Rostelecom Urals (Uralsvyazinform of Utel) (Microsoft Business Intelligence)

Situation

Now in the conditions of fierce competition telecom operators are forced to reduce rates. At this ARPU constantly falls, and operators should look for new ways of increase in profit. For solution of the problem of decline in yield there are two main ways: develop new highly profitable services, bringing to the market additional services before competitors and minimize the losses.

As told the leading specialist of the sector of system management of guaranteeing revenues of JSC Uralsvyazinform of Yuli Lebedev, "losses on different sections of a production process and sale of telecommunication services annually develop in the general shortage of money. On different expert evaluations, losses in revenues of operators average from 3–4% to 15%". A share of losses from a fraud from partners in connection with insufficient integration of systems, inefficient businesses processes / systems and errors of tariffing reach nearly a third at large telecom operators which Uralsvyazinform is. There is a question: how timely to detect losses, to determine their volume and the reasons and not to allow their emergence in the future? Traditionally this problem was solved by carrying out the analysis by own specialists of operator or third-party experts. Both approaches have certain shortcomings. In the first case the quality of researches is much lower, and in the second — too project high price. There is also the third method of solution: automate processes of detection of losses.

Solution

Heads of JSC Uralsvyazinform made the decision to automate processes of detection of losses. Strict requirements to a reference platform were made: it should provide an opportunity for work with large volumes of data, provide mechanisms of visualization of analysis results and at the same time it is easy to be integrated with the available information environment.

After the analysis of offers the choice fell on the Fraud detection system and loss analysis of revenues of the operator (SOMAP) of IBS company. The architecture of SOMAP is constructed on the integrated and adaptable solutions of Microsoft which allow to simplify development, implementation and management. The core of a system works based on Microsoft SQL Server DBMS 2005.

According to Yulia Lebedeva, for successful detection of losses of income a system should not only control consumption of services by subscribers and reveal cases of unauthorized use of network resources, but also in time detect loss of information about perfect calls from a switching equipment, distortion (intentional or accidental) it in the platform of interface and also to check correctness of tariffing. At the same time a system should be integrated with many external information sources. Microsoft developed on a product line, SOMAP has the developed abilities to integrate. After obtaining information on the rendered services a system performs a number of checks, beginning from data validation on reliability and consistency before creation of a profile of behavior of subscribers with the subsequent identification of signs of fraudulent activity.

Result of work of SOMAP is a set of the reports allowing to reveal problems which can lead to losses of income, and providing a workplace of the analyst with the developed means of data analysis. Results

Implementation of SOMAP allowed Uralsvyazinform:

* independently to control losses of income, to study and analyze their reasons;

• increase efficiency and accuracy of control;
• increase the speed of identification and elimination of problems;
• reduce losses of income.

At the moment SOMAP is implemented on city nodes of telecommunication of the regional and district centers, the operator carries out territorial replication of a system and expansion of its functionality. Principal users of a system are workers of a security service of the enterprise — both branches, and general directorate. Further information provided to SOMAP will be used also by other divisions: service of quality management, commercial and technical blocks, department of operation.

Technical details

Version of a product

SQL Server 2005 Enterprise Edition x64

Server platform:

Windows Server 2003 x64 Portal server: 4 CPU Xeon Dual Core, 14 GB of RAM Database server: 16 CPU Xeon Dual Core, 87 GB of RAM Application server: 4 CPU Xeon Dual Core, 14 GB of RAM Optical network of data storage using two IBM DS4800 disk arrays with a general capacity of 15.47 of Tb

Amount of data:

Amount of data — more than 15 Tb The biggest table — more than 3.5 billion lines The used technologies within SQL Server 2005 SQL Server Analysis Services Data MiningSQL Server Reporting Services SQLCLR и SQLXML Sectioning of tables Soft-NUMA

Loading indicators:

Number of users — more than 200 Up to 1000 analytical requests a day