SNF-PRP: A Covert Integrating Sensing and Communications Framework * Authors contributed equally to this work.
Dhrumil Bhatt *, Vidushi Kumar * · 2026
Read SNF-PRP: A Covert Integrating Sensing and Communications Framework * Authors contributed equally to this work. by Dhrumil Bhatt *, Vidushi Kumar * (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-03960.txt.
Introduction
Bhatt and Kumar build a covert way to sense while sending data. Joint sensing and communication lets one system do both at once. Its probe signals can be caught, though. A watcher may learn both the message and the fact of sensing. Older safety tools still left sensing open to view.
Their SNF-PRP method hides probes below the noise floor with pseudo-random spread. It gives a covertness bound based on KL divergence. It also uses a spreading gain that past wideband work missed. The team derives the least integration length for a target error bound. In 5G NR n78 tests, range error stayed under half a metre. Speed error stayed under half a metre per second. Those results sat well below the covertness limit, they report.
Catalog record
| CREATOR | Dhrumil Bhatt *, Vidushi Kumar * |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 25 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.03960
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
SNF-PRP: A Covert Integrating Sensing and Communications Framework * Authors contributed equally to this work. Dhrumil Bhatt *, Vidushi Kumar * Abstract Integrated sensing and communication (ISAC) enables simultaneous sensing and data transmission but exposes a critical vulnerability: probing signals may be intercepted, revealing both the transmitted information and the act of sensing itself. Existing physical layer security approaches mitigate interception yet operate with detectable signals, leaving sensing activity observable to a passive warden. This paper introduces sub-noise-floor pseudo-random probing (SNF-PRP), a covert sensing framework for OFDM-based ISAC systems under an energy-detection adversary model. SNF-PRP establishes an ϵ \epsilon -covertness guarantee via Kullback-Leibler (KL) divergence, exploits an N sc N_{\mathrm{sc}} -fold spreading gain absent from prior wideband analyses, and derives in closed form the minimum integration length required to achieve a target Cramér-Rao bound (CRB). Simulations under 5G NR n78 numerology confirm sub-0.5 m range and sub-0.5 m/s velocity accuracy with KL divergence 5.8 × 5.8\times below the covertness threshold, validating joint feasibility at − 12 -12 dB and − 15 -15 dB probing powers. I Introduction Integrated sensing and communication (ISAC) is a cornerstone of next-generation wireless systems, enabling environmental perception and data transmission to share spectrum and hardware [ 12 ] . Yet this integration introduces a security concern that has received little attention: sensing waveforms are deliberately radiated into the environment, exposing both the embedded information and the mere fact that sensing is occurring. Existing physical layer security (PLS) work in ISAC focuses almost exclusively on protecting the communication channel via beamforming, artificial noise (AN) injection, and reconfigurable intelligent surfaces [ 12 , 13 , 3 ] . In all cases the probe operates above the noise floor and the security objective concerns the data payload, not the sensing activity itself. A recent RIS-ISAC-PLS survey [ 14 ] confirms no existing work addresses covertness of the sensing waveform: Zou et al. [ 10 ] come closest conceptually but establish no formal covertness guarantee, and LPI radar techniques [ 11 , 6 ] offer only relative concealment without a KL-divergence analysis or communication overlay. A sufficiently sensitive