<p> Source localization using TDOA measurements has been an important research topic mainly because of its wide applications in practical localization systems. The TDOA localization accuracy is known to be very sensitive to the sensor position errors and even a small amount of sensor position errors can cause significant performance loss. This thesis considers the use of calibration emitters to improve the source location estimate from TDOAs when the sensor positions are subject to random errors. The principle behind is that with calibration emitters, extra TDOA measurements that contain information on the true sensor positions can be obtained and if properly exploited, they are able to mitigate the effect of sensor position errors and improve the source localization accuracy. This thesis investigates, in four different localization scenarios (scenarios (i)- (iv)), the effect of calibration TDOA measurements on the source localization accuracy and how to efficiently explore them in the localization algorithm design to achieve the optimum performance. </p><p> We consider in scenario (i) the TDOA localization of a source in the presence of sensor position errors when a calibration emitter at perfectly known position is available. The CRLB of the source location estimate for this case is derived and it indicates that the utilization of calibration TDOAs can lead to great improvement in localization accuracy over the case without a calibration emitter. The performance of the differential calibration (DC) method used in GPS to mitigate the effect of satellite position uncertainty and other errors using a calibration emitter is analyzed theoretically. It is shown that in general, the DC method cannot attain the CRLB accuracy. We then develop a novel algorithm that explores the TDOA measurements from both the calibration emitter and the unknown source to estimate the source location. The newly proposed solution is in closed-form and more importantly, we prove analytically that unlike the DC method, it can reach the CRLB accuracy under two mild conditions.</p><p> Scenario (ii) is practically more realistic than scenario (i) in the sense that besides the known sensor positions, the calibration positions also have random errors and the number of calibration emitters can be greater than one. We establish the CRLB of the source location estimate for this case and quantify the performance degradation due to the calibration position errors. The sensitivity of the source location estimate to the calibration position errors is investigated and the result indicates that the source localization accuracy can be severely degraded if the calibration position errors are simply ignored. Based on this observation, a closed-form TDOA localization algorithm that takes both the calibration position errors and the sensor position errors into account is developed. The proposed solution explores all the calibration TDOAs to improve performance and it is shown analytically to reach the CRLB accuracy under mild conditions.</p><p> In scenario (iii), multiple sources at unknown locations need to be localized from TDOA measurements when the sensor positions are not known accurately. This problem includes the scenario where a single source is located with calibration emitters at completely unknown positions as a special case. The multiple sources are assumed to be disjoint and for each source, a separate set of TDOA measurements is obtained. Recognizing that in literatures, only iterative methods are available for the multiple source localization problem, we propose a closed-form solution that does not require iterations and is computationally more attractive. The new concept of hypothesized source positions is introduced in the algorithm development. Through theoretical performance analysis, we show that the proposed solution is able to attain the CRLB localization accuracy under small noise conditions. The developed algorithm can also provide improved sensor positions as a byproduct. (Abstract shortened by ProQuest.)</p><p>
| Identifer | oai:union.ndltd.org:PROQUEST/oai:pqdtoai.proquest.com:13877178 |
| Date | 16 April 2019 |
| Creators | Yang, Le |
| Publisher | University of Missouri - Columbia |
| Source Sets | ProQuest.com |
| Language | English |
| Detected Language | English |
| Type | thesis |
Page generated in 0.0018 seconds