【作者】 刘干；

【导师】 朱光喜；

【作者基本信息】 华中科技大学 ， 信息与通信工程， 2005， 博士

【摘要】 随着无线通信与网络技术的迅猛发展,人们获得的移动服务越来越普及,越来越多样化,下一代无线通信网络将承载话音、数据、多媒体业务,如何保证多用户、多业务的服务质量(QoS)是无线网络中一个至关重要的问题。在另一方面,为了保证多业务的服务质量,常常以降低带宽利用率为代价。众所周知,在无线网络中,带宽资源是一个十分宝贵的资源。迫切需要一种有效的带宽分配机制,对这两个相互冲突的目标进行权衡:在保证多业务对服务质量(QoS)不同需求的前提下,尽可能充分地利用无线带宽资源。在本文,我们将探讨:在无线多媒体蜂窝网络中,保证多业务服务质量的带宽分配动态方法。分成两大部分:第一个部分研究基于呼叫连接级的动态呼叫接纳控制策略,这是粗粒度的带宽分配; 第二个部分研究基于PDU 级的自适应的带宽分配,属于细粒度的带宽分配。两者均是针对无线多业务蜂窝网络环境的QoS 控制:前者保证呼叫切换中断概率的服务质量,后者保证呼叫的时延、抖动、吞吐量。在此前提下,实现带宽资源的充分利用。在第一部分中,我们侧重于将自适应和动态的方法,应用于多业务无线蜂窝网络呼叫接纳控制中。新一代无线网络将能承载多媒体业务,在蜂窝小区越来越小,切换越来越频繁的趋势下,急需一种有效的呼叫接纳控制策略来保证无线多媒体业务的服务质量,同时又能保证无线带宽的利用率。传统的静态策略GC(保护信道策略),不能适应流量模式的变化,动态的方法利用传统的生灭过程理论,能适应流量模式的变化,但只能用于单业务,在多业务环境下,各业务的流量模式、呼叫占用信道、QoS需求各不相同,实际上是一个混杂多类业务的广义生灭系统。难以用传统的生灭过程理论加以解决。本文在第一部分的主要工作如下: 首先,提出针对多业务的参数合并方法和参数合并理论。通过建立一个伪随机生灭模型,将多维的随机生灭过程转化为传统的一维生灭过程问题,并且从理论上证明:转化后(一维空间)的呼叫中断概率CDP 近似计算值不大于实际值,而且近似相等。这一定量结果为严格的QoS 随机控制带来方便:只要使得该近似值小于预定的呼叫中断概率CDP 上限值,就能保证实际的呼叫中断概率CDP 严格满足QoS 要求。而且更多还原

【Abstract】 The great new era of information has witnessed rapid growth in the development and deployment of wireless services, evident from the proliferation of cellular data services and the emerging wireless multimedia applications. The new generation of wireless networks is expected to eventually carry multimedia traffic, including voice, video, images, data, or combinations of these, thus makes the quality of service (QoS) guarantee essential. However, to guarantee quality of multiservices usually comes at the expense of potentially poor bandwidth utilization. It’s well known that the bandwidth is extremely valuable resource in wireless networks. So an efficient bandwidth allocation scheme is urgent to balance these two conflicting requirements. This paper addresses issues of the effective bandwidth allocation policy to guarantee the quality of service (QoS) for multiple services in mobile wireless cellular networks. This thesis aims to apply dynamic and adaptive techniques to bandwidth allocation and consists of two parts. The first part concerns a dynamic and effective call admission control policy, which is belong to coarse-granularity bandwidth allocation based on call connection level, while the second part presents an intelligent bandwidth allocation scheme, which is belong to delicate-granularity bandwidth allocation based on PDU level, all for multi-services wireless cellular networks. In the first part, we apply dynamic stochastic control methods to call admission control for multi-services in wireless cellular networks. To guarantee the quality of service (QoS) for multi-services as well as achieve optimum utility of valuable bandwidth in wireless networks, an effective call admission control is urgent. Recently, the dynamic call admission mechanism was found to be more stable and precise. But, it is still hard to get an effective dynamic call admission mechanism for multi-services. The variety and complexity of traffic, the various requirements of bandwidth and QoS guarantee make it a complicated problem. Most of the relative studies cope with this problem by setting up a multiple dimensions’stochastic model with high computational complexity. In this paper, we propose some new methods to solve this problem. At first, we put forward a novel method to cope with the multi-services by setting up a fictitious stochastic model to turn the multi-dimensions stochastic processes into single-dimension one. We found a very interesting fact that the approximation of CDP is a little larger than the actual one. This fact facilitates our estimation for the acceptance ratio vector subject to QoS strict requirement. In broadband condition, the control is more precise. As a result, we get a multi-services dynamic call admission scheme called MDCA to adapt for multiple types of services in mobile wireless networks. Secondly, we study the boundary conditions impacting the control precision so as to get a more precise one. This novel study method and the fictitious stochastic model can be widely applied to many other areas with multi-dimensions birth death process. At last, we found it is enough to use a simplified stochastic states transition model to compute the key parameters, such as call dropping probability and admission probability. Although the computation is more complex than that of the MDCA, a fast numerical method has been found to decrease computational complexity and get a scheme named OMDCA steadily satisfies the hard constraints on call dropping probability for multi-services while maintaining a very high channel throughput. In the second part, we study PDU-based bandwidth allocation. The traditional methods, which allocate bandwidth according to the information of bandwidth application, are hard to avoid the allocation surplus or shortage because of delay. To overcome the shortcoming, we propose some adaptive forecast methods to improve bandwidth utility and feed back the information of bandwidth requirement by piggyback to further lower bandwidth cost. As a result, we get a high bandwidth utility while meet the multi-services QoS. In short, this paper aims to apply all kinds of adaptive technologies into bandwidth allocation. We propose some improved methods and put forward a novel stochastic model, which can be widely applied to other applications with multi-dimensions stochastic processes. Some practical control schemes are given. Numerical results of simulations show that these proposed schemes are better than the traditional schemes.更多还原