室內使用者定位在近幾年有蓬勃的發展。為了找到使用者的位置,相關的研究者已經發展出諸多的演算法,包含了卡爾曼濾器(Kalman filter)、粒子濾器(particle filter)以及隱藏式馬可夫鏈(hidden Markov chain)。儘管這些已經被發展出來的演算法並不相同,但目的皆為個別估計每位使用者的位置。這篇碩士論文所提出的位置估計演算法與前述既有的演算法不同,我們整合了使用者間合作與隱藏式馬可夫鏈演算法,成為三個不同的合作式定位演算法。這三個演算法分別為不同的計算複雜度需求而設計,共通點是使用相遇的使用者所獲得的估計結果來互相幫助以提升估計的準確度。接著,我們更提出了一個反複運算的機制來更加增進定位效能。對於傳統非合作式的演算法,模擬結果可以驗證我們的合作式演算法確實可以提供龐大的效能增進;而一如預期,模擬結果也顯示合作方式使得定位的準確度得以隨著使用者數目的增加而上升;最後,我們也發現反覆運算的機制可以提供更多的定位校能增進。總而言之,我們提供的演算法在合作式定位的領域樹立了一個新的典範。
Indoor user positioning has drawn lots of interests in recent years. To find the positions of the concerned users, researchers have developed numerous algorithms mostly based on Kalman filters, particle filters, or hidden Markov model (HMM). Despite the distinctions, existing works generally conduct the position estimation individually for each user. The inter-user information has been seldom considered. In this thesis, on the other hand, we integrate cooperative schemes into our proposed three HMM-based algorithms. Designed for different requirements for complexity, these algorithms generally utilize the estimation results of the encountered users to help improving accuracy. We further introduce an iterative method that can greatly enhance the positioning performance. For verification, simulation results show that our cooperative algorithms provide significant gains over the conventional non-cooperative works. As expected, the contribution of the cooperative scheme to positioning accuracy rises considerably with the increase of the number of collaborative users. By conducting the iterative positioning algorithm, an even more gain in accuracy can be achieved. In all, the proposed algorithms set new paradigms for the concept of cooperative positioning.