內容感知的應用在今日已經變的越來越熱門，而位置資訊的可知也因此衍生出許多研究的議題。這篇論文提出了一套精準的室內無線網路系統名為Precise Indoor Location System (PILS)。大部分擁有良好定位精準度的定位系統都必須在事情花費許多的人力在收集大量的訊號上面，使得定位系統的變的不實用與需求過多的人力資源。在這篇論文裡，我們將目標放在減少在建置訊號地圖上的人力資源耗費並且保持住定位系統的精準度在一個可以接受的範圍。我們也提出了在資料收集上、訊號內插上、以及位置估計上的模型。另外我們也考慮了一連串連續訊號的相關度來提高準確度。無線網路訊號傳遞的特性也是我們研究的一部份，大小範圍的遮蔽包含在我們所研究的訊號傳遞現象裡面。最後我們提出了一套學習的模型來調整我們的訊號地圖，以改進因為測量數目的減少所造成的精準度下降。 / Context-aware applications become more and more popular in today’s life. Location-aware information derives a lot of research issues. This thesis presents a precise indoor RF-based WLAN (IEEE 802.11) locating system named Precise Indoor Locating System (PILS). Most proposed location systems acquire well location estimation results but consume high level of manual efforts to collect huge amount of signal data. As a consequence, the system becomes impractical and manpower-wasted. In this thesis, we aim to reduce the manual efforts in constructing radio map and maintain high accuracy in our system. We propose the models for data calibration, interpolating, and location estimation in PILS. In the data calibration and location estimation models, we consider the autocorrelation of signal samples to enhance accuracy. Large scale and small scale fading are involved in the wireless channel propagation model. We also propose a learning model to adjust radio map for improving the accuracy down caused by calibrated data reduction.