以訊號強度為指紋辨識的定位方式,因每個位置的特徵亦會將障礙物遮蔽以及多重路徑等常見的室內訊號傳遞現象納入考量,而廣泛被使用在室內定位。在使用這樣的定位方式時,定位的準確度會被某些因素所影響,而不同於測試平台,在真實環境中這些因素往往被種種現實的限制所侷限,進而造成準確度的下降。因此,為了驗証室內定位系統於現實環境中的可行性,我們在世界貿易中心第三展覽館「快閃佈建」了一個從架設、運作到撤除僅僅為期五天的以訊號強度為指紋辨識的室內定位系統。 在這篇論文中,我們首先分享了包含在世貿所使用硬體以及定位機制描述的快閃佈建實際經驗,接著,討論快閃佈建所遭遇到的限制和挑戰,並呈現定位準確度受其影響的程度,最後針對這些限制和挑戰提出可能的改善方式。 從分析的結果顯示,佈建於世貿的室內定位系統的80百分位數為9.01公尺(在使用我們討論的多個定位機制中最準確者的情況下)。而在世貿的限制和挑戰包括訊號發射器擺放位置的限制、時間和空間的限制、環境改變的挑戰以及訊號不穩定的挑戰等四者,其中,訊號不穩定造成最大的定位誤差,也因如此,針對其所提出的可能改善方式 – 額外使用訊號強度以外的其他類「指紋」以及使用貝式過濾式(Bayesian filter),為改善快閃佈建情況下以訊號強度為指紋辨識室內定位系統準確度較為有效之方法。
RSSI fingerprinting-based localization technique is widely used for indoor localization because it could capture obstruction and multipath effect, which are commonly seen in indoor signal propagation, as unique feature for each location. Unlike testbeds, there are usually various constraints in the real world environment and the accuracy of location estimations may be reduced indirectly due to some factors limited by them while using RSSI fingerprinting-based technique. Therefore, we implemented a flash deployment of RSSI fingerprinting-based indoor localization system at World Trade Center (WTC) exhibition hall 3 for 5 days only to verify the feasibility of it in the real world environment. In this work, we shared our practical experience of flash deployment firstly. It includes the detail descriptions of the used hardware and the implemented localization mechanisms followed by the evaluations of them. Secondly, the challenges and limits of flash deployment at WTC are well discussed. In addition, the evaluations of the effects of them on the accuracy are presented and the possible improvements are proposed. The result indicates that the error distance of the indoor localization system deployed at WTC is 9.01m at 80th percentile with the most accurate localization mechanism among the ones we discussed. Beacon placement limit, time and space limit, environmental change challenge, and signal variation challenge are the four challenges and limits for flash deployment at WTC. Among these, signal variation introduces the most error distance. Thus, the possible improvements to it, using other kinds of fingerprint in addition to RSSI and using Bayesian filter, may be the most effective ways to improve the accuracy of location estimations for flash deployment of RSSI fingerprinting-based indoor localization system.