在無線感測網路(Wireless Sensor Networks, WSNs)以及視覺感測網路(Visual Sensor Network, VSNs)中，目標物追蹤一直是個非常重要的研究議題。近年來許多在WSNs中之目標物追蹤機制已被提出，相較於VSNs，WSNs具有成本低廉、容易佈建的特性。然而，WSNs只能提供目標物之移動軌跡。不同於WSNs，VSNs是由一組視覺感測節點(Camera Node)所構成，每個Camera Node皆具有收集目標物影像之能力，進而提供更豐富的目標物移動資訊。因此本論文分別對於WSNs以及VSNs提出保證追蹤品質之目標物追蹤機制。
    在WSNs中，本論文提出一分散式目標物追蹤(Target tracking)演算法，其根據使用者所要求之目標物追蹤品質(Quality of Tracking, QoT)，盡可能地喚醒較少數量的感測器進行監控與追蹤的任務。此外，為了避免感測器間頻繁地醒睡切換，本論文亦將感測器之醒睡排程機制納入考量，藉以延長感測器的生命週期。實驗結果顯示，本論文所提出之目標物追蹤演算法，確實可保證其追蹤目標物移動軌跡之精準度，並達到延長整個感測網路之生命期的目標。
    此外，除了WSNs，本論文亦結合WSNs與VSNs，提出一分散式演算法，由監控場景中感測器節點感測目標物存在與否，並合作決定出目標物之形狀，然後透過Camera Node進行追蹤照射，不同於近期的相關研究，本論文將不透明目標物的遮障問題納入考量，使得所挑選之Camera Node皆能滿足不透明目標物之完整照射，另一方面，本論文將二元感測節點先天感測的誤差值以及Camera Node所需的預留轉動或換手的時間亦納入考量，當目標物在場景中任意移動，所挑選的Camera Node也能保證連續追蹤，使得目標物在場景中任何時候的影像皆能被完整截取到。實驗結果顯示本論文可達到不透明目標物之完整照射，相較於以往之研究具有更高的目標物邊界覆蓋率，且當目標物移動時具有連續追蹤之效能。

Target tracking is one of the most important issues in wireless sensor networks (WSNs) and visual sensor networks (VSNs). A number of target tracking mechanisms have been proposed for WSNs in recent years. Compared to the VSNs, the WSNs have potential characteristics of low cost, easy deployment and quick response. However, the tracking mechanisms developed for WSNs can only provide the trajectory of the interested mobile target. Different to the WSNs, the VSNs is composed of a set of visual sensor nodes each of which can capture the images of the mobile target, and thus provide rich information of the mobile target. This thesis investigates the target tracking mechanisms for both WSNs and VSNs network environments.
For a given WSN, this thesis proposes a distributed target tracking approach which aims to satisfy the predefined quality of tracking while the number of awaked sensors is minimized. The sleep scheduling of each sensor is also taken into consideration to prevent the sensors from frequently switching between awake and sleep states. Performance study reveals that the proposed target tracking approach can efficiently obtain the accurate trajectory of the mobile target while the network lifetime can be prolonged in comparing with the existing works.
In addition to the WSNs, this thesis also takes into consideration the integrated network environment which is composed of the WSNs and VSNs. A distributed target tracking mechanism is proposed aiming at achieving the full-view of the target while the number of working camera nodes and the camera rotation angles are minimized. Unlike the existing approaches, this thesis takes the unsighted problem of the nontransparent target into consideration. Developing a target tracking mechanism for the integrated WSNs and VSNs has a big challenge since the binary sensor node can only detect whether or not the target is located in its sensing range but cannot identify the location of the target. The deviation of the binary sensor node might cause the target missing problem. Another challenge for the target tracking problem is the rotation delay of the camera nodes which might also cause target missing problem when the handover procedure of the camera nodes are performed. This thesis takes into account the target missing problem and proposes a target tracking mechanism which guarantees that the selected camera nodes can achieve full-view images of the target continuously. Performance study reveals that the proposed full-view target tracking approach can efficiently catch the full-view images of the target continuously.

