Positioning of mobile devices in wireless communication networks is nowadays being intensivelyinvestigated due to the combined benefit of location information and communication. Typical solutionsfor such scenario rely on robust algorithms that estimate position from indirect measurementsof the physical length of the communication links. Since these solutions do not require integrationof additional hardware into the mobile nodes, they are cheap and simple to implement. Asa price to pay, accuracy is typically lower in comparison to dedicated positioning systems. Thus,an important challenge to solve, that this dissertation aims at, is to enhance the accuracy of positionestimation in wireless communication networks without the use of additional hardware. Toreach this goal, the adopted approach is divided in three phases: (i) give a deeper insight on theproblems by analyzing current solutions; (ii) propose solutions for enhancing accuracy of positioninformation; and (iii) exploit the potential of such position strategies in cooperative schemes withinwireless communications.The dissertation starts by analyzing common localization and tracking solutions generally used inpositioning systems. Since the measurement sampling is also important for tracking moving devices,the first analysis concerns the latency time for obtaining location information in short-rangead-hoc networks. Bluetooth is the elected technology since it is an actual example of such kindof networks. Simulations of the communication protocols defined in the specifications tell us thattypical mechanisms present in current wireless communications, such as collision avoidance, candeeply influence the latency time. For accuracy analysis an experimental indoor setup was implemented,as a typical scenario where short-range communications are used in. The physical lengthof the communication links was estimated by using the measurements of signal strength permittedby the Bluetooth specifications. The final results lead us to the conclusion that due to the highnoise measurements and its inherent low frequency of fresh measurements, tracking wireless devicesis a challenging problem. In order to get better results, we propose a solution which restricts themovement models of the mobile devices. This restriction, by relying on the assumption that userstypically move in segment-wise linear trajectories, outperforms the standard solutions.Additionally, this dissertation develops the idea of mobile cooperative positioning. Instead of localizingmobile devices in cellular networks in an individualist manner, the proposed solutions exploitthe physical proximity of other mobile devices. By fostering the cooperation and subsequent communicationamong users, cooperative positioning strategies aim at localizing devices as a groupand not as individuals. In order to reach this goal it is necessary to combine measurements fromtwo domains: device-to-device links and cellular links. Since this combination of information is nota straightforward task, this dissertation proposes two different solutions for solving the problem.The main approach is to use a Bayesian filtering framework design in such a way that both typesof measurements can contribute in a balanced manner to a single estimation of positions for eachcooperative device. While in the first approach all the measurements are combined in a singleestimation procedure, in the second approach, the short- and long-range are separately treated andcombined in a subsequent step. The proposed solutions of combining additional information givenby the device-to-device links with the cellular links, results in considerably higher performance thanstandard non-cooperative solutions.
展开▼