radiometer can always detect a probe operating above the noise floor, exposing the act of sensing regardless of beamforming geometry or secrecy-rate constraints. The proposed solution answers the resulting open question can sensing itself be made provably undetectable at the physical layer? affirmatively, by proposing sub-noise-floor pseudo-random probing (SNF-PRP) for OFDM-based ISAC in sub-6 GHz bands. A cryptographically seeded PRN sequence is superimposed on a standard downlink waveform below the instantaneous noise floor. A legitimate receiver with the seed performs coherent matched filtering and multi-symbol accumulation to recover sensing parameters; a passive warden without the seed is confined to radiometric energy detection the optimal Neyman Pearson test for two complex Gaussians differing only in variance [ 2 , 5 ] -under which the probe is statistically indistinguishable from thermal noise. SNF-PRP makes three contributions. First , it is, to the best of the authors’ knowledge, the first framework to apply the KL divergence covertness criterion directly to the sensing probe waveform, establishing a provable covertness guarantee for sensing rather than communication. Second , it identifies an N_{\mathrm{sc}} -fold spreading covertness gain (Lemma 1) absent from all prior wideband energy-detection analyses, yielding a correction factor exceeding 10^{6} in the KL divergence. Third , it derives in closed form the three-way tradeoff among probing power \Delta_{p} , integration length N^{*} , and Doppler resolution \delta_{v} , and establishes the SNR invariance of Proposition 1: lower probing power simultaneously deepens covertness and preserves the sensing CRB target via proportionally increased N^{*} , with \mathrm{SNR}_{\mathrm{eff}} at N^{*} unchanged. At \Delta_{p}=-12 dB the required N^{*}=191 subframes corresponds to a 191 ms coherent integration window, feasible within the coherence time of a pedestrian target at 3.5 GHz and consistent with NR measurement gap structures. II System Model and Methodology Consider a 5G NR downlink in which a base station (BS) serves a communication UE while simultaneously probing the environment. A legitimate sensing node shares a pre-distributed cryptographic seed s with the BS. A passive warden (Willie) monitors the channel but has no knowledge of s . The BS must jointly satisfy: Sensing: achieve a target Cramér-Rao bound (CRB) on range and velocity after N OFDM symbols. Covertness: maintain D_{\mathrm{KL}}(\mathcal{H}_{1}\|\mathcal{H}_{0})\leq\epsilon , where \mathcal{H}_{0} and \mathcal{H}_{1} are Willie’s no-probe and probe-present hypotheses. The complete signal flow is formalised in Algorithm 1 . Algorithm 1 SNF-PRP: Sub-Noise-Floor Pseudo-Random Probing 0: Shared seed s , probing power ratio \Delta_{p} , integration length N , noise power \sigma^{2} 0: Range estimate \hat{R} , velocity estimate \hat{v} 1: // Transmitter (BS) 2: Generate PRN: c[k]\!\in\!\{-1,+1\} , k=0,\ldots,N_{\mathrm{sc}}-1 , seeded by s 3: for each symbol m=0 to N-1 do 4: Form overlay: X_{p}[k,m]=\sqrt{P_{p}/N_{\mathrm{sc}}}\,c[k]\,e^{j2\pi km/N_{\mathrm{sc}}} with P_{p}=\Delta_{p}\sigma^{2} 5: Transmit \mathbf{X}[m]=\mathbf{X}_{c}[m]+\mathbf{X}_{p}[m] \triangleright probe buried \Delta_{p}<1 below noise floor 6: end for 7: // Channel 8: Receive: \begin{aligned} Y[k,m]&=\alpha\,e^{-j2\pi k\tau_{s}/N_{\mathrm{sc}}}\\ &\quad\cdot e^{j2\pi f_{d}mT_{\mathrm{OFDM}}}\,X_{p}[k,m]\\ &\quad+W[k,m]\end{aligned} 9: // Legitimate Receiver 10: Regenerate template \tilde{X}_{p}[k,m] from seed s 11: for each symbol m do 12: Matched filter: Z[k,m]=Y[k,m]\cdot\tilde{X}_{p}^{*}[k,m]\,/\,|X_{p}[k,m]|^{2} 13: end for 14: Range profile: P_{R}[\ell]=\sum_{m}\bigl|\mathrm{IDFT}_{k}\{Z[k,m]\}[\ell]\bigr|^{2} 15: RD map: \mathrm{RD}[\ell,q]=\bigl|\mathrm{DFT}_{m}\{P_{R}[\ell,m]\}[q]\bigr|^{2} 16: \hat{R}=\arg\max_{\ell}\,P_{R}[\ell]\cdot\delta_{R} ; \hat{v}=\arg\max_{q}\,\sum_{\ell}\mathrm{RD}[\ell,q]\cdot\delta_{v} 17: // Willie (no seed s ) 18: Computes energy statistic T=M^{-1}\!\sum_{m,k}|Y[k,m]|^{2} ; cannot coherently combine subcarriers \Rightarrow probe appears as AWGN The frequency-domain transmitted vector on the m -th symbol is the superposition \mathbf{X}[m]=\mathbf{X}_{c}[m]+\mathbf{X}_{p}[m], (1) where the PRN overlay on subcarrier k is X_{p}[k,m]=\sqrt{\tfrac{P_{p}}{N_{\mathrm{sc}}}}\,c[k]\,e^{j2\pi km/N_{\mathrm{sc}}}, (2) with P_{p}=\Delta_{p}\sigma^{2} , \Delta_{p}<1 . Each subcarrier therefore carries only \Delta_{p}\sigma^{2}/N_{\mathrm{sc}} of probe power below the individual subcarrier noise floor \sigma^{2}/N_{\mathrm{sc}} by the factor \Delta_{p} . For a single point target at range R and radial velocity v , the received signal is Y[k,m]=\alpha\,e^{-j2\pi k\tau_{s}/N_{\mathrm{sc}}}\,e^{j2\pi f_{d}mT_{\mathrm{OFDM}}}\,X_{p}[k,m]+W[k,m], (3) where \tau_{s}=\lfloor 2RF_{s}/c\rceil (integer-sample delay), f_{d}=2vf_{c}/c (Doppler shift), and W[k,m]\!\sim\!\mathcal{CN}(0,\sigma^{2}/N_{\mathrm{sc}}) . The legitimate node, possessing seed s , regenerates X_{p}[k,m] exactly and applies a conjugate matched filter: \displaystyle Z[k,m] \displaystyle=\frac{Y[k,m]\cdot\tilde{X}_{p}^{*}[k,m]}{|X_{p}[k,m]|^{2}} (4) \displaystyle=\alpha\,e^{-j2\pi k\tau_{s}/N_{\mathrm{sc}}}\,e^{j2\pi f_{d}mT_{\mathrm{OFDM}}}+\tilde{W}[k,m]. An N_{\mathrm{sc}} -point IDFT along the subcarrier axis yields the range profile P_{R}[\ell] ; an N -point DFT along the slow-time axis yields the range-Doppler map \mathrm{RD}[\ell,q] (Steps 9-12 of Algorithm 1). Coherent accumulation over N symbols recovers an effective SNR of \mathrm{SNR}_{\mathrm{eff}}=\Delta_{p}\cdot N, (5) which grows linearly with N , compensating the \Delta_{p}<1 power deficit. Willie, lacking seed s , cannot form \tilde{X}_{p}^{*}[k,m] and is therefore restricted to the incoherent energy test in Step 14, gaining no processing advantage from the subcarrier structure. For uniform power allocation over bandwidth W , the standard OFDM-ISAC CRBs [ 9 , 4 ] under \mathrm{SNR}_{\mathrm{eff}} are \mathrm{CRB}_{R}=\frac{c^{2}}{8\pi^{2}\,\mathrm{SNR}_{\mathrm{eff}}\,B_{\mathrm{rms}}^{2}},\qquad\mathrm{CRB}_{v}=\frac{\lambda^{2}}{8\pi^{2}\,\mathrm{SNR}_{\mathrm{eff}}\,T_{\mathrm{obs}}^{2}}, (6) where B_{\mathrm{rms}}=W/\sqrt{3} and T_{\mathrm{obs}}=NT_{\mathrm{OFDM}} . Inverting ( 6 ) for target accuracies \sigma_{R}^{*} and \sigma_{v}^{*} gives the minimum integration lengths N_{R}^{*}=\frac{c^{2}}{8\pi^{2}\Delta_{p}B_{\mathrm{rms}}^{2}(\sigma_{R}^{*})^{2}},\qquad N_{v}^{*}=\!\left(\frac{\lambda^{2}}{8\pi^{2}\Delta_{p}T_{\mathrm{OFDM}}^{2}(\sigma_{v}^{*})^{2}}\right)^{\!1/3}\!, (7) with binding constraint N^{*}=\max(N_{R}^{*},N_{v}^{*}) . A key structural result follows directly from substituting ( 7 ) into ( 5 ): \mathrm{SNR}_{\mathrm{eff}}\big|_{N^{*}}=\Delta_{p}\cdot N_{R}^{*}=\frac{c^{2}}{8\pi^{2}B_{\mathrm{rms}}^{2}(\sigma_{R}^{*})^{2}}, (8) which is independent of \Delta_{p} . Decreasing \Delta_{p} increases N^{*} proportionally, keeping their product and hence the sensing performance constant. The covertness parameter controls integration latency, not estimation accuracy. Willie collects M=N\cdot N_{\mathrm{sc}} complex samples. Under \mathcal{H}_{0} each sample is \mathcal{CN}(0,\sigma^{2}/N_{\mathrm{sc}}) ; under \mathcal{H}_{1} it is \mathcal{CN}(0,(\sigma^{2}/N_{\mathrm{sc}})(1+\rho_{\mathrm{sc}})) with per-subcarrier SNR \rho_{\mathrm{sc}}=\Delta_{p}/N_{\mathrm{sc}} . For M i.i.d. complex Gaussian samples the KL divergence is [ 2 , 5