DECISION SUPPORT SYSTEMS: SMART WORKFLOWS : In the context of telecommunication networks
Year: 2013
Type: dissertation
Abstract: This thesis, conducted at Ericsson Software Research, investigates how workflows can be made to work more dynamically. The thesis further aims at recommending how such workflows can be applied in the context of network operation centres (NOC) - the steps needed for automation of trouble ticket handling. An agent-based solution is considered, where each agent implements a workflow, ruled by conventional methods, but the agents communicate in a way allowing new agents to appear as well as allowing disconnections and reconnections without disturbing other parts of the system. Furthermore, different artificial intelligence algorithms most suited for automatic information gathering are investigated, mainly for application in troubleshooting environments during telecommunication network management. The intention of this is to provide the workflows with the tools needed for automation of routine tasks – in a way users can easily understand and follow. The perception here is that Naïve Bayesian networks are preferred, as they are easy to train, scales good and inference from a Naïve net is easy to understand in an intuitive way. Further, ZeroMQ is recommended when designing a decoupled workflow system.
Author: Daniel Schiess
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W1597562199