Localização em ambientes internos utilizando múltiplas tecnologias sem fio

Abstract

Localization is a fundamental problem in many areas of computer science such as robotics, mobile and context-aware computing. Outdoors, the localization problem can be reasonable solved by the use of the global positioning system (GPS). However, the GPS signal can not be used indoors. The large number of applications that would benefit from location information indoors generates the need to develop solutions that work in these environments. Because of this, the problem of indoor localization has been the subject of a great number of researches in recent years. Recently, the use of radio frequency (RF) signal to perform localization in indoor environments has been proposed as an alternative to solve this problem. The main advantages of this kind of localization method are its reduced cost and the large number of wireless devices present in industry commercial buildings and university campuses today. Generally, only one of the several available wireless technologies (Wi-Fi, Bluetooth, Zigbee, etc.) is used and little has been said about the use of multiple wireless technologies to perform the localization. Therefore, in this work we investigate the advantages and challenges of using multiple RF technologies (more specifically, Wi-Fi, Bluetooth and Zigbee compliant devices are used) to perform localization indoors. Two localization methods are used in order to do the research. The first method is based on fingerprinting. We introduce a merge process that can be used to fuse information from multiple wireless technologies when using this method. The merge process consists of concatenating each individual technology map. The K Nearest Neighbors (K-NN) is used to produce the final location estimates. The experimental evaluation of the fingerprinting method shows that the combination of technologies can reduce the localization error. It is investigated the performance of each technology separately and also the performance of the combination of technologies. It is also considered how the number of beacons used affects the localization quality. The results show that for all technologies, more beacons lead to less error. Finally, we show how interference among technologies may lead to lower localization accuracy. The second method is based on a RF propagation model. The model estimates the distance between a beacon and the robots location based on the received signal strength (RSSI). Then, this estimated distance is used by a localization algorithm. We use two algorithms: the first one is trilateration and the second one is the Extended Kalman Filter (EKF). We use simulation to perform a qualitative study in order to evaluate the main factors that affect the quality of localization. The two main factors are the beacons geometric configuration in the environment and the quality of the distance estimates generated by the RF propagation model. We also present results of real experiments that corroborate the results obtained in simulation and show that the method produces better results than those obtained using only the robot\\\\\\\'s odometer. Finally, we show how the use of multiple technologies can increase the availability of the localization service. The two main advantages of using multiple RF technologies to perform localization are the possibility of increase the accuracy and availability of the localization service. The two main challenges are finding the best way to integrate the multiple technologies (in other words, it is necessary to understand what is the best way to fuse information from different RF technologies) and to avoid the adverse effects of this integration that can diminish the localization quality (for example, how to prevent that the interference among technologies affects the results). The further exploration of these advantages and the overcome of these challenges are essential to create more effective localization systems based on multiple RF technologies.