Mining frequent patterns to identify vertical handover parameters in cellular networks

Abstract

The heterogeneous environment of cellular networks imposed the need for finding different vertical handover (VHO) mechanisms. VHO algorithms are presented in literature with different input parameters for various services related to different quality of service (QoS) classes (conversational, streaming, interactive and background), such as: cost, available bandwidth (ABW), MT speed, received signal strength (RSS), etc. These parameters are suggested by researchers or experts based on simulation analysis. Furthermore, the concept of identifying the effective parameters for VHO has not been widely studied. For this purpose, we proposed a new approach to extract the effective parameters and their priorities by using data mining (DM). Since voice call service is one of the most sensitive delay services, we collected real log data from servers of two mobile telecom companies in Lebanon, Touch and Alfa for Global System for Mobile Communications (GSM) and Universal Mobile Telecommunications System (UMTS) networks. After preprocessing and discretizing the data, the log file was then summarized and the frequent patterns (FP) were extracted. In this research, two new algorithms were designed. The first, is the Data Summarization Algorithm (DaSA) to summarize the preprocessed data. The second, is the Frequent Pattern Calculation Algorithm (FPCA) to find the frequency of the simultaneous occurrence of the obtained parameters.