Short Description

Both for wireless communication systems and radio-based localization systems non-line of sight (NLOS) situations between transmitter and receiver nodes can cause major problems. For localization systems, distance estimates between transmitter and receiver can show large biases, causing the localization to fail. However, if the receiver is aware of the NLOS condition, robustness can be substantially enhanced.

To this end, the receiver needs to perform LOS/NLOS classification. This task can be based on the instantaneous channel impulse response (CIR) between transmitter and receiver. We have such measurements available for a representative indoor environment (two CIRs are shown in the plot). Based on those measurements, two or more methods (depending on performing a project or thesis) from the literature should be analyzed, implemented and tested in Matlab. Example methods include decision-theoretic methods like [1] or support vector machines [2]. Especially in the latter case, features that can discriminate between LOS and NLOS need to be found and extracted from the CIR.

Your Tasks

• Literature review concerning NLOS identification methods ([1] and [2] might be suitable starting points)
• Implement and evaluate some promising candidates in Matlab based on real measurements

Profile of prospective student (Requirements)

• Interest in radio channels, lecture Mobile Radio Systems is beneficial
• Interest in classification algorithms and/or statistical signal processing

References

• [1] Guvenc, I., Chong, C., Watanabe, F, and Inamura, H.: “NLOS identification and weighted least-squares localization for UWB systems using multipath channel statistics” in EURASIP Journal on Advances in Signal Processing, 2008
• [2] Marano, S., Gifford, W., Wymeersch, H. and Win, M.: “NLOS identification and mitigation for localization based on UWB experimental data” in IEEE Journal on Sel. Areas in Comm., 2010