Network Selection In Heterogeneous Networks

Due to the rapid development of radio access technology with great variety, networks which are different in capacity, coverage, QoS level, and cost integrate into a complicated wireless communication environment."HETEROGENEITY" has become the most important characteristic of the next generation mobile communication system. At the same time, the increasingly sophisticated radio technology makes the multi-channel multi-mode terminal available in reality. In this context, we investigate the network selection in this dissertation, which is a new issue of the radio resource management in heterogenous networks. The paper includes both the establishment of the CRRM architecture and the algorithm design. We intend to design a common network selection strategy which is independent of any specific RAT.In this dissertation, we study the subject of network selection gradually as the following order:At first, we discuss the radio resource management architechture design in heterogeneous networks, which is the basis of algorithm development. Next, because the network selection procedure always causes extensive signaling overhead and access delay, we try to deal with this problem by designing an appropriate strategy. Then, we discuss the single-target network selection algorithm from either a network-centric or a user centric perspective. After that, further study of multi-target network selection algorithm is developed. The usage of fuzzy logic, neural network, multi-objective decision making and bayesian forcasting for network selection is analized respectively. At last, the network selection based on multi-connection is proposed, and consequently, the traffic synchronization issue is also studied. During our analysis, we make full use of mathematics and simulation tools to modeling and verifying our work.The main contribution of this dissertation includes the following aspects:At first, the Common Radio Resource Management (CRRM), the basic issue of network selection is studied. In the second chapter, a distributed CRRM architechture with central control is proposed, which is used for loosely-coupled networks with universal adaptability. This architechture benefits both from the traditional distributed and the centralized architechture.Then, the single-objective network selection of single-connection is studied from both the network-centric and the user-centric perspective, and is discussed in the third and the forth chapter, respectively. In the third chapter, a semi-preactive network selection mechanism is described from the network-centric perspective, which can reduce the signal load and accessing delay significantly. Meanwhile, a load balance network selection algorithm is proposed for performance analysis. In the forth chapter, a user-centric network selection algorithm is proposed, which can reduce the outage probability of SIR and thus improve the communication performance.Next, the multi-objective network selection is discussed in the fifth chapter. A bayesian dynamic linear delay forcasting model is built, and a network selection algorithm based on multi-objective fuzzy inference is proposed. This method can improve both the connection level and the packet level QoS.Finally, the network selection in a multi-connection scenario is studied. A dynamic data rate allocation strategy which considerably e nhances the synchronazition performance of sub-traffic streams is proposed.