Dynamic Quota Calculation System (DQCS): Pricing and Quota Allocation of Telecom Customers via Data Mining Approaches

Dynamic Quota Calculation System (DQCS): Pricing and Quota Allocation of Telecom Customers via Data Mining Approaches

Ulaş Çelenk (Innova IT Solutions Inc., Turkey), Duygu Çelik Ertuğrul (Eastern Mediterranean University, North Cyprus), Metin Zontul (Istanbul Aydin University, Turkey), Atilla Elçi (Hasan Kalyoncu University, Turkey) and Osman Nuri Uçan (Istanbul Kemerburgaz University, Turkey)
DOI: 10.4018/978-1-5225-5384-7.ch019
OnDemand PDF Download:
No Current Special Offers


One of the most important IT sectors that requires big data management is mobile data communication systems (MDCS) of GSM companies. In the charging mechanism of current MDCS, a subscriber “surfs” on the internet that creates data traffic and a counter subtracts the amount of data used by the user from the subscriber's quota. In other words, instant constant quota values are assigned to subscribers without concern for their previous amount of internet usage in current MDCS. Moreover, constant quota values cause constant charge calls in control traffic that are repeated for all new quota requests. Thus, performance degradation occurs because of the repetition of quota request calls and allocations. In this chapter, a dynamic quota calculation system (DQCS) is proposed for dynamic quota allocations and charging in MDCS using data mining approaches as two cascaded blocks. The first block is self-organizing map (SOM) clustering based on a sliding window (SW) methodology followed by the second block, which is the markov chain (MC); the overall system is denoted as “SOM/SW and MC.”
Chapter Preview

Literature Background

Quota-based charging of mobile users was investigated in several studies. Abidogun & Omlin (2004) presented a Self-Organizing Map (SOM) model for outlier detection in call data from subscribers, over a period of time in a mobile telecommunication network so that suspicious call behavior could be isolated in order to identify abnormal call patterns from subscribers. The researchers applied the SOM model to the unsupervised classification of call data for prepaid service subscribers from a real mobile telecommunication network. They indicated that the ideas presented in their study might be used for clustering call patterns in order to label them as normal or abnormal.

Next similar study by Lehtimäki & Raivio (2005) presented an analysis process based on SOM to visualize MDCS network performance data. They applied SOM in the analysis of 3G network performance, including advanced network monitoring and cell grouping. After outlining the overall SOM based analysis process for MDCS performance data, they demonstrated the use of the analysis process in two problem scenarios in which the capacity problems in the signaling and traffic channels were analyzed.

In the study of Ozianyi & Ventura (2005), an approach was proposed for charging mobile Internet use based on subscriber profiles. The researchers attempted to determine the Internet use habits of subscribers by considering their economic and social status, and they established a model for charging subscribers temporarily by considering usable Internet bandwidth and service quality as parameters.

Multanen, Raivio, & Lehtimäki (2006) used a SOM-based model to develop a method for exploring the data of an entire MDCS network. They preferred SOM because it has the ability of reducing highly visual data. In their study, SOM was used both in clustering and in visualization.

Key Terms in this Chapter

Self-Organizing Map: The self-organizing map (SOM) is a well-known neural network and certainly one of the most popular unsupervised learning algorithm. Since its invention by Finnish Professor Teuvo Kohonen in the early 1980s, more than 4000 research articles have been published on the algorithm, its conception, and uses. The SOM mapping is preserving, namely the most similar two data samples are in the input space, and the closer they will appear together on the final displayed map. This allows the user to identify clusters such as large sets of a specific type of input pattern.

LTE: Long term evolution (LTE) is a 4G wireless broadband technology developed by the third-generation partnership project (3GPP), an industry trade group.

Session: Logical connection between parties involved in a packet-switched-based communication. This term is used for IP connections rather than the term “call” that is normally used for a connection over conventional (circuit switched) systems.

Clustering: Clustering is a grouping of a particular set of objects based on their characteristic, that is, aggregating them according to their similarities. Regarding to data mining, this methodology partitions the data implementing a specific join algorithm, most suitable for the desired information analysis.

GSM: Global system for mobile communication is a digital mobile telephony system that is widely used in Europe and other parts of the world. GSM uses a variation of time division multiple access (TDMA) and is the most widely used of the three digital wireless telephony technologies (TDMA, GSM, and CDMA). GSM digitizes and compresses data, then sends it down a channel with two other streams of user data, each in its own time slot. It operates at either the 900 MHz or 1800 MHz frequency band.

Charging: A function whereby information related to a chargeable event is formatted and transferred in order to make it possible to determine usage for which the charged party may be subsequently billed.

Complete Chapter List

Search this Book: