M. Klemettinen. A Knowledge Discovery Methodology for Telecommunication Network Alarm Data. PhD thesis, University of Helsinky (Finland), 1999.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... the complexity and built in redundancy often results in a large number of alarm events being raised and cascaded to the Element Controller (the manager) The behavior of the alarms is so complex it appears non deterministic [13] making it very difficult to isolate the true cause of the fault [14]. Failures in the network are unavoidable but quick detection and identification of the faults responsible is essential to ensure robustness. To this end the ability to correlate alarm event messages becomes very important [15] The major telecommunication equipment manufacturers deal with event ....

M. Klemettinen, "A knowledge discovery methodology for telecommunication network alarm databases", Ph.D. Thesis, University of Helsinki, Finland, 1999.

....be it by means of correlation, filtering, patching of flawed IDS signatures, blocking of attackers at the firewall, or something else. In the world of telecommunication networks, Klemettinen uses association rules and episode rules to support the development of alarm correlation systems [31]. Hellerstein and Ma pursue the same goal by means of visualization, periodicity analysis, and m patterns (a variant of association rules requiring mutual implication) 26] Garofalakis and Rastogi investigate bounded error lossy compression of network management events [18] Note that a priori, ....

....on the occurrence of other alarms. For example, an episode rule might state that in 50 percent of all cases, an Authentication Failure alarm is followed within 30 seconds by a Guest Login alarm. In network management, researchers have successfully used episode rules in a framework like ours [31]. Therefore, episode rules are a natural candidate for the data mining step of Figure 1. In this section, we report our experience with this approach. In addition, we summarize the insights we gained into the nature of intrusion detection alarms. 4.1 Definitions To formally define episode ....

[Article contains additional citation context not shown here]

M. Klemettinen. A Knowledge Discovery Methodology for Telecommunication Network Alarm Data. PhD thesis, University of Helsinky (Finland), 1999.

....can relate to the rules that are expressed about their datasets. Association rules can be representations of simple causal relationships within the domain. Thus, rule based approaches are often used to understand the relationships within the dataset to the best of the domain expert s knowledge [42]. It is essential that the rule based algorithm used, is as e#cient and as fast as possible, while providing the user with the minimal set of the most interesting rules. As mentioned by Das et al. in [16] a problem with finding association rules in sequences is that as the number of items in the ....

....many of those describe above. This research was inspired by several data mining techniques: Normalizing, discretizing, and clustering time series, such as work done by Goldin [26] Recognizing sequences of events, or episodes, such as work done by Mannila [51, 52, 53] and Klemettine [42], Finding representatitive association rules using closures, such as work done by Kryszkiewicz [44, 45] Pasquier [66] and Saquer [77] Using user specified constraints to drive the analysis, such as work done by Srikant [80] Discovering associations (rules) in time series, such as ....

[Article contains additional citation context not shown here]

Klemettinen, M., A Knowledge Discovery Methodology for Telecommunication Alarm Network Databases , PhD Thesis, University of Helsinki, Finland, 1999.

....life span would greatly assist in managing and assessing maintenance strategies. Fault management is an important but difficult area of telecommunications network management. Networks produce large amounts of alarm information that must be analysed and interrupted before the faults can be located [23]. As has been stated earlier alarm correlation is the central technique in fault identification [18] The instance of a fault can cause numerous alarm events to be raised from an individual network element (NE) this means that the alarms are often interrelated. Also a fault may trigger numerous ....

M. Klemettinen, "A Knowledge Discovery Methodology for Telecommunication Network Alarm Databases", PhD Thesis, University of Helsinki, Finland, 1999

....priori versus a posteriori filtering of association rules (extended abstract) Bart Goethals and Jan Van den Bussche Limburgs Universitair Centrum fgoethals; vdbussg luc.ac. be 1 Introduction The concept of inductive database, proposed by Mannila [8, 11], is a beautiful formalization of the interactive mining process. In the concrete setting of association rule mining, an inductive database provides virtual tables containing virtually all itemsets and rules over the data. The user does not care how these inductive tables are implemented; for him, ....

M. Klemettinen. A Knowledge Discovery Methodology for Telecommunication Network Alarm Databases. PhD thesis, University of Helsinki, 1999.

....versus a posteriori ltering of association rules (extended abstract) Bart Goethals Limburgs Universitair Centrum bart.goethals luc.ac.be Jan Van den Bussche Limburgs Universitair Centrum jan.vandenbussche luc.ac. be 1 Introduction The concept of inductive database, proposed by Mannila [8, 11], is a beautiful formalization of the interactive mining process. In the concrete setting of association rule mining [1] an inductive database provides virtual tables containing virtually all itemsets and rules over the data. The user does not care how these inductive tables are implemented; for ....

M. Klemettinen. A Knowledge Discovery Methodology for Telecommunication Network Alarm Databases. PhD thesis, University of Helsinki, 1999.

No context found.

M. Klemettinen. A Knowledge Discovery Methodology for Telecommunication Network Alarm Data. PhD thesis, University of Helsinky (Finland), 1999.

No context found.

M. Klemettinen, "A knowledge discovery methodology for telecommunication network alarm databases", Ph.D. Thesis, University of Helsinki, Finland, 1999.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC