HoloTrace: a Location Privacy-Preserving Framework for mmWave MIMO-OFDM Systems
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Electrical Engineering. The day in electrical engineering: signals, systems and the hardware that carries them.
Rosa: We're looking at the paper: "HoloTrace: a Location Privacy-Preserving Framework for mmWave MIMO-OFDM Systems".
Dev: Here's the gist of the paper. High-resolution localization in beyond-5G cellular systems creates new opportunities for location-based services, but also enables base stations (BSs) to infer user location from routine pilot transmissions, raising significant privacy concerns<ref:2509.23444#pg2> HoloTrace introduces a signal-level location-privacy framework for single-BS localization based on mmWave MIMO-OFDM communication<ref:2509.23444#pg2> Framework Introduction HoloTrace is a signal-level location-privacy framework for single-BS localization based on mmWave multipleinput multiple-output (MIMO)-orthogonal frequency division multiplexing (OFDM) communication<ref:2509.23444#pg2> The framework modifies userside pilot symbols while keeping the analog precoder fixed to bias localization-relevant channel features, including angle of arrival (AoA), angle of departure (AoD), and time difference of arrival (TDoA), toward a prescribed fictitious geometry<ref:2509.23444#pg2> This work proposes a signallevel framework where the UE modifies its transmitted pilots to bias the localization-relevant channel features estimated by the BS<ref:2509.23444#pg2> The proposed solution is referred to as HoloTrace, short for holographic trace, as it operates by slightly reshaping the geometric imprint (or trace) left by the transmitted waveform at the physical layer, thereby creating a fictitious (holographic) image<ref:2509.23444#pg2> Spoofing Regimes The paper formulates pilot-level spoofing as a subspace projection-based signal design problem<ref:2509.23444#pg2> The framework breaks down into two regimes: a coherent regime where the UE knows both CSI and geometric parameters (called oracle solution) and a non-coherent regime where the UE knows only the geometric parameters (called blind solution)<ref:2509.23444#pg2> Oracle Spoofing Capabilities Assuming known CSI at the transmitter, HoloTrace can induce the BS to estimate arbitrary feasible positions within its coverage<ref:2509.23444#pg2> The oracle design shows that waveform-level pilot modification can provide location privacy in the evaluated single-BS setting when CSI is available<ref:2509.23444#pg2> Specifically, the unified spoofing approach with CSI allows for closed-form HoloTrace pilot designs that exactly reproduce the desired noiseless received signal corresponding to prescribed AoA, AoD, ToA, and their combinations<ref:2509.23444#pg2> Blind Spoofing Limitations The blind spoofing methods require no channel-gain knowledge while relying on geometric information<ref:2509.23444#pg2> For angular spoofing, perfect blind spoofing is generally possible only when L = 1 and when each element of b(θ0) is nonzero<ref:2509.23444#pg2> For TDoA spoofing, perfect spoofing requires either L = 1 or a specific delay shift τ = τ + ∆τ1L<ref:2509.23444#pg2> In the multiple path case, blind perfect spoofing is generally impossible due to the pairing problem<ref:2509.23444#pg2> Performance Evaluation Simulation results show that oracle HoloTrace can steer the inferred position toward a desired spoofed location in a two-path geometry<ref:2509.23444#pg2> The communication impact is geometry-dependent, meaning an appropriate spoofed location choice can preserve channel-rate quality while achieving effective location privacy<ref:2509.23444#pg2> Oracle HoloTrace closely realizes the target spoofed position in the evaluated two-path scenario, whereas blind variants are limited by the angle-delay pairing ambiguity<ref:2509.23444#pg2> The communication penalty is geometry-dependent rather than intrinsic<ref:2509.23444#pg2> The proposed framework is validated by theoretical analysis and numerical evaluation, characterizing achievable spoofing performance under oracle and blind operations<ref:2509.23444#pg2> Future work should address robust blind spoofing under richer multipath dynamics, digital and hybrid arrays, multi-anchor settings, and adversaries whose positions and orientations are unknown to the UE<ref:2509.23444#pg2> The communication penalty is not fixed but geometry-dependent: locations compatible with the selected beams preserve rates close to the no-spoofing case<ref:2509.23444#pg2> The oracle HoloTrace remains effective over the tested location- and CSI error ranges, with accuracy degrading as uncertainty grows<ref:2509.
Title and authors: Rosa: So we're looking at this paper today, it’s called HoloTrace: a Location Privacy-Preserving Framework for mmWave MIMO-OFDM Systems. It sounds pretty technical, but basically, it tackles the problem that high-resolution location tracking in 5G networks lets base stations see where you are just by looking at your routine pilot signals <ref:2509.23444#pg1>.
