Ontology-driven knowledge based autonomic management for telecommunication networks : theory, implementation, and applications
MetadataShow full item record
Current telecommunication networks are heterogeneous, with devices manufactured by different vendors, operating on di↵erent protocols, and recorded by databases with different schemas. This heterogeneity has resulted in current network managements system becoming enormously complicated and often relying on human intervention. Knowledge based network management, which relies on a universally accepted knowledge base of the network, has been discussed extensively as a promising solution for autonomic network management. To build an autonomic network management system, a universally-shared and machine interpretable knowledge base is required which describes the resources inside the telecommunication system. Semantic web technologies, especially ontologies, have been used for many years in building autonomic knowledge based systems in Artiﬁcial Intelligence. There is a pressing need for a standard ontology to enable technology agnostic, autonomic control in telecommunication networks. Network clients need to describe the resource they require, while resource providers need to describe the resource they can provide. With semantic technologies, the data inside complex hybrid networks can be treated as a distributed knowledge graph, where an SQL-like language – SPARQL is ready to search, locate, and conﬁgure a node or link of the network. The goal of this thesis is two-fold. The ﬁrst goal is to build a formal, machine interpretable information model for the current heterogeneous networks. Thus, we propose an ontology, describing resources inside the hybrid telecommunication networks with different technology domains. This ontology follows the Device-Interface-Link pattern, which we identiﬁed during the modelling process for networks within different technology domains. The second goal is to develop a system that can use this ontology to build a knowledge base automatically and enable autonomic reasoning over it. We develop a Semantic Enabled Autonomic management system of software deﬁned NETworks (SEANET), a lightweight, plug-and-play, technology-independent solution for knowledge-based autonomic network management that uses the proposed ontology. SEANET abstracts details of network management into a formally deﬁned knowledge graph augmented by inference rules. SEANET’s architecture consists of three components: a knowledge base generator, a SPARQL engine, and an open API. With the open API developed, SEANET enables users without knowledge of Semantic Web or telecommunication networks to develop semantic-intelligent applications on their production networks. Use cases of the proposed ontology and system are demonstrated in the thesis, ranging from network management task and social applications.