基於骨架的混合無線定位與參考點部署

Abstract

地圖的空間骨架可以幫助室內定位，避免一些不合理的情況發生， 並實現準確的定位估計。在本論文中，我們提出了一種基於骨架部署 參考點的自動方法，稱之為骨架參考點自動部署演算法。基於接收信 號強度和骨架最短路徑，我們提出了接收信號強度與骨架最短路徑聯 合相似度聚類演算法對被骨架參考點自動部署演算法部署的參考點進 行了聚類。我們模擬了接收信號強度與骨架最短路徑聯合相似度聚類 演算法的最佳權重為八成的接收信號強度加上兩成的骨架最短路徑。 集群的信息被存儲進了群集數據庫中。群集數據庫由骨架最短路徑矩 陣、群集與接收信號強度測量值的信息組成。此外，我們提出了候選 參考點的選擇函數，選擇相似實時接收信號強度測量值的群集。在本 論文中，為了搭配候選參考點時的定位表現可以更好，我們重新設計 了骨架加權最近鄰居法。最後，基於骨架參考點自動部署演算法、接 收信號強度與骨架最短路徑聯合相似度聚類演算法與骨架加權最近鄰 居法及低功耗藍牙小區辨識，我們提出了基於骨架的混合無線定位與 參考點部署。根據實驗結果，基於骨架的混合無線定位與參考點部署 可以實現準確的位置估測。

Spatial skeleton of map can assist indoor localization to avoid some unreasonable cases and achieve accurate location estimation. In this thesis, we proposed an automatic method, based on skeleton to deploy reference points (RPs), called skeleton-based automatic RP deployment (S-RPD) algorithm. Based on received signal strength (RSS) and skeleton-based shortest path (SSP), we proposed joint similarity of RSS and SSP (JSSP) clustering algorithm to cluster the RPs deployed by S-RPD. The best weights we simulated of JSSP clustering algorithm are 0.8 for RSS and 0.2 for SSP. The information of clusters is stored into clustered database (CDB). CDB is made up of the information of SSP matrix, clusters, and RSS measurements. Moreover, we proposed the selection function of candidate RPs, the function which select the clusters of CDB similar to the real-time RSS measurements. In this thesis, in order to make the positioning performance better with candidate RPs input, we redesigned skeleton-based weighted KNN (S-WKNN) algorithm. Finally, we proposed skeleton-based positioning and reference point deployment for hybrid wireless localization (S-PDH) based on S-RPD, JSSP clustering, S-WKNN, and Bluetooth Low Energy (BLE) cell-id. According to the result of the experiment, S-PDH can achieve accurate location estimation.