Dev: Right, the core idea here is that this system creates new chances for location based services, but it also lets those base stations figure out your location from just normal pilot transmissions, which is a big privacy concern.
Rosa: Exactly. And HoloTrace steps in by offering a signal level framework for single base station localization using mmWave MIMO OFDM communication to address that privacy issue. It’s trying to stop the base station from inferring your location without asking permission first.
Taro: I'm interested in how it works—is this something that actually happens outside of a controlled lab setting, or does it rely on perfect lab conditions?
Dev: That's a good question, Taro. The paper breaks down the spoofing aspect into two regimes: one where the user equipment knows both the channel information and its geometric parameters, which they call an oracle solution. And another one where they only know those geometric parameters, which is called a blind solution.
Rosa: So it’s not just one method; it has these two ways to do the spoofing based on what information you have available at the user equipment when trying to mislead the base station about your position.
Taro: The oracle setting sounds powerful, but I wonder how realistic that is when you're actually out there in a messy world where things aren't perfect.
The paper's summary: Dev: Well, assuming the user equipment has knowledge of its channel information at the transmitter, this oracle setting lets HoloTrace induce the base station to estimate any feasible position within its coverage area. It shows that by modifying those pilot symbols at the physical layer, you can actually provide location privacy in this single base station setup when that channel information is available.
Rosa: What’s interesting is that for this unified spoofing approach with channel information, they derived closed form HoloTrace pilot designs. This means you get an exact way to reproduce the desired noiseless received signal corresponding to specific angle of arrival, angle of departure, time difference of arrival, and even combinations of those features.
Taro: That sounds like a very precise manipulation at the physical layer—they’re not just making noise; they are shaping the actual signal signature left by your transmission.
Dev: Exactly. They are slightly reshaping that geometric imprint or trace left by the transmitted waveform, which creates what they call a fictitious holographic image, and this happens at the physical layer, not just in some higher protocol stack aspect.
Rosa: So this is about biasing those key localization features—angle of arrival, angle of departure, and time difference of arrival—toward a specific fake geometry that the base station will then use to guess your location. It’s about making the trace look like it came from somewhere else entirely.
Taro: If you can trick the system into thinking you are at location A when you are actually at location B, that’s a serious problem for security and privacy in these high-resolution systems.
The paper's improvements: Rosa: The paper points out a couple of ways they can improve or generalize this framework. One major thing is the split between the oracle and blind spoofing regimes, which lets them understand what’s possible depending on whether the user equipment has channel knowledge.
Dev: And for those who are only in the blind setting, they address limitations that exist when you don't know your channel gain knowledge at all. For instance, for angular spoofing to work perfectly in a blind way, they found that it generally only works if the parameter L is equal to one and each element of b of theta zero is nonzero.
Taro: That sounds like a lot of specific conditions to meet just for the blind case to even work well. Are these conditions always going to be met in real life?
Rosa: Not necessarily, but they show the limitations clearly. And for time difference of arrival spoofing, perfect spoofing requires either L is one or a specific delay shift tau equals tau plus delta tau l. In the case with multiple paths, they found that blind perfect spoofing is generally impossible because of this pairing problem.
Dev: So when you look at the performance evaluation, simulation results show that oracle HoloTrace can steer the inferred position toward a desired location in a two-path geometry. But for blind variants, they are limited by this angle delay pairing ambiguity.
Taro: So the communication penalty isn't just an intrinsic part of the system; it depends on which spoofed location you choose, and how that interacts with the channel conditions.
Conclusion: Rosa: To wrap up, the HoloTrace paper validates this entire framework through both theoretical analysis and numerical evaluation. They’ve characterized exactly what kind of spoofing performance you can expect under both oracle and blind operations.
Dev: The main implication for us is that this isn't a fixed solution; the communication penalty isn't set in stone; it depends on the geometry you are trying to achieve, meaning locations that match the selected beams can keep your data rate quality close to what it would be without any spoofing.
Taro: From my side, I think it’s important to remember that the oracle HoloTrace remains effective even when there are some location and channel errors, but accuracy starts dropping as those uncertainties grow. It's not perfect under stress.
Rosa: So in a nutshell, this paper introduces HoloTrace: a Location Privacy-Preserving Framework for mmWave MIMO-OFDM Systems, which is about using signal level manipulation to stop base stations from inferring user location from routine pilot transmissions. It’s a solid piece of work for understanding the practical limits of location spoofing in these advanced communication systems.
Dev: Good stuff. We've got a lot more to unpack next time as we look at how this feeds into larger AI models for wireless autonomy and privacy.
