Triangulateration uses geometric distance to estimate the location of user by employing techniques like received signal strength (RSS), time-of-arrival (TOA), time-difference-of-arrival (TDOA) and image processing.
Radio frequency (RF) signal positioning using TOA or TDOA techniques generally requires timing synchronization of the anchors and/or the anchors and targets. If the desired position accuracy is high and coverage area is large, timing synchronization will be an extremely challenging issue. The first part of this dissertation focuses on improving the performance or deployability of indoor localization systems. Specifically, we propose a synchronization-free positioning network architecture that eliminates the need of timing synchronization. Another problem that remains unsolved in RF based localization is the non-line-of-sight (NLOS) problem, which greatly degrades the positioning performance. We propose a semidefinite programing (SDP) with a soft-minimal method and an NLOS link identification method with bias deduction to mitigate the NLOS error TOA systems. For TDOA systems, NLOS mitigation is more difficult since a reference should be fixed first. To overcome this problem, we propose a method to transform the TDOA architecture into a TOA one, and then form a SDP problem with new constraints.
To avoid the special problems and difficulties in RF signal positioning, such as the synchronization and NLOS problems, in the second part of the dissertation, we propose an image-tag based localization using image processing and convolutional neural network (CNN). In the proposed method, after the segmentation of the tag from the image, information such as the tag ID, the distance, and the angle with reference to the camera are retrieved through CNNs. The camera position is finally reliably and accurately estimated from such retrieved information. The proposed method simplifies the system and provide good accuracy compare to RF based system. In addition, the proposed method effectively resolve those issues that exist in the traditional image-based localization, like the high cost, blind spot problems and unreliable and not scalable for in changing environments.
description.provenance : Submitted by Zhenqiang Su (email@example.com) on 2017-06-15T18:51:14ZNo. of bitstreams: 1SuZhenqiang2017.pdf: 3206000 bytes, checksum: ba820e0f41e09cf4639af142bd454b53 (MD5)
description.provenance : Approved for entry into archive by Julie Kurtz(firstname.lastname@example.org) on 2017-06-20T16:03:43Z (GMT) No. of bitstreams: 1SuZhenqiang2017.pdf: 3206000 bytes, checksum: ba820e0f41e09cf4639af142bd454b53 (MD5)
description.provenance : Made available in DSpace on 2017-06-20T19:48:48Z (GMT). No. of bitstreams: 1SuZhenqiang2017.pdf: 3206000 bytes, checksum: ba820e0f41e09cf4639af142bd454b53 (MD5) Previous issue date: 2017-05-31
description.provenance : Approved for entry into archive by Steven Van Tuyl(email@example.com) on 2017-06-20T19:48:48Z (GMT) No. of bitstreams: 1SuZhenqiang2017.pdf: 3206000 bytes, checksum: ba820e0f41e09cf4639af142bd454b53 (MD5)