Network management and application design in software defined wireless networks

Abstract

With the development of big data technologies, the abundant mobile network information and traffic data provide opportunities for researchers to analyze and understand mobile networks better. Software Defined Networking (SDN) is a flexible, scalable and programmable way for network management. Combining big data technology and SDN in wireless networks can exploit the potential of big data analysis for network management and provide a wide range of applications.

In this thesis, we focus on network management and application design in software defined wireless networks. We firstly propose a new software-defined cellular network (SDCN) architecture, namely, Big-Data-Enabled Architecture (BDEA), which can support big mobile data analysis and storage for efficient cellular network resource allocation. Based on BDEA, we propose a virtuous network management cycle of data collection, data analysis, and network deployment, which can increase the efficiency of network feedback and improvement.

Secondly, we introduce data prefetching as an application of network resource allocation in BDEA. Taking advantage of centralized computing capability and packet path control, data packets are prefetched to possible target base stations according to the prediction of a user's next location. Numerical results show that the average delay of the whole user group can be reduced dramatically with the data prefetching strategy.

Finally, we study topology management in Software-Defined Vehicular Network (SDVN) and discuss the feasibility of deploying SDN in a network with frequent topology changes. We introduce the procedures of topology discovery in SDVN. Four models are introduced to analyze the features of topology changes in vehicle-to-RSU (V2R) and vehicle-to-vehicle (V2V) communications on highways and in urban areas. Simulation results show that a reasonable topology update interval can provide enough topology information to ensure successful data transmission in SDVN. We conclude the thesis with suggestions for further research.