2013-09-092008-07-152013-09-092007-09-052007-07-152008-07-10Bihina Bella, M 2007, A fraud management system architecture for next-generation networks, MSc dissertation, University of Pretoria, Pretoria, viewed yymmdd < http://upetd.up.ac.za/thesis/available/etd-07102008-145544/ >http://hdl.handle.net/2263/30984Dissertation (MSc (Computer Science))--University of Pretoria, 2007.The telecommunications industry is shifting to a new type of network generally referred to as next-generation networks (NGNs). The concept of NGNs implies the convergence of voice, video and data networks onto the same infrastructure. It enables the offering of a new class of services that has the potential to increase revenue for operators and enhance customers’ experience. Unfortunately NGNs are highly likely to favour the rise of telecommunications fraud. Due to some of their key characteristics such as being based on the Internet Protocol, NGNs create new challenges for effective fraud detection. Besides, as they enable the provision of innovative services, NGNs may also give rise to new fraud scenarios that cannot be addressed by existing fraud management systems (FMSs) as these systems are application-dependent. They heavily depend on the service types and their underlying network platform. Consequently, FMSs need to be revised to effectively tackle these emerging issues. This thesis presents an architecture for a next-generation network fraud management system (NGN FMS) specifically designed to satisfy the requirements of flexibility and application-independency that cannot be met by traditional FMSs. The architecture has a thorough multi-stage detection process that analyses billing records in Internet Protocol Detail Record (IPDR) format – an emerging IP-based billing standard – for signs of fraud. In addition to its high level of flexibility, the proposed architecture has the benefit of being largely scalable and is able to help uncover new fraud types. This is achieved through the added combination of intrusion detection and neural network technology in the architecture design. Neural networks are implemented in the form of Self-Organising Maps (SOMs) in one component of the FMS, appropriately named the SOM Analyser. A prototype of the SOM Analyser has been implemented and is discussed in the thesis. Copyright 2007, University of Pretoria. All rights reserved. The copyright in this work vests in the University of Pretoria. No part of this work may be reproduced or transmitted in any form or by any means, without the prior written permission of the University of Pretoria. Please cite as follows: Bihina Bella, MA 2007, A fraud management system architecture for next-generation networks, MSc dissertation, University of Pretoria, Pretoria, viewed yymmdd < http://upetd.up.ac.za/thesis/available/etd-007102008-145544 / > E853/agen© 2007, University of Pretoria. All rights reserved. The copyright in this work vests in the University of Pretoria. No part of this work may be reproduced or transmitted in any form or by any means, without the prior written permission of the University of Pretoria.UCTDSelf-organising mapProtocol detail recordIpdrNext-generation networkNgnIpInternet protocolFmsFraud management systemTelecommunications fraudBilling systemSomDissertationhttp://upetd.up.ac.za/thesis/available/etd-07102008-145544/