行動計算環境中初始及交遞連線之效能模型與模擬

Abstract

行動計算之通訊環境可以蜂巢狀╱微蜂巢狀電話系統或個人通訊服務( PCS) 網路為基礎建構。在這樣的環境下，交遞的發生就無法避免。當交 遞發生時，行動電腦所使用的頻道可能會被強制終結。交遞連線要求的強 制終結機率對行動計算環境的效能有很大的影響。此外，初使連線要求的 攔阻機率也是一項重要的效能指標。目前已有許多關於蜂巢狀電話系統及 PCS 網路之初使與交遞的研究。然而，這些研究中的分析模型並不適合行 動計算，因為每部電話只使用一個頻道，而每部行動電腦可能使用多個頻 道。在本論文中，我們發展了一個分析模型以分析型動計算環境中的初使 及交遞連線行為。這個模型是由一個多維連續時間馬可夫鏈建構。這個模 型的正確性可由模擬來做確認。此外，我們也發展了另外一套模擬模型來 評估經常性中斷對行動計算環境的效能所產生的影響。模擬的結果顯示， 提高經常性中斷的機率會降低系統的效能。最後，我們提出一個平衡初始 連線攔阻機率與工作不完成機率的方法。 The communication infrastructure of a mobile computing environment can be built on the basis of a cellular/ microcellular telephone system or a PCS network. Under this circumstance, the occurences of handoffs cannot be avoided. With the occurences of handoffs, wireless links held by mobile computers crossing cell boundaries may be forced terminated. The forced termination probability of handoff access requests has a significant effect on the performance of a mobile computing environment. Moreover, the blocking probability of initial access requests is also an important metric of performance measurement. Several researches have been done on the analysis of the initial and handoff accesses in cellular/ microcellular telephone systems and PCS networks. However, the analytical models used in these researches are not appropriate for mobile computing since each telephone uses only one channel but a mobile computer may use several channels. In this thesis, we develop an analytical model to analyze the initial and handoff accesses in a mobile computing environment. This model is constructed with a multi-dimensional continuous time Markov chain. The accuracy of this model is verified by simulation. Besides, we also develop another simulation model to examine the effects of frequent disconnections. The simulation results show that increasing the probability that a wireless link will be disconnected may seriously degrade the performance of a mobile computing system. Furthermore, we provide an approach to balance the initial access blocking probability and the task incompletion probability.