行動裝置平台上之樣本比對定位演算法的快取記憶體管理

Abstract

隨著行動裝置與區域無線網路的普及，位置相關系統與服務獲得越來越多的注意。現今熱門的GPS 定位系統由於在室內無法定位的缺點，導致其不適用於位置相關服務。相反地，類似RADAR 的定位系統在室內外都可以定位，此外，不需額外硬體成本使的它更適用於位置相關服務。然而目前以RADAR 為基礎的定位演算法都需要大量的資料與龐大的計算量來達到可接受的精準度，因此並不適合在硬體資源有限的行動裝置上運行。為了解決此問題，本論文提出了一個新穎的記憶體管理機制與預測下載機制。透過這些方法，我們希望能實現一具有一定精準度且較不須依賴後端伺服器的行動裝置端樣本比對定位系統。

The proliferation of mobile computing devices and local-area wireless networks has fostered a growing interest in location-aware systems and services. Current positioning technique, GPS, are not suitable for more and more LBS(Location Based Service) because of its incapability of indoor positioning. In contrast, RADAR-like system can work both indoor and outdoor. Furthermore, they only need to use existing hardware and are, thus, more suitable for LBS. However, existing RADAR-like algorithms all need extensive data and large amount of computation to achieve acceptable accuracy. Therefore, not suitable for positioning on resource limited portable device. In order to solve this problem, we propose a novel cache management scheme and prediction method in this paper. By using these techniques, we hope to achieve a client-based pattern-matching positioning system with fair enough accuracy and less dependent to server.