Covert Routing with DSSS Signaling Against Cycle Detectors

This paper investigates covert multi-hop communication in wireless networks where an adversary employs a cyclostationary (cycle) detector to reveal hidden transmissions. The covert route employs direct sequence spread spectrum (DSSS) signaling to ensure either maximum end-to-end covertness maximization or minimum latency minimization-under quality-of-service (QoS) and link budget constraints. Optimal bandwidth, transmit power, and spreading gain for each hop jointly satisfy reliability and either rate or covertness requirements. We show the equivalence between the covertness and the detection SNR gain-based widest-path formulations, and, hence, enabling efficient route computation. Numerical simulations in a realistic 3D environment illustrate that (i) end-to-end latency increases exponentially with the covertness requirement, (ii) the end-to-end latency increase is super-linear with the packet size M, and (iii) cycle and energy detectors impose different latency behavior as a function of the message length and the covertness requirement. The proposed framework provides important insights into resource allocation and routing design for covert networks against advanced detection adversaries.

💡 Research Summary

The paper addresses covert multi‑hop communication in wireless networks where a passive adversary, Willie, employs a cyclostationary (cycle) detector that exploits periodic spectral correlations to uncover hidden transmissions. To conceal the communication, each hop uses Direct Sequence Spread Spectrum (DSSS) signaling. The spreading code length (or spreading gain η) expands the signal bandwidth, providing an η‑fold processing gain at the intended receiver while still leaving a cyclostationary signature that Willie can try to detect.

The authors formulate two complementary routing problems under realistic quality‑of‑service (QoS) and link‑budget constraints: (i) end‑to‑end covertness maximization, which seeks to maximize the weakest detection‑error probability (DEP) among all hops, and (ii) end‑to‑end latency minimization, which seeks the smallest total transmission time while guaranteeing a minimum DEP per hop. For each hop, the transmitter can adapt its bandwidth Ω, transmit power P, and spreading gain η. The signal‑to‑noise ratio at the legitimate receiver (Bob) is SNR_B = P|h_T,R|²/(N₀Ω)·η, while Willie’s SNR is SNR_W = P|h_T,W|²/(N₀Ω). Prior work showed that DEP is approximately inversely proportional to SNR_W (DEP ∝ 1/SNR_W) and also improves with longer observation time (i.e., larger packet size M).

Covertness‑maximization: The objective can be written as max_ψ min_{v∈ψ} DEP_v. Because DEP_v is a monotonic decreasing function of the detection‑SNR gain θ_v = SNR_B / SNR_W, maximizing the minimum DEP is equivalent to maximizing the minimum θ_v along the route. This equivalence (Theorem V.2) allows the problem to be transformed into a classic widest‑path (max‑bottleneck) problem on a graph where each edge weight is θ_v. The authors solve it with a Dijkstra‑based widest‑path algorithm, which runs in polynomial time and avoids the costly Monte‑Carlo DEP estimation. The optimal per‑hop parameters are: use the maximum allowed bandwidth Ω_max, set the spreading gain to η_opt = Ω_max / D_req (where D_req is the required data rate), and choose the transmit power that meets Bob’s SNR requirement while minimizing Willie’s SNR, i.e., P_opt = (SNR_req·N₀Ω_max)/( |h_T,R|²·η_opt ).

Latency‑minimization: Here the objective is min_ψ Σ_v λ_v, where λ_v = M/D_v is the per‑hop transmission time (M is the number of bits). The constraints now include DEP_v ≥ DEP_req, BER_v ≤ BER_req, and the same power/bandwidth limits. By rearranging the SNR constraints, the authors derive a lower bound on λ_v: λ_min_v = M·SNR_req·SNR_W,max / (|h_T,W|²·|h_T,R|²·Ω_max). The optimal solution uses the maximum bandwidth and power (Ω_max, P_max,req) and a spreading gain η_opt = Ω_max·λ_min_v / M. Lemma V.1 formalizes this result.

Performance evaluation: Simulations are performed in a realistic three‑dimensional urban environment with Ray‑tracing‑derived channel gains. Key findings include:

1. End‑to‑end latency grows exponentially with the covertness requirement because higher covertness forces lower transmit powers and/or longer routes, increasing per‑hop delays.
2. Latency scales super‑linearly (and eventually exponentially) with packet size M, especially under stringent covertness constraints.
3. Cycle detectors and conventional energy detectors exhibit different latency trends: at low covertness levels the cycle detector is more limiting, while at high covertness levels the energy detector becomes the dominant constraint.

Implications: The work demonstrates that covert routing against sophisticated cyclostationary detectors can be efficiently tackled by converting the problem to a widest‑path formulation, enabling low‑complexity routing decisions. It also highlights the trade‑off surface among covertness, latency, and resource consumption, providing practical guidelines for designing covert networks that must operate under strict QoS and power budgets. The methodology is readily extensible to other spreading techniques or detector models, making it a valuable contribution to the emerging field of low‑probability‑of‑detection (LPD) networking.