Researches On The Theory And Application Of Complex Network

The intention of this research is to give a systematic study for some real complex networks in theory and application. Chinese phrase network is selected as our first focus of real complex network. Its organization and topological properties are studied in detail and an evolution model to describe the behaviors is put forward. Ad hoc network is our key chosen as a sample of complex network application. The concept of'the view of complex network'is suggested to depict the role of complex network playing in ad hoc net, which is used to optimize DSR routing protocols in our work. And also a generalized theoretical model for coupling network is proposed and analytical results are obtained.A generalized evolutional model for coupling network is proposed based on the model introduced by Zheng et.al. Coupling in real networks is analyzed and the connections between old nodes in existing networks are considered during modeling the evolution of coupling. Rate equation is set up, the degree distributions of network are solved exactly and asymptotic power-law results with corresponding exponents are obtained. When the parameters are selected properly, our generalized model can give consistent results with that of some real networks, and can be adapted to more real situations.Language networks are important sample of complex network research. Chinese language network is in relative an unexplored field as a study object. Chinese phrase networks are defined based on the idea of complex network and three groups of real phrases, structure and organization of phrase networks are analyzed, topological properties are calculated and some evolutional behaviors are uncovered. We get important results which show that Chinese phrase networks present small-work patterns and power-law degree distributions, and the average separation is 3. A random text model—Monkey language model for Chinese phrase networks is defined to analyze and predict their dynamic behaviors. It is found that direct random Monkey model can not depict real networks well, and as some ingredients such as word selection probability and the length of phrase are taken into consideration, i.e. the randomness of Monkey model is reduced in proper ways, it will present more similar properties as natural language. At the same time, our model reveals some general principles of human language during its evolutions, for example,'the least effort principle'can be induced .from our Monkey model. The performance of the model is simulated at every step and compared with that of real networks. And also, comparisons of small world effects among Chinese phrase network, English networks and other few language networks are made, and we conclude that Chinese phrase network and English concept network belong to a special class of small world networks not included by Amaral. A brief discussion for the potential application of language network researches is given too.Application of the ideas of complex network is a new research interest in recent year. MANET(Mobile Ad hoc NETwork)is chosen as a place to realize our attempt which includes applying complex network concepts to optimize routing protocols and improve performance of networks. Starting from basic architecture of MANET, we first try to find the intersection of complex network and Ad hoc network, and taken as a generalized topology control method, a sub-layer called'complex network view'is added to the protocol stacks. Then a model is introduced and a small world network is constructed based on application scenes of MANET.The view of complex network can extract useful information from the basical network topology or construct a complex network over the basical network topology. It is a kind of generalized topology control which can provide routing protocols up layer with special characterized topologies. DSR (Dynamic Source Routing) protocol is studied in detail and optimized in its route discovery and route maintenance. When network emerge small-world and scale-free patterns through the view of complex network, DRS can be optimized further. Degree of separation property is used to improve performance of DSR in route discovery phase, and congestion problem caused by hub nodes is analyzed and a modification in DSR is suggested. Simulations are conducted for all optimization process mentioned above using ns-2 network simulation tool. Results indicate that the view of complex network can give important helps in routing protocols design, and the properties of complex network can be useful to improve network performance.Our works about Chinese phrase network and the application of complex network in MANET are entirely new research areas. We believe that this study is significative to complex network, Chinese languages, and Ad hoc network researches. Simultaneously, many problems are put forward for further studies in the future.