Detection of distorted IR-UWB pulses in low SNR NLOS scenarios

In dense environments, for example residential, office or industrial, the predominant type of link within a real time localization system (RTLS) is non-line-of-sight (NLOS). In some NLOS scenarios the signal-to-noise-ratio (SNR) of the first incoming Ultra Wideband (UWB) pulse content is much below 0 dB and the shape of the pulse is distorted. Accordingly the detection of the first pulse is critical. In those cases a possible solution is to detect several sample values which show a high probability to be the center of an UWB pulse. In Time-of-Flight (ToF) based ranging, this leads into a number of pulse or rather distance candidates. The aim of the paper is to present a novel scheme for an improved detection of distorted UWB pulses, especially with SNR below 0 dB. A major objective is to minimize the number of necessary pulse candidates for a defined pulse detection quality. The robustness of the approach has been analyzed by a simulation with a huge number of synthetic pulses in dependency of the SNR and a further series with the degree of distortion as parameter. For a wall of brick and a thickness up to 25 cm an average improvement of 32 % against traditional matched filter (MF) method is achieved.