整合服務DS-CDMA蜂巢式系統之智慧型呼叫允諾控制器

Abstract

為了達到提供整合服務及有效率地利用珍貴頻寬資源的目的，呼叫允諾控制在直接序列-分碼多工擷取蜂巢式系統中扮演了一個很重要的角色。在本論文中，我們發展了一個智慧型呼叫允諾控制器。 這個智慧型呼叫允諾控制器主要由平行迴路類神經網路(pipeline recurrent neural network)干擾平均值預測器，乏晰等效干擾評估器以及呼叫允諾處理器所組成。智慧型呼叫允諾控制器利用平行迴路類神經網路干擾平均值預測器來預測下一時刻的干擾平均值，用乏晰等效干擾評估器來預估新進呼叫的訊務特性，並且參考系統的損壞機率(outage probability)來保證多媒體服務的服務品質(QoS)。 我們考慮以兩種不同的人工智慧理論來實現呼叫允諾控制器，分別稱為乏晰呼叫允諾控制器以及類神經乏晰網路呼叫允諾控制器。實驗的結果顯示我們所發展的乏晰呼叫允諾控制器以及類神經乏晰網路呼叫允諾控制器比起傳統以訊號干擾比為基礎呼叫允諾演算法可以有較高的系統容量以及較低的阻滯率(blocking probability)，並且仍然可以維持服務品質。由於類神經乏晰網路呼叫允諾控制器同時具有參數及乏晰法則架構學習的能力，所以類神經乏晰網路呼叫允諾控制器可以比乏晰呼叫允諾控制器達到較高的系統容量以及較低的阻滯率。在此論文中，我們也討論在變動的訊務情形中，智慧型呼叫允諾控制器的處理能力。

In order to provide the integrated services and utilize the precise radio resource efficiently, a sophisticated call admission control scheme is essential to DS-CDMA cellular systems. In this thesis, we develop an intelligent call admission controller (ICAC) for DS-the multimedia services. We design two alternative intelligent call admission controllers, FCAC and NFCAC. The FCAC uses the fuzzy logic CDMA cellular systems. The intelligent call admission controller consists of the PRNN interference mean predictor, the equivalent interference estimator and the call admission processors. The ICAC employs the PRNN to predict the next-time interference level, the fuzzy logic controller to catch the traffic characteristics of the new call user, and takes the outage probability as a input parameter to guarantee the QoS which is a necessary factor for theory to implement the call admission processors and the NFCAC employs the self-constructing neural fuzzy inference network (SONFIN) to do that. SONFIN has the structure learning ability to optimize the fuzzy rules structure and parameter learning ability to fine tune the parameters of the membership functions. From the simulation results, the NFCAC and FCAC have lower blocking probability and higher system capacity than the SIR-based CAC algorithm while keeping the QoS requirements. On the other hand, the NFCAC outperforms the FCAC in system capacity and blocking probability.In this thesis, we also consider various traffic conditions in the ICAC.