Ambient IoT Backscatter Devices as Passive Anchors for NLOS Cellular Positioning: Fundamental Limits

Technical Summary: Ambient IoT Backscatter Devices as Passive Anchors for NLOS Cellular Positioning

Problem Statement
Cellular positioning in non-line-of-sight (NLOS) environments is fundamentally challenged by the absence of a direct line-of-sight (LOS) path between the base station (BS) and user equipment (UE). While Reconfigurable Intelligent Surfaces (RIS) have been proposed to create controllable "virtual LOS" paths, they typically require calibrated, coherent control and a common phase reference. In contrast, Ambient Internet-of-Things (IoT) backscatter devices (BDs) are low-cost, passively powered, and independently controlled. They lack a common phase reference and have uncalibrated reflection gains and residual phases.

The core problem addressed is determining the fundamental localization limits when these uncalibrated, known-location BDs are used as passive anchors in an uplink cellular scenario. Specifically, the paper investigates how much positioning information survives when the direct NLOS path and the BD-assisted paths share a common, unknown scatterer, and when the BDs introduce nuisance parameters (unknown gains, relative responses, and phases) that cannot be pre-calibrated.

Methodology
The authors formulate a shared-scatterer NLOS localization model where the BS observes a direct NLOS component and multiple BD-assisted components. The model assumes a dominant scatterer-to-BS segment is common to all paths, while the UE-to-scatterer and BD-to-scatterer segments vary based on known BD locations.

1. Signal Model: The system extracts measurements from the uplink Sounding Reference Signal (SRS). The direct NLOS component and BD-assisted components are separated via known slow-time modulation signatures (Doppler bins). The measurements are modeled as complex gains multiplied by path coefficients, plus noise.
2. Nuisance Parameter Handling: The model treats the common complex gain (absorbing the scatterer-to-BS path), the relative BD response coefficient, and the residual per-BD phase offsets as nuisance parameters.
3. Fisher Information Analysis: The authors derive closed-form Equivalent Fisher Information Matrices (EFIMs) for the range-domain parameters under five distinct calibration regimes:
   • Fully calibrated (all parameters known).
   • Unknown BD phases.
   • Unknown common gain.
   • Unknown relative BD coefficient.
   • Fully uncalibrated (all nuisance parameters unknown).
4. Mapping to Position: The range-domain EFIMs are mapped to the UE and scatterer position domain using a Jacobian transformation. The Cramér–Rao Bound (CRB) is used to establish lower bounds on the estimation error (Position Error Bound, PEB).

Key Contributions

• Closed-Form EFIMs: The paper provides analytical expressions for the EFIMs under various calibration scenarios, quantifying exactly how the loss of specific information (phase, gain, or relative response) degrades the Fisher information.
• Information-Loss Laws:
  • Phase Uncertainty: Unknown BD phases completely eliminate carrier-phase ranging information from the BD-assisted paths, leaving only bandwidth-dependent delay and amplitude-gradient information.
  • Gain/Response Uncertainty: Joint uncertainty in the common gain and relative response couples the BD-assisted paths and removes narrowband amplitude-gradient information from the direct NLOS path, leaving it with only bandwidth-dependent delay information.
  • Relative Response Uncertainty: If only the relative response is unknown (but gain and phases are known), the direct NLOS path retains its information through a calibrated relative reference.
• Identifiability Conditions: The study derives necessary conditions for joint single-snapshot identification of the UE and scatterer positions. It establishes that at least two BDs are required in 2D and three in 3D. Crucially, it notes that the number of devices is insufficient without geometric diversity; the bearings of the BDs to the shared scatterer must ensure the geometry Jacobian is full rank.
• RIS Analogy: The framework is shown to apply to separable RIS or subpanel-based architectures where individual element paths are observable, providing a calibration dictionary for such systems.

Results
Numerical evaluations based on a corridor-to-room NLOS scenario (3.5 GHz carrier, 12.47 MHz RMS bandwidth) reveal:

• Calibration Impact: In a fully calibrated regime, adding BDs rapidly reduces the PEB to centimeter-level accuracy (e.g., 0.98 cm with 4 BDs). In contrast, when BD phases are unknown, the PEB plateaus at a much higher level (54 cm with 4 BDs), demonstrating that carrier-phase information is critical for high precision.
• Geometric Diversity: The number of BDs alone does not guarantee accuracy. Deployments where BDs are collinear with the scatterer (poor angular diversity) result in ill-conditioned geometry, leading to large PEBs even with many devices. For instance, 20 collinear BDs performed worse than 4 well-separated BDs.
• Shared Response: The results show that narrowband information relies heavily on a shared response structure across devices. As the number of independent unknown gain groups increases (i.e., less sharing), the PEB degrades significantly.

Significance and Claims
The paper claims to be the first to quantify the fundamental limits of uncalibrated, known-location passive anchors observed over a shared-scatterer NLOS channel, a regime representative of Ambient IoT.

The significance lies in identifying the specific "calibration losses" that prevent low-cost, uncalibrated backscatter devices from achieving the high accuracy of engineered RIS solutions. The results indicate that for Ambient IoT positioning to be effective in NLOS:

1. Geometric Diversity is Paramount: Passive anchors must observe the common scatterer from sufficiently diverse directions; device count alone is insufficient.
2. Calibration Trade-offs: While full calibration yields the best bounds, the analysis clarifies that even partial calibration (e.g., knowing relative phases or gains) preserves significant information compared to the fully uncalibrated case.
3. Practical Deployment: The findings suggest that practical deployments should prioritize resolvable passive signatures, phase calibration, sufficient sounding bandwidth, and diverse anchor bearings to the dominant scatterer to approach the theoretical bounds.

The paper concludes that while uncalibrated BDs can support positioning, their utility is strictly bounded by the loss of carrier-phase information and the geometric conditioning of the deployment, distinguishing them fundamentally from coherent, engineered RIS anchors.