Lorenzo Italiano, Alireza Pourafzal, Hui Chen, Mattia Brambilla, Gonzalo Seco-Granados Fellow IEEE, Monica Nicoli Senior Member IEEE, Henk Wymeersch Fellow IEEE
Politecnico di Milano Department of Electronics, Information and Bioengineering · Chalmers University of Technology Department of Electrical Engineering · Autonomous University of Barcelona Department of Telecommunication Engineering
eess.SP
Submitted: 2025-09-27
Updated: 2026-11-02
Comments: submitted to IEEE Transactions on Vehicular Technology (IEEE TVT)
License: http://creativecommons.org/licenses/by/4.0/
Importance score: 89/100
The gist: High-resolution localization in beyond-5G cellular systems creates new opportunities for location-based services…
Key concepts
- HoloTrace
- This is a signal-level framework that modifies user transmitted pilot symbols. It reshapes the physical signal imprint into a fictitious (holographic) image, effectively biasing localization features estimated by the base station toward a chosen target geometry.
- Oracle Spoofing
- This regime assumes the user knows both the channel state information (CSI) and geometric parameters. Using this knowledge, HoloTrace can design closed-form pilot modifications that exactly reproduce a desired noiseless received signal corresponding to specific angles (AoA, AoD, ToA).
- Blind Spoofing
- This method allows the user to spoof location without knowing the exact channel gains. It relies solely on geometric information and requires no knowledge of path-gain details. However, perfect blind spoofing is often limited by ambiguities in angular or time-delay pairings.
- AoA/AoD/TDoA
- These are key channel features used by the base station to determine a user's location. HoloTrace manipulates the transmitted signals to bias these specific measurements toward a predetermined, false geometric configuration.
Terminology
Summary
High-resolution localization in beyond-5G cellular systems creates new opportunities for location-based services, but also enables base stations (BSs) to infer user location from routine pilot transmissions, raising significant privacy concerns<ref:2509. p. 0 1
HoloTrace: a Location Privacy-Preserving
Framework for mmWave MIMO-OFDM Systems
Lorenzo Italiano, Graduate Student Member, IEEE, Alireza Pourafzal, Member, IEEE, Hui Chen, Member, IEEE,
Mattia Brambilla, Senior Member, IEEE, Gonzalo Seco-Granados, Fellow, IEEE, Monica Nicoli, Senior
Member, IEEE, and Henk Wymeersch, Fellow, IEEE
Abstract—High-resolution localization in beyond-5G cellular
systems creates new opportunities for location-based services,
but also enables base stations (BSs) to infer user location from
routine pilot transmissions, raising significant privacy concerns. This paper introduces HoloTrace, a signal-level location-privacy
framework for single-BS localization based on mmWave multipleinput multiple-output (MIMO)-orthogonal frequency division
multiplexing (OFDM) communication. 1
HoloTrace: a Location Privacy-Preserving
Framework for mmWave MIMO-OFDM Systems
Lorenzo Italiano, Graduate Student Member, IEEE, Alireza Pourafzal, Member, IEEE, Hui Chen, Member, IEEE,
Mattia Brambilla, Senior Member, IEEE, Gonzalo Seco-Granados, Fellow, IEEE, Monica Nicoli, Senior
Member, IEEE, and Henk Wymeersch, Fellow, IEEE
Abstract—High-resolution localization in beyond-5G cellular
systems creates new opportunities for location-based services,
but also enables base stations (BSs) to infer user location from
routine pilot transmissions, raising significant privacy concerns. This paper introduces HoloTrace, a signal-level location-privacy
framework for single-BS localization based on mmWave multipleinput multiple-output (MIMO)-orthogonal frequency division
multiplexing (OFDM) communication.Where the paper says it
Where the paper says it
Framework Introduction
HoloTrace is a signal-level location-privacy framework for single-BS localization based on mmWave multipleinput multiple-output (MIMO)-orthogonal frequency division multiplexing (OFDM) communication<ref:2509.23444#pg2> The framework modifies userside pilot symbols while keeping the analog precoder fixed to bias localization-relevant channel features, including angle of arrival (AoA), angle of departure (AoD), and time difference of arrival (TDoA), toward a prescribed fictitious geometry<ref:2509. p. 0 This paper introduces HoloTrace, a signal-level location-privacy
framework for single-BS localization based on mmWave multipleinput multiple-output (MIMO)-orthogonal frequency division
multiplexing (OFDM) communication. HoloTrace modifies userside pilot symbols while keeping the analog precoder fixed to bias
localization-relevant channel features, including angle of arrival
(AoA), angle of departure (AoD), and time difference of arrival
(TDoA), toward a prescribed fictitious geometry. A critical case arises when the
base station (BS) infers the user equipment (UE) location
opportunistically from routine uplink pilot transmissions,
originally intended for communication and channel estimation,
without an explicit positioning request. Within this line of
