Footnote:
Diese Datenquelle enthält auch Bestandsnachweise, die nicht zu einem Volltext führen.
Description:
Mobile devices, such as sensor nodes, smartphones and smartwatches, are now widely used in many applications. Localization is a highly important topic in wireless networks as well as in many Internet of Things applications. In this thesis, four novel localization schemes of mobile devices are introduced to improve the localization performance in three different areas, like the outdoor, indoor and underwater environments. Firstly, in the outdoor environment, many current localization algorithms are based on the Sequential Monte MCL, the accuracy of which is bounded by the radio range. High computational complexity in the sampling step is another issue of these approaches. Tri-MCL is presented, which significantly improves on the accuracy of the Monte Carlo Localization algorithm. To do this, three different distance measurement algorithms based on range-free approaches are leveraged. Using these, the distances between unknown nodes and anchor nodes are estimated to perform more fine-grained filtering of the particles as well as for weighting the particles in the final estimation step of the algorithm. Simulation results illustrate that the proposed algorithm achieves better accuracy than the MCL and SA-MCL algorithms. Furthermore, it also exhibits high efficiency in the sampling step. Then, in the GPS-denied indoor environment, Twi-Adaboost is proposed, which is a collaborative indoor localization algorithm with the fusion of internal sensors such as the accelerometer, gyroscope and magnetometer from multiple devices. Specifically, the datasets are collected firstly by one person wearing two devices simultaneously: a smartphone and a smartwatch, each collecting multivariate data represented by their internal parameters in a real environment. Then, the datasets from these two devices are evaluated for their strengths and weaknesses in recognizing the indoor position. Based on that, the Twi-AdaBoost algorithm, an interactive ensemble learning method, is proposed to improve the indoor localization accuracy by fusing ...