Optimizacija svojstava usluge primjenom inteligentnih agenata

scientific master's theses

• Author: Denis Sablić
• Mentor: Dragan Jevtić (mentor)
• Committee member: Marijan Kunštić (predsjednik povjerenstva)
• Committee member: Dragan Jevtić (član povjerenstva)
• Committee member: Antun Carić (član povjerenstva)
• Committee member: Vjekoslav Sinković (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; (Department of Telecommunications); Zagreb
• Defense date and country: 2005-06-08, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Electrical Engineering; Telecommunications and Informatics
• Universal decimal classification (UDC): 621.3 - Electrical engineering

Abstract

Rad obrađuje problematiku uvođenja inteligentnog agenta u općeniti sustav posluživanja, sa zadaćom usmjeravanja poziva, odnosno poslova poslužiteljima. Poseban naglasak stavljen je na istraživanje ponašanja agenta temeljenog na učenju ojačanjem. Konkretno, riječ je o metodi Q-učenja, koja se pokazala posebno pogodnom i interesantnom u sustavima koji zahtijevaju brzu prilagodbu. Razvijeno je nekoliko nadgradnji osnovnog modela agenta, s ciljem poboljšanja performansi u dinamičnoj okolini, kakvu predstavlja sustav posluživanja. Uz središnji model inteligentnog agenta, prikazani su i neki tradicionalni modeli temeljeni na često korištenim algoritmima za usmjeravanje. Na temelju modela sustava posluživanja, razvijen je simulator sustava posluživanja, pomoću kojeg su dobiveni usporedni rezultati ponašanja različitih agenata u identičnoj okolini, s ciljem isticanja prednosti i nedostataka uvođenja dodatne inteligencije u sustav posluživanja, to jest distribuciju poziva odnosno poslova.

Abstract (english)

This paper covers the problem of an intelligent agent in a general purpose service system, acting as a call transfer or job distribution mediator. The focus is on the behaviour of a reinforcement learning based agent. More specifically, the agent relies on Q-learning method, which has proven to be especially interesting and well suited for systems that require fast adaptation. A couple of additions to the basic Q-learning agent model were developed, in order to improve the agent's performance in a dynamic environment, such as service system. Besides the central result of this work – the intelligent agent, some of the more traditional agent models were presented. These models corespond to the well known and widely used call and job distribution algorithms. Based on a model of service system, a simulator was developed to produce results of various agent models behaviour in the identical environment. These results illustrate the advantages and disadvantages which come with the introduction of intelligence in a service system, or more precisely distribution of calls and jobs.

• Keywords: Inteligentno; agent; poziv; usmjeravanje; ojačanje; učenje; Q-učenje; telekomunikacije; optimalno
• Keywords (english): Intelligent; agent; call; transfer; reinforcement; learning; Q-learning; telecommunications; optimal
• Language: croatian
• URN:NBN: urn:nbn:hr:168:104477
• Study programme: Title: Postgraduate master programme in electrical engineering; Study programme type: university; Study level: postgraduate; Academic / professional title: Magistar znanosti elektrotehnike (Magistar znanosti elektrotehnike)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2019-07-02 09:05:39