∑_v p_v(t) in [Θ(1/poly N), O(1/ε²)], ensuring a constant success probability per non-jammed slot. Empirical simulations validate these guarantees: AntiJam achieves steady-state throughputs in the 20–40% range for n from 50 to 5,000 nodes, while 802.11 collapses to near zero for ε < 0.8 (Richa et al., 2010).

4. Physical and SINR Layer Considerations

Protocols such as SADE extend the anti-jamming paradigm from abstract collision models to physical SINR interference models. Successful reception at a receiver j requires

$$SINR_{i\to j} = \frac{P}{d(i,j)^\alpha} \bigg/ \Big(ADV(j) + \sum_{k \in S \setminus \{i\}} \frac{P}{d(k,j)^\alpha}\Big) \geq \beta,$$

with P as transmit power, d(·,·) the Euclidean distance, α > 2 the path-loss exponent, ADV(j) as adversarial noise, and β the SINR decoding threshold.

The optimality of SADE is analyzed by geometric zone decomposition around a receiver, bounding aggregate transmission probabilities in disks and annuli, and employing concentration bounds (Chernoff, Markov) to argue that, except in an ε-fraction of subframes, receivers see bounded aggregate interference. This, combined with the adversary’s bounded jamming fraction, ensures a constant-fraction throughput (Ogierman et al., 2013).

The multi-hop protocol Jade also applies sector-based decomposition to maintain, with high probability, an O(1) contention sum in any disk neighborhood of adequate density, critical for resisting local starvation even under k-uniform adversarial jamming.

5. Empirical Evaluation and Synchronization

AntiJam’s simulation results demonstrate robust performance across a wide range of realistic jamming strategies, outperforming IEEE 802.11 by orders of magnitude in high-jamming regimes. Convergence to steady-state occurs within O(10^4–10^5) slots for n = 1,000, ε = 0.5, and ĥp = 1/24; once converged, more than 90% of slots satisfy p_t ∈ 1/(2ε), 2/ε. Fairness is ensured as the per-node throughput histogram concentrates tightly.

Jade empirically attains 20–35% throughput of the available non-jammed slots in multi-hop, provided all nodes meet the network density threshold. Rapid convergence (O(10–100) rounds) is consistently observed in simulations (Richa et al., 2010).

6. Practical Extensions and Limitations

These protocols require minimal hardware assumptions: no GPS, no centralized controller, only local counters/state, and only approximate slot synchronization. The MAC performance remains robust to moderate synchronization jitter.

Limitations arise for sparsely connected networks (node degree o(1/ε)), where adversaries can force isolation and starvation, and for multi-hop under adversaries with higher group selectivity (e.g., k ≥ 2-uniform) unless local density is sufficient. These impossibility results are information-theoretic and apply regardless of algorithmic improvements (Richa et al., 2010, Ogierman et al., 2013).

Protocols such as Jam-X (Boano et al., 2012) extend the anti-jamming approach at the handshake/agreement level using deliberate jamming signals for ACKs, but their primary function is agreement rather than ongoing medium access under adversarial jamming.

7. Key Analytical Formulas

• SINR Success Condition (SADE/physical model):

$$SINR_{i\to j} = \frac{P}{d(i,j)^\alpha} \Big/ \left(ADV(j) + \sum_{k \in S \setminus \{i\}} \frac{P}{d(k,j)^\alpha}\right) \geq \beta$$

• Adversarial Noise Constraint:

$$\sum_{t\in I} ADV_t(v) \leq B\,T, \quad \text{for any window } I, |I| = T.$$

• Competitive Throughput:

Let $f_v(F)$ = # non-potentially-busy or non-jammed rounds at $v$ in frame $F$; $s_v(F)$ = # successful receptions at $v$; then

$\sum_{v \in V} s_v(F) \geq c \sum_{v \in V} f_v(F)$

for the protocol-specific constant $c$.

In summary, anti-jamming MAC protocols such as AntiJam, Jade, and SADE provide rigorous, distributed solutions for maintaining throughput in shared wireless media under powerful, adaptive, and reactive jamming adversaries. These protocols leverage randomized multiplicative-increase/decrease feedback, local contention estimation, and probabilistic contention management, yielding constant-fraction throughput with minimal fairness and convergence time penalties, and achieve information-theoretic optimality under frequently hostile networking environments (Richa et al., 2010, Richa et al., 2010, Ogierman et al., 2013).