Wireless sensing transforms everyday communication infrastructure into a radar-like system. By transmitting a known signal and listening to its reflections, a receiver can map the surrounding environment, calculating a target's distance via Time-of-Flight (ToF) and its speed via Doppler shifts.
In a traditional monostatic setup, the transmitter and receiver share a single local oscillator, making signal alignment straightforward. However, next-generation paradigms—such as Integrated Sensing and Communication (ISAC), cooperative passive radar, and distributed RFID networks—physically decouple the transmitter and the receiver. Without a shared physical clock, independent hardware oscillators inevitably drift, introducing timing and frequency errors that corrupt sensing measurements.
The Anatomy of Clock Drift
No two hardware crystals vibrate at the exact same frequency. When a transmitter and receiver operate on separate clocks, this discrepancy manifests as two distinct challenges:
- Timing Offset (TO): A shift in the perceived arrival time of a signal. Because radio waves travel at the speed of light, even a minuscule nanosecond-level timing discrepancy translates to significant distance measurement inaccuracies. As Sidorenko et al. demonstrate in ultra-wideband (UWB) ranging, these synchronization errors accumulate over the duration of a signal exchange, degrading centimeter-level localization.
- Carrier Frequency Offset (CFO): A mismatch in the baseline mixing frequencies of the transceivers. CFO injects a time-varying phase rotation into the received signal.
Ideal Signal: |--- ToF ---| [Reflected Pulse]
Asynchronous: |--- ToF + Timing Offset ---| [Shifted Pulse] + Phase Rotation (CFO)
This phase distortion severely complicates environmental mapping. Chen et al. note that in uplink ISAC systems, these offsets distort the phase of the Channel State Information (CSI), which simultaneously degrades target range estimation and data recovery.
Furthermore, CFO directly interferes with velocity estimation. A moving target causes a Doppler frequency shift, but an asynchronous clock creates a hardware frequency shift. Because both effects alter the signal's frequency in the exact same manner, a standard receiver cannot distinguish between a target's physical velocity and its own internal clock drift. According to Hu et al., this ambiguity severely limits Doppler sensing accuracy, particularly for low-speed targets. The cumulative effect of these random time and phase offsets is a "blurred" sensing image, a phenomenon characterized by Zhao et al..
Why Naive Calibration Fails
A common workaround is to use a direct Line-of-Sight (LOS) path between the transmitter and receiver as a reference. If the receiver knows the exact position of the transmitter, it can calculate the expected propagation time, isolate the hardware clock offset from the direct signal, and subtract it from the target reflections.
However, Wu et al. highlight a fundamental vulnerability in this approach: real-world environments are dynamic and frequently obstructed. In non-line-of-sight (NLOS) scenarios where the direct path is blocked, the reference signal disappears. Without a stable reference path, traditional calibration frameworks collapse, leaving the system unable to separate environmental reflections from hardware jitter.
Mechanism 1: Mathematical Cancellation via Ratios and Cross-Correlation
To eliminate clock offsets without a direct reference path, researchers use differential signal processing. The core intuition is simple: if multiple signal paths or components are captured by the same asynchronous receiver, they will experience the exact same hardware-induced phase rotation and timing delay. By mathematically comparing these signals, the shared error term can be canceled out.
One prominent implementation of this concept is the CSI-Ratio technique. Instead of analyzing raw channel measurements, the system divides the CSI of one subcarrier or packet by another. Because the oscillator's phase error is a multiplicative term common to both measurements, the division operation strips away the hardware-induced phase rotation. Hu et al. utilize this mechanism to isolate clean Doppler shifts, bypassing clock asynchronism entirely to expose the true velocity of moving targets.
A parallel approach relies on Cross-Correlation. In cooperative passive sensing frameworks, such as those proposed by Wei et al. and Jiang et al., multiple base stations or receivers capture reflections of the same communication signal. By performing cross-correlation on the received waveforms, the independent timing and frequency offsets of the transceivers are mathematically canceled out during the fusion process, leaving only the geometric parameters of the target.
Signal A: Target_Reflection * Clock_Error
Signal B: Reference_Reflection * Clock_Error
Cross-Correlation / Ratio ---> (Target_Reflection) / (Reference_Reflection) [Clock_Error Cancels Out]
Mechanism 2: Spatial Anchors and Geometric Constraints
When mathematical cancellation is insufficient, systems can exploit the physical geometry of the environment. Hu et al. introduce a framework that leverages static "anchor points"—environmental landmarks with known, fixed coordinates.
Instead of treating clock drift as an isolated error to be removed, the system treats it as an unknown variable within a larger geometric equation. Because the distance between the base station and the anchor point is constant and known, any deviation in the measured reflection from the anchor point must be caused by clock drift. By anchoring the mathematical model to these fixed spatial points, the system can jointly solve for the clock offsets and the unknown target positions simultaneously.
Mechanism 3: Statistical Tracking and Subspace Filtering
Because hardware clocks drift dynamically over time due to thermal fluctuations and component aging, offsets cannot be treated as static constants.
To track this time-varying noise, Chen 0029 et al. apply a Kalman Filter-based framework to uplink sensing. By modeling the physics of oscillator drift as a continuous state-space equation, the filter dynamically predicts and suppresses timing and carrier frequency offsets from packet to packet, preventing hardware noise from corrupting range and location estimates.
[Raw Asynchronous Signal]
│
▼
┌──────────────┐
│ Kalman Filter│◄─── Dynamically tracks time-varying drift
└──────┬───────┘
│
▼
[Clean Target Parameters]
For high-resolution imaging, Zhao et al. approach the problem through subspace-based super-resolution sensing. Clock asynchronism typically smears the eigenvalues of the signal covariance matrix, blending target reflections with noise. By separating the signal subspace from the noise subspace corrupted by asynchronism, super-resolution algorithms can mathematically "unblur" the sensing image, restoring sharp target detection even under severe oscillator drift.
Quantifying the Limits of Asynchronous Sensing
To guide the design of these mitigation strategies, researchers have established the fundamental theoretical boundaries of asynchronous sensing. Zhao et al. derived the Fisher Information Matrix (FIM) and Cramér-Rao Bounds (CRBs) tailored specifically to asynchronous ISAC systems.
These bounds provide a mathematical benchmark, defining the absolute limit of sensing accuracy achievable for a given level of clock jitter. By quantifying exactly how timing mismatches degrade parameter estimation, these theoretical frameworks allow engineers to determine whether a distributed sensing system can achieve centimeter-level precision using commodity, un-synchronized hardware.
Key Foundational Work
Toward a Low-Cost Software-Defined UHF RFID System for Distributed Parallel Sensing (2021) — contributed 1 concept elements to this direction.
Error Corrections for Ultrawideband Ranging (2020) — contributed 1 concept elements to this direction.