Knowledge Graph Based Intelligent Service Discovery for Distributed Fog Computing

Abstract

To realize distributed fog computing, intelligent service discovery solutions should be designed to efficiently detect appropriate services with good performance and high Quality of Service (QoS) in different and varying environments. The goal is to come up with a Knowledge Graph Based Intelligent Service Discovery solution that will be useful for enhancing semantic service discovery, QoS-based service selection and resource allocation in distributed fog computing environments. The methodology used for developing such a solution entails request preprocessing, ontology mapping, construction of knowledge graph, semantic reasoning, graph traversing, QoS evaluation, intelligent service ranking and dynamic resource allocation. New ideas introduced involve using the knowledge graph semantics in combination with context-aware reasoning and graph-based service ranking for efficient service discovery that goes beyond the traditional keyword and ontology-based methods. For performance evaluation of the proposed solution, QWS Dataset Version 2.0 was used where the data set contains 2,507 web services that have been characterized using their Response Time, Availability, Throughput, Successability, Reliability, Compliance, Best Practices, Latency and Documentation features. The results of experiments included service discovery time of 31 ms, 99% service success rate, 98% QoS satisfaction, throughput of 4,218 requests/s and query execution time of 17 ms.