| Summary: | 博士 === 國立臺灣大學 === 資訊管理學研究所 === 93 === Throughout the world, wireless communications are experiencing extraordinary growth – encompassing and transforming every aspect of our lives, and creating an “always connected” society. To meet ever-growing user demand for much higher data rates, the ongoing migration of voice services from wired to wireless systems, as well as ever-increasing data applications, figure prominently in the drive toward fulfillment of the CDMA-based third-generation (3G) system. This dissertation investigates a number of topics related to mobile wireless communications in DS-CDMA networks. They are integrated in a research framework that, from a macro perspective to a micro perspective, consists of network planning, resource management, and call admission control problems. The planning problem considers adaptive sectorization and a hybrid F/CDMA scheme under QoS/GoS constraints. For resource management problems, adaptive load balancing and network survivability when the network is in operation are discussed. Call admission control problems are comprised of three issues: optimal-revenue, real-time processing, and prioritized handoff control. All problems are formulated as mathematical optimization models, and Lagrangean relaxation is applied as the solution approach.
The contributions of this dissertation are highlighted by the following. The results of the planning problem show that costs are reduced by as much as 55%, when survivability is considered; the achievement of the survivability model is a reduction in time complexity from O(N2) to O(1). The model also offers capacity expansibility. The performance of the proposed adaptive scheme (AS) for load balancing outperforms the non-adaptive (NA) approach by up to 68%. In admission control problems, the optimal revenue model is approximately 8% better than other approaches. In addition, we propose a real-time model to deal with long term revenue analysis. The prioritized model outperforms other approaches in the reduction of the blocking rate. The improvement is up to 70% better than the cutoff priority scheme (CPS), and 75% better than the complete sharing scheme (CSS).
Based on the results, this dissertation suggests several engineering guidelines as follows: network planning that consists of cell planning, which addresses user coverage, and a topology design, which plays an important role for the integrated network planning problem, can solve global optimization. The survivability model fits the requirement of capacity expansion by locating mobile/portable base stations in the places they are most needed. Load balancing achieves significantly more when the AS approach is combined with sectorization. In the optimal revenue admission control model, to accommodate as many as users as possible, the strength of power-controlled signals needs to be properly assigned; in the real-time model, the time budget is a more significant factor than the mean call holding time. The proposed dynamic guide channel (DGC) scheme outperforms other approaches when a cell has fewer channels, and reduces the handoff blocking rate more easily when handoff traffic is heavy.
|
|---|
