Research On Ontology- And CPN-based Formalization Of Resource Allocation Problems

With the ever-rising of Internet technology and emerging network applications,the traffic in the network is increased explosively.Elastic Optical Networks(EONs)based on optical orthogonal frequency division multiplexing could provide large transmission capacity and dynamic flexible configuration capabilities,and it also supply the flexible resources for software-defined networks,data center networks and other networks.This thesis is sponsored by the National Natural Science Foundation of China's project entitled “Novel intelligent algorithms and formalization of resource optimization in hybrid elastic optical networks”,studying the formalization of the issues related to resource allocation in EONs.Although a lot of researches have been done on EONs at home and abroad,these studies have focused on micro aspects,exploring solutions to some kinds of problems in specific network scenarios.There is a lack of research on the formalization description of problems with traffic source and intelligent algorithms in EONs from macro aspects.This thesis focuses on the formalization of problems such as traffic source models and intelligent algorithms in EONs.By making the comparison and selection of suitable formalization tools,the above issues are described formally and corresponding knowledge bases are established to deepen the cognition of some aspects of the above issues.The main contents and innovations in this thesis are listed as following aspects:By establishing different mathematical models such as Poisson model,time series model and wavelet model,the traditional traffic source description methods could describe the flow characteristics of each traffic source.These source description methods could establish a specific traffic source model for specific problems such as traffic forecasting problem,and specific algorithms are proposed to solve certain problems.Due to the lack of a comprehensive description of the traffic source models from a macro aspect,it is difficult to form a relatively complete knowledge base,which makes against its knowledge sharing and reuse.In this thesis,from the perspective of knowledge representation,we apply the Ontology on the formalization description of traffic source models domain knowledge and build the Traffic Source Models Ontology(TSMO).First,this thesis extracts and analyzes the core concepts and related properties of the traffic source models field,then constructs TSMO based on the Web Ontology Language(OWL).Second,by using the Simple Query and RDF Query Language(SPARQL)and Ontology inference engine combined with specific application scenarios,we perform the knowledge retrieval and inference of various types on the established ontology,and verify the rationality,practicality and versatility of TSMO which is used to serve as an Ontology database.Besides,TSMO can provide a semantic platform prototype of traffic source models domain for researchers and users.There are many resource allocation algorithms in EONs.The algorithms of same kind could be used to solve different specific problems,which results in it difficulty for each module in different algorithms to share and reuse.However,we could find the commonness of some algorithms by abstracting them,for instance,the compute modes of these algorithms are similar to each other,and partial modules in these algorithms are universal.Based on this analysis,this thesis uses the formal method of the Colored Petri Nets(CPN)and its formalization tool(CPN Tools),and takes the genetic algorithm that is one of intelligent algorithms,as an example.By the abstraction of crossing operators,mutation operators and sub-process of the algorithm,the formal description of each module in the genetic algorithms is constructed.In addition,the state space tool of CPN Tools is used to verify the formal model of the genetic algorithm from two aspects of functionality and correctness.