Description:
This is a method for classifying Line-of-Sight (LOS) and Non-Line-of-Sight (NLOS) conditions in radio frequency (RF) communication
Background
Indoor positioning systems struggle with accurately identifying LOS and NLOS signal conditions ‑ a critical capability since most localization algorithms assume the presence of a direct signal path. However, indoor environments introduce numerous obstacles, leading to multipath propagation that distorts signal characteristics. Traditional classification approaches are hindered by high computational demands, sensitivity to environmental changes, or require large, up-to-date datasets for training. ML-based techniques using channel impulse response can theoretically achieve high classification accuracy but fail in real-time due to latency and environmental dependence. Moreover, their reliance on learned signal patterns makes them less robust when applied in varying real-world scenarios. This method addresses those limitations by leveraging an inherent statistical feature of phase modulation that persists across environments. Unlike time- or power-domain features, phase distribution remains stable and indicative of LOS/NLOS conditions, particularly when beamforming is used to align signal phases. As a result, this method provides a computationally efficient and environmentally resilient solution to a longstanding IPS challenge.
Technology Description
This method for classifying radio frequency (RF) signals into line-of-sight (LOS) and non-line-of-sight (NLOS) categories uses characteristics of signal phase histograms derived from commonly used modulation schemes like QAM and PSK. This classification is achieved through a histogram-based analysis of the received signal’s phase. Unlike machine learning-based solutions, this approach is highly accurate and computationally efficient—making it suitable for real-time applications in indoor positioning systems (IPS) and Internet-of-Things (IoT) environments. The classification process detects distinctive patterns in the phase histogram to distinguish LOS conditions and NLOS conditions. These histogram patterns are robust to environmental changes and can be extracted from a single raw RF waveform without the need for extensive datasets or complex network models. This method solves a critical challenge in indoor navigation and wireless communication—accurate classification of LOS and NLOS signal paths, which dramatically impacts the precision of IPS. By integrating this method into systems, dynamic adaptation to changing propagation conditions is possible, significantly enhancing system reliability and localization accuracy. The method has been experimentally validated and achieves near-perfect classification accuracy, even in low-SNR environments. The method is both scalable and easily integrated into existing commercial RF systems.
Features & Benefits
- Real-time LOS/NLOS classification, improving system reliability and performance
- Dynamic identification of base stations with LOS channels, improving data throughput and reducing latency
- Does not use machine learning approaches, improving computational efficiency
- Improved accuracy
- Enables adaptive management of resources within wireless networks to optimize signal paths, improving connectivity
Applications
- Internet‑of‑Things (IoT) deployment, especially in position‑sensitive use cases
- Real‑time indoor navigation (IPS)
- Autonomous navigation through built environments
Opportunity
OSU is seeking partners to increase TRL and commercialize.
Status
Patented 63/708,202