In the telecommunication market, the analysis of the mobile network is essential to get detailed information about the behaviour of the subscribers. In this work, a network analyser tool is discussed, which inspects the network traffic and provides further information regarding the users.
This thesis is looking for a sufficient possibility to increase the performance of the current data processing method regarding the network traffic analyser application. The detailed workflow and the description of the analyser tool are part of this study, where the focus is especially on the post-processing segment of the actual workflow. The post-processing phase contains the evaluation of the results, where a lot of queries can be executed on the database.
As it is discussed later, the slowness of the actual post-processing process is originated from the query execution time, when usually a large dataset is queried. In this work a solution is proposed, which can improve the efficiency, while the possibilities of the modification of the current workflow are discussed.
Theoretically, complex event processing (CEP) technology is able to process and aggregate information in real-time even from multiple data sources. In this thesis, CEP is analysed to decide whether this emerging technology is able to increase the performance.
CEP is proposed as an excellent technology to improve the data processing workflow by implementing an application, which aggregates information in nearly real-time. The implemented software is based on Microsoft StreamInsight, Microsoft's platform for developing CEP-based applications. The architecture of Microsoft StreamInsight platform, the basic development possibilities and constraints are also discussed in this study.
According to the analysis, modification of the current workflow is proposed. The tests and the evaluation of the results are mainly based on the query execution time measured in both the original and the modified workflows in our test environment, which simulates nearly the real situation. The results show that, although this new technology has wider implications and it is applicable in different areas, it is also able to solve the described problem. Thanks to the pre-aggregation, the size of the database is drastically decreased, and the reduction of the required query execution time is achieved.