In an ideal vacuum, radio frequency (RF) signals propagate outward in perfect, concentric spheres. Under these pristine conditions, the Received Signal Strength Indicator (RSSI) decreases predictably as a function of distance, following the elegant physics of the inverse-square law. This deterministic relationship underpins the intuitive appeal of wireless localization: if you can measure how much a signal has attenuated, you should be able to calculate exactly how far away the transmitter is.
However, as Jae-Hun Lee et al. observe, these idealized circular propagation models completely break down in real-world deployments. Instead of acting as a clean proxy for distance, signal strength behaves as a highly volatile, non-linear variable. Understanding why this instability happens—and how researchers work around it—is the central challenge of modern wireless positioning and proximity sensing.
The Physical Mechanics of Volatility
Why does signal strength fluctuate so wildly, even when both the transmitter and receiver are completely stationary? The instability stems from several distinct physical phenomena that distort the radio environment:
- Multipath Fading: In indoor or dense urban environments, radio waves bounce off walls, floors, and obstacles. A receiver does not just catch a single direct line-of-sight wave; it catches a chorus of reflections arriving at slightly different times and phases. As explained by Hejun Wu et al. and Amin Kargar et al., these waves interfere with each other constructively or destructively. This creates severe signal volatility (fading) that can drastically alter RSSI without any physical movement.
- Shadowing and Obstructions: Physical objects absorb and scatter RF energy. The human body, composed mostly of water, is an exceptionally effective attenuator. C. Galperti documents how environmental noise and body shadowing create a phenomenon where a stationary user appears to rapidly enter and exit a zone because the signal dips below thresholds.
- Device State and Orientation: The physical context of the hardware introduces severe bias. Arkadiusz Stopczynski et al. demonstrate that signal strength measurements are highly sensitive to whether a sensor is tucked away in a pocket or held out in a hand. This is reinforced by the authors of Tackling Multipath and Biased Training Data for IMU-Assisted BLE Proximity Detection, who note that these varying "carriage states" mask true physical boundaries. Furthermore, M. Coppola et al. observe that the physical orientation of a device's antenna alone can inject massive noise into distance estimations.
- Dynamic Environmental Drift: Wireless environments are rarely static. Jiannong Cao notes in BaG: Behavior-aware Group Detection in Crowded Urban Spaces using WiFi Probes that people moving through a space constantly alter the signal path. Even in completely static environments, long-range low-power signals like LoRa experience severe baseline instability, as documented by Chaojie Gu, proving that time-varying environmental noise is inescapable.
The Cascading Failures of Naive Ranging
When localization or routing algorithms treat raw RSSI as a direct ruler, physical volatility triggers cascading system failures.
First, geometric positioning errors compound rapidly. As Bing Li et al. argue, converting RSSI directly to distance is an inherently error-prone step that amplifies initial measurement noise into massive coordinate errors. Second, network routing suffers. In mesh networks, routing protocols rely on signal strength to choose the best access point. Gang Yao et al. show that relying on these unstable metrics causes frequent handoff failures and suboptimal routing because a device might mistake a temporary signal spike for a close, reliable node.
Finally, this instability breaks behavioral models. In contact tracing and epidemiology, Chunlin Huang et al. point out that wireless signal range does not map cleanly to the biological mechanisms of disease transmission, meaning naive RSSI thresholds can severely misestimate exposure risks.
Embracing the Noise: Modern Mitigation Strategies
To build reliable systems, researchers have shifted away from treating RSSI as a clean ruler. Instead, they employ several key paradigms to "tame" or bypass signal instability.
Statistical Filtering and Fuzzy Logic
Rather than accepting a single, volatile peak value—which Ying Cheng et al. show leads to high error rates due to signal drift—modern systems smooth the data over time. Bing Li et al. use Gaussian Mixture Models to statistically reconstruct true RSSI values from noisy distributions. Similarly, Gang Yao et al. apply fuzzy logic functions to handle the inherent vagueness of RF data, transforming noisy fluctuations into probabilistic signatures rather than hard distance metrics.
Fingerprinting over Ranging
Instead of calculating distances, fingerprinting treats environmental noise as a unique spatial signature. By mapping the specific, complex pattern of visible access points at a given location, systems bypass the need for a path-loss model entirely. For instance, Piotr Sapieżyński et al. leverage lists of visible WiFi access points to infer proximity with high accuracy, though they caution that using secondary wireless signals like Bluetooth as ground truth caps performance due to its own underlying noise.
Multi-Sensor Fusion
Fusing RF signals with complementary, non-RF sensors allows systems to cross-reference data and self-correct. Pierre-Yves Lajoie et al. demonstrate in PEOPLEx: PEdestrian Opportunistic Positioning LEveraging IMU, UWB, BLE and WiFi that combining smartphone Inertial Measurement Units (IMUs) with multiple wireless sources via factor graphs allows a system to dynamically correct for signal noise and carriage states without requiring prior environmental maps.
Bypassing RSSI for Richer Metrics
Finally, some researchers abandon raw signal power altogether in favor of metrics that capture the underlying physics of wave propagation. Yunbing Hu et al. utilize Channel State Information (CSI), which records the fine-grained phase and amplitude variations across multiple subcarriers to reconstruct the physical environment with sub-meter accuracy. For long-range networks where signals frequently drop below the noise floor, Chaojie Gu highlights phase-based localization as a far more stable alternative to power-based metrics.