現今生活中行動定位服務已越來越受歡迎。因此,定位精準度是必需要的。在複雜的市區或是室內環境,射頻資料的建立需投入大量人力資源才能達到滿意的效能,我們希望減少量測數據的人力及時間成本。我們的方法是建立是在一個零配置且穩健的定位系統。其中,多變量回歸是一個線性方法,我們用來呈現訊號強度與距離的關係。在這篇論文中,我們提出了三個方法來量化量測點的重要性。當我們可以量化量測點的重要性,我們就可以避免盲目的收集資料,可以更有效的運用資源。實驗結果顯示,不管在任何定位平台,我們的方法都可以有效的量化量測點。我們可以選擇最好的方法來強化定位系統。因此,我們提出的方法可以有效的運用在資源有限的情況,減少資料收集的時間及人力資源。
The location-based service become more and more popular in today's life. Therefore, the accuracy requirements are highly. RF database was build that need a lot of time and human resource to achieve satisfactory performance. We reduce the time and human resource of data collected. Our proposed methods are based on a zero-configuration, robust localization system. Multivariate Regression (MR) performs a theoretical study on linear method. It can represent the relationship between distance and signal strength. In this paper, we propose three methods to quantify the importance of training points. When we quantify the importance of training points, we can avoid blind collecting data. The results present that our proposed methods can effective quantify the importance of training points on any localization system. We can select the best methods to strengthen the positioning system. Hence, our proposed methods can be used when resources are limited. We also can e ective solve the time and human resource of data collected.