Title Page
Contents
Abstract 8
초록 9
1. Introduction 10
2. Reduced-Order Track-to-Track Fusion for Multisensory Linear Systems 15
2.1. Problem Statement 15
2.2. Derivation of Reduced-Order Fusion Equations 17
2.3. Recursive Equations for Local Cross-Covariances 22
2.4. Computation Algorithm for Implementation of ROF 26
2.5. Reduced-Order Fusion for Identical Sensors 28
2.5.1. Derivation of Effective Formula for MSE of ROF: Identical Sensors 29
2.5.2. Accuracy of ROF for Different Numbers of Identical Sensors 30
2.5.3. Comparative Analysis of ROF and FOF for Identical Sensors 31
3. Numerical Simulations and Comparative Analysis 33
3.1. Computational Complexity of ROF, FOF and CI 33
3.1.1. Computational Complexity of T2TF Filters for 2D Motion Model with Constant Velocity and Five Sensors 35
3.1.2. Computational Complexity of T2TF Filters for 3D Motion Model with Five Sensors 36
3.2. Performance of ROF 38
3.2.1. Performance of ROF for 2D Motion Model with Identical Sensors 38
3.2.2. Performance of ROF for 2D Motion Model with Different Sensors 41
CONCLUSION 43
REFERENCES 44
Table 3.1. Examples of ratio k=QFOF(n, L)/QROF(l, L)[이미지참조] 34
Table 3.2. Comparison of computational times with FOF, ROF and CI for n=4, l=2 and L=1~5 35
Table 3.3. Comparison of computational times with FOF, ROF and CI for n=9, l=3 and L=1~5 36
Figure 1.1. Centralized fusion structure 10
Figure 1.2. Decentralized fusion structure 11
Figure 3.1. Comparison of MSEs for 2D motion with ROF versus number of sensors (L) 37
Figure 3.2. Comparison of MSEkCKF, MSEkROF and MSEkCI for 2D motion model using ROF, CI and CKF with 3 identical sensors[이미지참조] 40
Figure 3.3. Trajectories of 2D motion using CKF, ROF and CI with 3 identical sensors 40
Figure 3.4. Comparison of MSEkCKF, MSEkROF and MSEkCI for 2D motion model using ROF, CI and CKF with 3 different sensors[이미지참조] 42
Figure 3.5. Trajectories of 2D motion model using ROF, CI and CKF with 3 different sensors: two sensors are identical, and the third sensor is more accurate 42