physical-layer privacy research, this paper proposes a signallevel framework where the UE modifies its transmitted pilots
to bias the localization-relevant channel features estimated
by the BS: in this context, obfuscation degrades or prevents
localization, whereas spoofing induces a fictitious location
estimate. Our key insight is that, despite these constraints,
a single perturbation block, embedded in the UE’s pilot
generator, can bias the geometric inversion procedure
employed by the BS for positioning, while leaving the
nominal precoder unchanged. We refer to this solution as
HoloTrace, short for holographic trace, as it operates by
slightly reshaping the geometric imprint (or trace) left by
the transmitted waveform at the physical layer, thereby
creating a fictitious (holographic) image.Where the paper says it
Where the paper says it
Where the paper says it
Spoofing Regimes
The paper formulates pilot-level spoofing as a subspace projection-based signal design problem<ref:2509.23444#pg2> The framework breaks down into two regimes: a coherent regime where the UE knows both CSI and geometric parameters (called oracle solution) and a non-coherent regime where the UE knows only the geometric parameters (called blind solution)<ref:2509. p. 4 C(α
′
, ρ|Xe, α, ρ) = 0
for some α′ ̸= 0. Since H(ρ, α) ⊙ Xe depends on the values of the complex
channel gains α and the location-dependent geometric channel
parameters ρ, pilot spoofing breaks down into two regimes: a
coherent regime where the UE knows both α and ρ (called
oracle solution) and a non-coherent regime where the UE
knows ρ but not α (called blind solution), in which case (13)
and (14) should hold ∀α.Where the paper says it
Oracle Spoofing Capabilities
Assuming known CSI at the transmitter, HoloTrace can induce the BS to estimate arbitrary feasible positions within its coverage<ref:2509.23444#pg2> The oracle design shows that waveform-level pilot modification can provide location privacy in the evaluated single-BS setting when CSI is available<ref:2509. p. 1 We refer to this solution as
HoloTrace, short for holographic trace, as it operates by
slightly reshaping the geometric imprint (or trace) left by
the transmitted waveform at the physical layer, thereby
creating a fictitious (holographic) image. Our results show
that waveform-level pilot modification can provide location
privacy in the evaluated single-BS setting; the achievable
spoofing accuracy and communication-rate loss depend on
the available CSI and the selected spoofed geometry. , sounding reference
signals (SRSs)), which occupy dedicated time-frequency
resources distinct from those used for data transmission
(physical uplink shared channel and its associated
demodulation pilots). • Unified spoofing approach with CSI: Assuming
channel-gain knowledge at the UE, we derive closed-form
HoloTrace pilot designs that exactly reproduce the desired
noiseless received signal corresponding to prescribed
AoA, AoD, ToA, and their combinations.Where the paper says it
Where the paper says it
Blind Spoofing Limitations
The blind spoofing methods require no channel-gain knowledge while relying on geometric information<ref:2509.23444#pg2> For angular spoofing, perfect blind spoofing is generally possible only when L = 1 and when each element of b(θ0) is nonzero<ref:2509. p. 6 (24)
Proposition 4. Blind perfect spoofing of the AoA is generally
possible when L = 1 and when each element of b(θ0
) is nonzero, with xe = λ b(θ0
)⊘b(θ0
), for a power control parameter
λ ∈ C. Proposition 5. Achieving a perfect spoofing on a target ToA
vector τ in absence of knowledge α has no solution, except
when: (i) L = 1, or (ii) τ = τ +∆τ1L
, with ∆τ a delay shift. Perfect spoofing is possible only for L = 1, with
pilot xe = λ c(ϕ0
) ⊘ c(ϕ0). For L ≥ 2, blind perfect spoofing
is generally impossible, and the attacker must instead resort
to approximate spoofing, as in the AoA case.Where the paper says it
Where the paper says it
Where the paper says it
Performance Evaluation
Simulation results show that oracle HoloTrace can steer the inferred position toward a desired spoofed location in a two-path geometry<ref:2509.23444#pg2> The communication impact is geometry-dependent, meaning an appropriate spoofed location choice can preserve channel-rate quality while achieving effective location privacy<ref:2509. p. 0 Notably, the communication impact is not fixed
by the proposed method, but depends on the selected spoofed
geometry. An appropriate spoofed location choice can preserve
channel-rate quality while achieving effective location privacy. We further
develop blind designs that operate without complex path-gain
information, while relying on geometric side information, and
explicitly characterize their limitations in multipath scenarios. Simulation results with a two-path single-anchor geometry show
that oracle HoloTrace can steer the inferred position toward a
desired spoofed location, whereas blind variants mainly provide
obfuscation or approximate spoofing due to angle-delay pairing
ambiguities. A spoofed location along the
