⚡ electrical engineering

Backscatter Assisted Indoor NLOS Positioning

This paper proposes a robust, noncoherent power-domain tracking method using passive asynchronous backscatter devices as virtual anchors to achieve practical sub-meter indoor non-line-of-sight positioning without requiring phase synchronization or power calibration.

Original authors: Kalle Ruttik, Hüseyin Yiğitler, Jingyi Liao, Alexander Sheverdyaev, Riku Jäntti

Published 2026-06-17

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Kalle Ruttik, Hüseyin Yiğitler, Jingyi Liao, Alexander Sheverdyaev, Riku Jäntti

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are walking down a long, narrow hallway in a building. You can't see the exit, and there are no GPS satellites above you. Usually, figuring out exactly where you are in this situation is like trying to find a needle in a haystack while wearing blindfolded goggles. The walls block the signals, and the echoes bounce around so much that standard location tools get confused.

This paper presents a clever new way to solve this problem using invisible "ghost" helpers and a bit of mathematical magic.

The "Ghost" Helpers (Passive Backscatter Devices)

Think of the hallway as a room filled with tiny, silent mirrors (called Backscatter Devices or BDs) stuck to the walls.

They don't have batteries: These mirrors are completely passive. They don't send out their own signals.
They don't talk: They don't need to be synchronized with each other or the main system. They are just "dumb" objects.
How they work: Imagine a lighthouse (the transmitter) shining a beam of light down the hall. When that light hits a mirror, the mirror wiggles slightly to change the color (frequency) of the reflected light just a tiny bit.
The Result: Even though the mirrors are silent, the receiver (your phone or a tracking device) can hear them because their "wiggled" light is distinct from the main lighthouse beam. It's like hearing a specific bird chirp in a noisy forest; even if you can't see the bird, you know exactly where it is based on that unique sound.

The Problem: The Hallway is a Mess

In a real hallway, the signal bounces off walls, floors, and ceilings. It's like shouting in a canyon; your voice echoes back in a confusing mess.

The "Phase" Problem: Usually, to track something precisely, you need to know the exact timing and phase of the wave. But because these mirrors are "dumb" and the hallway is messy, the timing gets scrambled.
The Solution: Instead of trying to listen to the timing (which is messy), the researchers decided to listen to the loudness (power). They realized that even if the signal is bouncing everywhere, the average loudness still gets quieter the farther you are from a mirror.

The "Smart Map" (The Algorithm)

The researchers built a computer program that acts like a very strict, smart detective. Here is how it works:

The Grid: Imagine the hallway is covered in a grid of invisible tiles (like a chessboard).
The Rules: The detective knows you can't walk through walls. So, if a tile is inside a wall, the detective ignores it.
The Guessing Game:
- The detective looks at how loud the signals are from the four mirrors.
- It guesses, "If I were standing on this tile, how loud would the mirrors sound?"
- It compares that guess to what it actually heard.
The "Huber" Filter: Sometimes, a signal might get weirdly loud or quiet because of a weird echo (like a sudden shout in the canyon). The detective uses a special rule (called Huber estimation) that says, "Okay, that one weird reading is probably a fluke; I'll ignore the extreme parts and focus on the general trend."
The Motion Rule: The detective also knows you are walking, not teleporting. If the last guess was at tile A, the next guess can't be at tile Z (the other end of the hall). It must be a nearby tile. This "motion regularization" smooths out the path.

The Results: How Well Did It Work?

The team tested this in two ways:

Computer Simulations: They built a virtual hallway on a computer.
- The Result: The system was incredibly accurate, usually guessing the position within 23 to 27 centimeters (about the length of a ruler). Even in the worst cases, it was rarely off by more than half a meter.
Real-World Test: They went into a real office hallway at Aalto University in Finland. They set up four of these "dumb mirrors" and walked a receiver down the hall.
- The Result: It worked! The system guessed the position with a median error of 50.5 centimeters (about 20 inches).
- Comparison: They compared it to a simple, old-school method (just averaging the signal strengths). The new "smart detective" method was significantly better, cutting the error almost in half compared to the simple method.

Why This Matters

The paper claims this is a breakthrough because:

No Batteries Needed: The mirrors don't need power.
No Sync Needed: The mirrors don't need to talk to each other or the main system.
Works in the Dark: It works perfectly in "Non-Line-of-Sight" (NLOS) situations where you can't see the transmitter.
Uses Existing Tech: It can work with standard cellular signals (like 5G or LTE) without needing special hardware modifications.

In short, the paper shows that you can turn a hallway full of "dumb," battery-free mirrors into a highly accurate GPS system, even when you can't see the source of the signal, by using a smart algorithm that focuses on signal strength rather than perfect timing.