圖目錄	VI
表目錄	VIII
第一章、	簡介	1
第二章、	國內外相關研究	6
第三章、	Quality of Tracking Guaranteed Approach (QTG)	9
3-1 網路環境與問題描述	10
3.1.1 網路環境	10
3.1.2 問題描述	10
3.2 Ideal QTG (I-QTG) Algorithm	13
3.2.1 Monitoring Phase	13
3.2.2 Handoff Phase	28
3.3 Enhanced QTG (E-QTG) Algorithm	34
3.3.1 Contribution First Policy：	34
3.3.2 Density First Policy：	40
3.4 效能評估	42
3.4.1 Simulation Model	42
3.4.2 Performance Study	43
第四章、	Seamlessly Full-View Tracking Approach	52
4.1 網路環境與問題描述	53
4.1.1 網路環境	53
4.1.2 問題描述	53
4.2 Seamlessly Full-View Tracking (SFT) Approach	57
4.2.1 Initial Phase	57
4.2.2 Non-Transparent Aware Phase	59
4.2.3 Monitoring Phase	64
4.2.4 Tracking Phase	69
4.3 效能評估	74
4.3.1 Simulation Model	74
4.3.2 Performance Study	75
第五章、	結論	83
參考文獻	84
附錄-英文論文	88

圖目錄
圖(一) 無線視覺感測網路與視覺感測節點。	2
圖(二) Camera Node判斷非透明物體可實際照射範圍。	4
圖(三) I-QTG Procedure。	17
圖(四) 將目標物包圍於 內，以滿足QoT。	18
圖(五) I-QTG演算法滿足QoT及節省醒睡成本之挑選範圍。	21
圖(六) 兩圓交點符號說明。	23
圖(七) 判斷具最大貢獻度之working node的位置。	25
圖(八) Monitoring Phase Algorithm。	27
圖(九) 目標物移動出防禦區域所越過之弧線。	30
圖(十) 目標物移動時,感測器之挑選方法。	30
圖(十一) Handoff Phase Algorithm。	33
圖(十二) E-QTG之感測器挑選範圍，與貢獻度的判斷。	35
圖(十三) E-QTG目標物移動之換手方法。	37
圖(十四) 先挑選貢獻度最大的感測器為 。	39
圖(十五) 先挑選鄰居密度低的感測器為 。	39
圖(十六) Hybrid QTG 從切割區塊開始進行 之挑選。	40
圖(十七) 各演算法追蹤誤差值之比較(Qreq=1/8)。	44
圖(十八) 各演算法追蹤誤差值之比較(Qreq=1/5)。	44
圖(十九) 挑選一組 成功率之比較。	45
圖(二十) 追蹤過程中每一瞬間平均活動感測器數量之比較。	46
圖(二十一) 均勻佈點(U)。	47
圖(二十二) 非均勻佈點(NU)。	47
圖(二十三) CFP以及DFP在不同網路場景之活動感測器數量之比較。	48
圖(二十四) DFP在不同密度之網路場景之切割區塊數n值大小不同之效能比較。	49
圖(二十五) 各演算法網路總耗能之比較。	50
圖(二十六) 目標物移動速度以及δ之設定對於感測器耗能影響之比較。	51
圖(二十七) 感測器節點決定目標物形狀。	58
圖(二十八) Camera Nodes判斷 中那些線段被目標物所遮障。	61
圖(二十九) Non-Transparent Aware Phas Algorithm。	64
圖(三十) Camera Node model。	65
圖(三十一) Camera Node 判斷自己對目標物邊界之貢獻度。	67
圖(三十二) 不適當的Camera Node選擇，將無法應付目標物隨機的移動。	69
圖(三十三) 二元感測器先天誤差對於camera Node覆蓋目標物將造成影響。	71
圖(三十四) Camera Node透過低估自計的照射半徑，以及Buffer Area的設計以保證連續追蹤。	72
圖(三十五) 不具移動性之目標物覆蓋率之比較。	76
圖(三十六) 具移動性之目標物覆蓋率之比較。	77
圖(三十七) 挑選一組Camera Nodes完整覆蓋目標物邊界之成功率之比較。	78
圖(三十八) 演算法成功追蹤以及連續追蹤程度之比較。	79
圖(三十九) 目標物不同移動速度下，SFT之連續追蹤程度的比較。	80
圖(四十) Camera Node在滿足50%平均覆蓋率的條件下不同FoV，平均工作工作節點數量之比較。	81
圖(四十一) dmax大小對於工作節點數量之影響。	82

