Summary: | 碩士 === 國立成功大學 === 系統及船舶機電工程學系 === 104 === Three microphone-based indoor localization designs developed by combining novel estimation designs and tag sound sources, which are with distinct frequencies, are proposed in this thesis. From the experiments in practice, results summarized a key point for achieving a satisfactory indoor localization: an accurate estimation of the instantaneous sound pressure level (SPL) that is inevitably affected by the random variation of environmental corruptions dominated the indoor localization performance. Based on this concept, three algorithms including Kalman filter based design, H2 estimation design, and H∞ estimation design which accompany with a sound pressure level model are developed for effectively mitigating the effects of received signal strength (RSS) variations due to the reverberation, the reflection, refraction and so on. From the simulation results and practical experiments, it is obvious that these proposed methods deliver really promising localization performance even under heavy environmental corruptions.
The physical characteristics of the proposed methods include: 1.High indoor localization accuracy, 2.Robustness to complicated and unpredictable environmental corruptions, and 3.Low cost designs.
|