Extraction of Contact Lilst From CDR

Abstract

Modern Technology has revolutionized the way we communicate. The most widely used communication mechanism of modern times is the cellular networks. The cellphones used are supported with multiple services which include a way to save your contacts in your phone. But in-case your phone suffers any tragedy for example gets lost, you can be in trouble if you need to contact someone. Database queries and techniques generally used by telecom companies to retrieve ones contacts are not efficient due to large data. Analysis performed on 25 GB call details record takes more than 75 minutes to extract any user’s contacts from database. The project is to focus on the user's personal data and CDR, and develop an application which can discover contact list of a particular subscriber, given the CDR for a certain period of time. In this project we are employing modern algorithms from the areas of data-mining and their parallel implementations to devise solutions which can do the job of extracting contact-lists efficiently. Efficiency and high-performance is a key goal in this project because only then the said application will be a solution that can be acceptable to the big telecommunications service providers. The software is intended to generate contact list from CDR (Call Detail Record). The works focuses on databases searching mechanism. To perform efficient searching, Searching is done on the bases of parameters which are faster than others. Clustered indexes are defined on the database as to save data in order that it’s easier to retrieve. The whole data is divided into different file groups so that while searching instead of traversing whole data only limited portion of data is scanned. Normalization and de-normalization operations are also performed on tables to improve performance. After performing all these operation on database, Parallel implementation algorithm is developed to access database simultaneously by making several connections in parallel to database. These techniques improved performance in a way that the time taken to extract any user’s contact on 25 GB data is reduced to average 2 minutes.