表目錄
Table I : Simulation parameters	42
Table II : Simulation parameters	74

[1]	J. CHEN and M. Matsumoto, “EUCOW: Energy-Efficient Boundary Monitoring for Unsmoothed Continuous Objects in Wireless Sensor Network,” in Proceedings of the 6th IEEE International Conference on Mobile Adhoc and Sensor Systems(MASS), pp. 906 – 911, October 2009.
[2]	J. Teng, H. Snoussi, and C. Richard, “Decentralized Variational Filtering for Target Tracking in Binary Sensor Networks,” IEEE Transactions on Mobile Computing, vol. 9, no. 10, pp. 1465-1477, October 2010.
[3]	Z. Zhong, T. Zhu, D. Wang, and T. He, “Tracking with Unreliable Node Sequences,” IEEE INFOCOM, pp. 1215-1223, April 2009.
[4]	Y. C. Tseng, S. P. Kuo, H. W. Lee, and C. F. Huang, “Location Tracking in a Wireless Sensor Nework by Mobile Agents and Its Data Fusion Strategies,” IEEE IPSN, 2003.
[5]	J. Chen, K. Cao, Y. Sun, and X. Shen, “Adaptive Sensor Activation for Target Tracking in Wireless Sensor Networks,” IEEE ICC, pp 1-5, June 2009.
[6]	Q. Ren, J. Li, and H. Gao, “TPSS: A Two-phase Sleep Scheduling Protocol for Object Tracking in Sensor Networks,” IEEE MASS, pp 458-465, October 2009.
[7]	Z. Wang, E. Bulut, and B. K. Szymanski, “Distributed Target Tracking with Directional Binary Sensor Networks,” IEEE GLOBECOM, pp 1-6, December 2009.
[8]	Z. Wang, E. Bulut, and B. K. Szymanski, "Distributed Energy-Efficient Target Tracking with Binary Sensor Networks," ACM Transaction on Sensor Networks, vol 6, 2010.
[9]	F. Hsin and M Liu, “Network Coverage Using Low Duty-Cycled Sensors: Random & Coordinated Sleep Algorithms,” Information Processing in Sensor Networks (IPSN), pp. 433-442, April 2004.
[10]	J. Hill and D. Culler, “A Wireless Embedded sensor Architecture for System-level Optimization,” Technical Report, Computer Science Department, University of California at Berkekey, 2002.
[11]	I. F. Akyildiz, T. Melodia, Kaushik R. Chowdhury, “A Survey on Wireless Multimedia Sensor Networks,” Computer Networks, October 2006, pp. 921-960
[12]	S. Soro and W. B. Heinzelman, “On the coverage problem in video based wireless sensor network,” in Proc. IEEE Cof. Broadband Advanced Sensor Networks (BaseNets ’05) , Oct 2005, pp. 932-939.
[13]	L. Liu, H. Ma, and X. Zhang, “Analysis for Localization-Oriented Coverage in Camera Sensor Networks,” in Proc. IEEE Cof. Wireless Communications and Networking Conference (WCNC), Mar. 2008, pp. 2579-2584.
[14]	L. Liu, H. Ma, and X. Zhang, “Collaborative Target Localization in Camera Sensor Networks,” in Proc. IEEE Cof. Wireless Communications and Networking Conference (WCNC), Mar. 2008, pp.2403-2407.
[15]	K. Y. Chow, K. S. Lui, and E. Y. Lam, "Maximizing angle coverage in visual sensor networks," in Proc. IEEE Cof. International Conference on Communications(ICC), June 2007, pp. 3516–3521.
[16]	K. Y. Chow, K. S. Lui and E. Y. Lam ,“Achieving 360 Angle Coverage with Minimum Transmission Cost in Visual Sensor Networks,” in Proc. IEEE Cof. Wireless Communications & Networking Conference(WCNC), October 2007, pp. 4112-4116.
[17]	T. S. Chen , C. P. Chen , H. W. Tsai , “Object Coverage with Camera Rotation in Visual Sensor Networks”, in Proc. Cof. National Computer Symposium(NCS) , Nov 2009
[18]	K. P. Shih, C. M. Chou, and I. H. Liu, “On Barrier Coverage in Wireless Camera Sensor Networks”, in Proc. Cof. National Computer Symposium(NCS) , Nov 2009.
[19]	S. Kloder, S. Hutchinson, “Partial Barrier Coverage: Using Game Theory to Optimize Probability of Undetected Intrusion in Polygonal Environments,”  in Proc. Cof. International Conference on Robotics and Automation, (ICRA), 2008, pp. 2671-2676.
[20]	A. Newell , K. Akkaya, ”Self-actuation of Camera Sensors for Redundant Data Elimination in Wireless Multimedia Sensor Networks”, in Proc. IEEE Cof. International Conference on Communications (ICC), pp. 1-5, June 2009.
[21]	T. S. Chen, J. J. Peng, D. W. Lee, and H. W. Tsai, "Prediction-based Object Tracking and Coverage in Visual Sensor Networks," in Proc. The 7th International Wireless Communications and Mobile Computing Conference (IWCMC 2011) Istanbul, Turkey, July 5-8, 2011.