The System Of Mining Weighted Association Rules For Alarm Data In Operations Supporting Network

Today, communication industry has developed rapidly, the competition of MobileCommunications Operator is intense, network maintenance is an important guarantee for operators.Operation supporting network is used to maintain and support the operators network. In order towork better, the network management of operation becomes more and more important, the faultmanagement is an important branch of network management,and a fault may cause a lot of alarmdata. The network maintenance persons’ artificial alarm data analysis can not meet the needs of theoperation and maintenance work. So the paper chose to use association rules algorithm of datamining to analyze the alarm data of the operation supporting network. The purpose is to extractvaluable information from the alarm, and use the information to find the root causes of the alarm.The principal activities of this paper is the establishment of Weighted Association RulesMining System (WARMS).The function of WARMS is mining the association rules of alarms ofOperation supporting network.The Weighted Association Rules Mining System (WARMS)uses the sliding time windowalgorithm and analytic hierarchy process (AHP) to preprocess the alarm data of the operationsupporting network. Sliding time window algorithm is used to convert the alarm data of theoperation supporting network into alarm transaction data. Analytic hierarchy process (ahp) usesattribute of the alarm data to determine the importance of the data between the two to distribute theweights for the alarm and alarm transaction data.Then,the preprocessed alarm transaction data are inserted into a weighted frequent pattern tree(WFP-Tree), the step includes the advantage of frequent pattern growth (FP-growth) and removesits shortcomings. Finally, using the weighted frequent patterns algorithm (WFP) to mine theassociation rules from the weighted frequent pattern tree. The algorithm absorbs the advantages ofAprior algorithm (pruning, connections, etc.) and dismisses the disadvatage ofAprior algorithm.