Core Concepts
A novel method to identify non-line-of-sight (NLOS) conditions in ultra-wideband (UWB) positioning systems, distinguishing between the presence and absence of the direct path component.
Abstract
The paper proposes a two-step NLOS identification method for UWB positioning systems. The first step classifies the propagation conditions as either line-of-sight (LOS) or NLOS using a Support Vector Machine (SVM) algorithm based on various signal features. The second step further distinguishes between two NLOS scenarios:
Direct-Path NLOS (DP-NLOS), where the delayed direct path component is available, and
Non-Direct-Path NLOS (NDP-NLOS), where the direct path component is completely blocked.
The key novelty of the method is its ability to recognize the absence of the direct path component, which introduces much higher biases and is harder to mitigate compared to the DP-NLOS scenario.
The signal features used for classification include received signal power, power ratio of the received signal to the first path, signal energy, mean excess delay, root mean square delay spread, mean value, variance, kurtosis, amplitude, and the variance of the signal preceding the first path detection. Experimental results in a furnished apartment demonstrate the method's ability to accurately identify LOS, DP-NLOS, and NDP-NLOS conditions.
Stats
The bias introduced in DP-NLOS conditions is usually less than 2 ns, while in NDP-NLOS conditions the bias can be several nanoseconds, leading to ranging errors of even a few meters.
Quotes
"In case of harsher NLOS propagation conditions, in which the direct path component is not available or is its level is too low to be properly received and detected the bias tends to be much higher."
"In such case the safest option would be to exclude those results from location calculation."