Technical Summary: Backscatter Assisted Indoor NLOS Positioning

Problem Statement
Modern mobile networks require robust positioning capabilities for applications such as emergency services, industrial automation, and asset tracking. While cellular systems can achieve sub-meter accuracy under favorable line-of-sight (LOS) conditions, performance degrades significantly in non-line-of-sight (NLOS) environments like indoor corridors, urban canyons, and tunnels. Conventional time- and angle-based measurements are often limited by blockage, diffraction, and multipath, particularly when bandwidth and antenna aperture are constrained. Existing solutions for NLOS mitigation, such as multipath-assisted methods or fingerprinting, frequently demand wide bandwidths, large antenna arrays, dense calibration data, or substantial infrastructure coordination, which hinders scalability. Furthermore, prior work on backscatter localization has largely focused on localizing the tags themselves using dedicated readers or phase-coherent processing, leaving the use of passive backscatter devices (BDs) as known-location virtual anchors for tracking a separate user equipment (UE) largely unexplored.

Methodology
The paper proposes a backscatter-assisted positioning framework for indoor corridor environments where passive BDs act as engineered virtual anchors. The system operates under a bi-static topology: an external transmitter illuminates the environment with a continuous wave (CW), and passive BDs modulate and reflect this signal with specific frequency shifts. This allows the receiver to separate BD-dependent backscatter components from the direct carrier in the Doppler domain using standard channel estimation, without requiring active transmission from the anchors or phase synchronization.

Key methodological components include:

Noncoherent Power-Domain Formulation: To address the lack of phase synchronization between BDs and residual carrier offsets, the approach utilizes power-domain measurements rather than phase-coherent signals.
Measurement Model: The BD-dependent measurements are modeled using a log-distance law with unknown BD-specific offsets ( $G_k$ ) and a common path-loss exponent ( $p$ ). This model treats each BD as an effective transmitter with unknown power, absorbing stationary link components (transmit power, antenna gains, fixed propagation loss) into the offset parameter. This eliminates the need for transmit-power calibration.
Tracking Algorithm: The authors develop a corridor-constrained Maximum A Posteriori (MAP) tracker implemented as a Viterbi-style dynamic programming algorithm on a discrete grid.
- State Space: The UE trajectory is discretized into grid cells, with validity constraints ensuring the UE remains within the corridor.
- Cost Function: The objective minimizes a negative log-posterior comprising an emission cost and a motion regularization term.
- Robustness: The emission cost employs Huber loss to handle non-Gaussian noise and multipath fading residuals.
- Parameter Estimation: The algorithm jointly estimates the UE trajectory, the path-loss exponent (via coarse grid search), and the BD-specific offsets (via Huber Iteratively Reweighted Least Squares, IRLS).

Key Contributions

Virtual Anchor Concept: The paper introduces and experimentally validates the use of passive, asynchronous BDs as known-location virtual anchors for continuous UE tracking, a scenario distinct from prior tag-localization work.
Infrastructure-Light Approach: The method is compatible with existing LTE and New Radio (NR) systems by exploiting standard reference signals, requiring no radio hardware modifications at the receiver. It avoids the need for BD phase synchronization or transmit-power calibration.
Robust Tracking Framework: The development of a noncoherent, corridor-constrained MAP tracker that integrates motion regularization and Huber-robust estimation to handle residual carrier offsets and strong multipath.
Experimental Demonstration: The work provides one of the first experimental demonstrations of continuous UE positioning using passive asynchronous BDs in a real-world office corridor.

Results
The proposed method was evaluated through both ray-tracing-inspired simulations and real-world measurements:

Simulation: In an L-shaped corridor model with U-shaped and S-shaped trajectories, the method achieved median positioning errors of 0.23–0.27 m with 90th-percentile errors below 0.45 m. The results indicated that while a more accurate channel model (ray-tracing benchmark) could improve performance, the log-distance surrogate provided robust sub-meter accuracy.
Measurements: In an office corridor campaign at 866 MHz using four passive BDs, the system achieved an aggregated median error of 0.505 m and a 90th-percentile error of 0.933 m.
Baseline Comparison: In the measurement campaign, the proposed method significantly outperformed a simple weighted-average baseline (which computed position as a convex combination of BD locations based on received power), which yielded an aggregated median error of 0.818 m.

Significance and Claims
The authors claim that passive asynchronous BDs can provide practical sub-meter indoor NLOS tracking while remaining compatible with existing channel-estimation pipelines and energy-autonomous BD deployments. The paper emphasizes that the remaining performance gap is primarily attributable to propagation-model fidelity (specifically deviations from the single-slope log-distance surrogate due to local multipath) and limited spatial sampling density, rather than flaws in the tracking architecture itself. Consequently, the proposed framework serves as a viable basis for incorporating richer, environment-aware fingerprints (e.g., from digital twins) in future indoor deployments. The work demonstrates that infrastructure-light, calibration-free positioning is feasible in challenging NLOS corridor environments.