true LoS bearing but at a different range therefore moves
the inferred position while preserving the communication
beams, and hence the rate. The communication impact is thus
geometry-dependent rather than intrinsic.Where the paper says it
Where the paper says it
Where the paper says it
Conclusion and Future Work
The proposed framework is validated by theoretical analysis and numerical evaluation, characterizing achievable spoofing performance under oracle and blind operations<ref:2509.23444#pg2> Future work should address robust blind spoofing under richer multipath dynamics, digital and hybrid arrays, multi-anchor settings, and adversaries whose positions and orientations are unknown to the UE<ref:2509.23444#pg2> The communication penalty is not fixed but geometry-dependent: locations compatible with the selected beams preserve rates close to the no-spoofing case<ref:2509. p. 9 All
results are compared to the ideal channel rate with perfect CSI
(dashed black line). The results show that the communication
penalty is geometry-dependent: spoofing position #2 lies in
a direction compatible with the selected beams and therefore
achieves a rate close to the no-spoofing case, whereas position
#1 induces a larger rate loss. Oracle spoofing, therefore, remains effective as
long as this residual is small compared with the projected
receiver noise. In practical terms, the required channelgain accuracy becomes more stringent as the effective SNR
increases.Where the paper says it
Where the paper says it
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Signal-level location privacy framework for mmWave MIMO-OFDM systems can be implemented in next-generation cellular networks to prevent base stations from inferring user location from routine pilot transmissions without explicit consent.
-
The AI system can actively modify its transmitted pilot symbols to bias the localization relevant channel features, such as angle of arrival, angle of departure, and time difference of arrival, toward a prescribed fictitious geometry.
-
When equipped with channel gain knowledge at the user equipment (UE), the system can utilize closed-form designs to exactly reproduce a desired noiseless received signal corresponding to prescribed AoA, AoD, ToA, and their combinations for perfect location spoofing.
-
The system can employ blind spoofing methods that operate without complex path-gain information by relying on geometric side information to induce obfuscation or approximate spoofing.
-
The AI system can perform communication optimization by selecting combiner and precoder beams based on different strategies, such as maximizing total received power, while ensuring the resulting position estimate is displaced away from the true location.
Abstract
High-resolution radio localization is becoming increasingly important for next-generation vehicular and mobile networks, enabling infrastructure-assisted positioning for connected and automated mobility. However, the same mmWave multiple-input multiple-output (MIMO)-orthogonal frequency division multiplexing (OFDM) links used for communication may allow a serving base station (BS) to infer the location of a vehicle-mounted or mobile user equipment (UE) from routine uplink pilot transmissions, even without an explicit positioning request. This paper introduces HoloTrace, a signal-level location-privacy framework for single-BS localization in mmWave MIMO-OFDM systems. HoloTrace modifies user-side pilot symbols while keeping the analog precoder fixed to bias localization-relevant channel features, including angle of arrival (AoA), angle of departure (AoD), and time difference of arrival (TDoA), toward a prescribed fictitious geometry. We formulate pilot-level spoofing as a subspace projection-based signal design problem and derive closed-form oracle designs assuming UE-side channel-gain knowledge. We further develop blind designs that operate without complex path-gain information, while relying on geometric side information, and explicitly characterize their limitations in multipath scenarios. Simulation results with a two-path single-anchor geometry show that oracle HoloTrace can steer the inferred position toward a desired spoofed location, whereas blind variants mainly provide obfuscation or approximate spoofing due to angle-delay pairing ambiguities. The communication impact is geometry-dependent rather than intrinsic to the proposed method: selecting a spoofed location compatible with the serving beams can preserve channel-rate quality while providing effective location privacy.
Related papers
- Open-H-RF: A Large-Scale Ultrasound Raw Channel Dataset Initiative
- Scalable Attention for 5G NR Channel Estimation
- Low-Altitude Wireless Networks: A Comprehensive Survey
- Sensing-Assisted OFDM with Index Modulation: A Waveform for Optical Wireless Integrated Communication and Localisation
- Iteration-Free Candidate Pruning and an Index-Domain VLSI Architecture for Sort-Free MIMO Box Decoding up to 1024-QAM
- A Foundation Model for Large-Scale Wireless Network Planning, Operation and Optimization