標題: 行動電話系統之區域規劃Location Area Planning in a Cellular Phone Network 作者: 李佩君Pei-Chun Lee蔡中川林一平Jong-Chuang TsayYi-Bing Lin資訊科學與工程研究所 關鍵字: 區域規劃;整數線性規劃;類神經網路;location area planning;integer linear programming;artificial neural network 公開日期: 1999 摘要: 在一個行動電話系統中，面對廣大且持續增加的行動用戶，系統裡的信令 (signaling) 流量勢必會隨之大幅增加。因此降低系統裡的信令流量，以減輕系統運作負擔，便成為系統設計及運作時一個重要的考量。系統裡主要的兩種信令型態分別為行動電話搜尋信令 (paging signaling) 和註冊信令 (registration signaling)，因為區域規劃 (location area planning) 的不同，這兩種信令對系統造成的負擔形成一個互為取捨 (tradeoff) 的局面，換句話說，當區域規劃使行動電話搜尋信令流量降低時，註冊信令流量就會增高；反之亦然。這使得區域規劃的問題在本質上乃是一個最佳化的問題。 在區域規劃這個領域裡，已有許多研究者使用各樣不同的方法來解決這個問題。本論文將使用整數線性規劃 (Integer Linear Programming) 以及類神經網路 (Artificial Neural Network) 這兩種方法來解決這個問題。 首先，我們先定義區域規劃問題;接著，我們分別採用整數線性規劃和類神經網路的方法來解此問題。雖然採用整數線性規劃方法可解得最佳解，但其解問題之時間複雜度很高。反之，採用類神經網路方法雖然僅能求得近似最佳解，但其時間複雜度比整數線性規劃方法小很多。In a cellular phone network, mobility management may cause heavy traffic to signaling network. Two kinds of signaling, paging signaling and registration signaling, constitute the net signaling load in a cellular phone network. Depending on the location area planning, when the paging signaling load (paging cost) is reduced, the registration signaling load (registration cost) increases; and vice versa. Thus location area (LA) partitioning is an optimization problem, which should balance against two conflicting signaling costs. Several approaches have been proposed to solve the LA partitioning problem. In this thesis, we propose two novel approaches, integer linear programming (ILP) approach and artificial neural network (ANN) approach, to solve the LA partitioning problem. We formulate the LA partitioning problem as an optimization problem. Next, we formulate the cost function to an integer linear programming problem and then an artificial neural network energy function. Then, we adopt the integer linear programming approach and the artificial neural network approach to solving the problem. Although the integer linear programming approach can find the optimal solution, its time complexity is high. On the other hand, although the artificial neural network approach yields only near optimal solution, it can solve the problem with much less time complexity. URI: http://140.113.39.130/cdrfb3/record/nctu/#NT880392008http://hdl.handle.net/11536/65403 Appears in Collections: Thesis