Performance Analytical and Monitoring System for SMS
SUMMARY
BACKGROUD
PAMSiS is a CDR analytical tool designed specifically as a support system for measuring performance of SMS Service. PAMSiS tool provides a set of functions that enables user not only to get service performance, but also to be able to do in-depth troubleshooting and analysis from CDR files using general SQL Query. For management level, PAMSiS has Business Intelligent Tools to provide graphical web-based summary to review KPI service. This document describes the key innovation of successfully loading large CDR data and performing near real-time KPI summary.
Figure 1: The SMS data flow
Figure 1 illustrates the basic flow of sending an SMS in GSM network, an SMS is sent from cell phone to Base Transmitter Station (BTS), and then the message is transmitted into BSC (Base Station Center). From BSC, the message is transmitted to Mobile Service Center (MSC), then to SMS Center (SMSC). From SMSC, the message then is transmitted to destination number using the same reversal path.
Each path (BSC, MSC, SMSC) generates CDR data and statistical log in order to ease users to do troubleshooting and view performance. SMSC provides service-level CDR, while MSC or BSC usually provides protocol-level CDR.
Most statistical data is displayed based on CDR log files. This is because CDR log files are much smaller than original CDR files, but CDR log files do not have detailed information of SMS traffic.
PAMSiS generates statistical data from CDR files. The system also holds all CDR data for a certain period of time. By using CDR as source for statistical data, users are able not only to retrieve performance summary of machine, but also to do troubleshooting directly to CDR data that have been loaded to database.
The key of PAMSiS performance in managing a very large amount of data lies on its architecture:
- The data loader uses a technique called Stream-pipe processing. By using this technique, system can avoid disk processing. All processes are processed inside memory, thus allowing system to avoid ten-thousand-times-slower disk access.
- The database is designed using data warehouse technology, and optimized to maintain very large data load. System is open to any database, but it is recommended using Column-oriented database. Our optimization method covers not only in designing logical database, but also in designing physical storage to balance system requirements and hardware profiles.
- PAMSiS is designed to have open-ended architecture. By having this architecture, our system is open to any forms of CDR data, either binary or ASCII. The system is also open to any forms of reports, as long as the reports can be generated from provided CDR.
BUSINESS VALUE
Revenue Enhancement
Focusing on making more profit for customers, PAMSiS provides analysis tools that can help management of company in determining a better company policy that can influence company revenue.
Accelerate Cash Flow and Reduce Organization Cost
PAMSiS can accelerate cash flow by providing information that can help management in providing more accurate SMS demand forecasting. In the other hand, organization can also reduce cost from mediation storage investment for their CDR.
Analytical Productivity
In the past, business users, analysts and knowledge workers had to spend most of their time in gathering data, and only less time left over to perform analysis. With PAMSiS, we are seeing less time in gathering data and much more time for analysis.
Fraud Reduction
PAMSiS have been used to detect fraudulent SMS usage
Better and Faster Decision
PAMSiS produces OLAP Business Intelligence Report from very huge SMS data at very fast time. This helps user in making decision better and faster.
Comparation Source
PAMSiS can be used as a second opinion against any system conducting SMS CDR. For example billing system and report for SMSC Vendor
Business Intellegent
The system provides near-real time summary in interactive graphical web-based user interface Dashboards. The system also enriches user experience to analyze SMS Performance KPI by using rich-colored-clickable graphic. There are no queries needed, only point and click to the chart, and the other graphs will change to the corresponding data.
Figure 2: PAMSiS System Architecture
Figure 2 describes basic architecture of PAMSiS. CDR data that are generated in timely fashion are grabbed using ftp transfer. The CDR data are then processed to create a new database-friendly format file (usually in the form of CSV files). The new file will be loaded into database using bulk load command (may be in the form of SQL loader for Oracle, Oracle external table, or a psql load script). The processes include removing duplicate records that sometimes exists in single CDR file.
The loaded database is called staging database. The staging database contains all CDR fields, and extra supporting fields to allocate the original CDR files, loading time, and some other information that might be necessary but not exists in CDR data. The staging database is very large and suitable for troubleshooting purpose.
There is another database schema called summary database. The summary database converts staging database into a data warehouse star-schema. The summary process only takes necessary fields to be inserted into Data Warehouse fact-table. The size of the summary database is usually about 5% of the staging database. This database is suitable for building service-level data summary that makes the management able to check the service-level performance of current system.
STREAM-PIPE PROCESSING
Since CDR files are processed once an hour, the size of memory modules can be reduced. For example, a 70 GB CDR files/day will require only 7 GB dedicated memory, assuming that there is 10% of total message at peak hour.
Number of processor also plays important role in this process. We do not want to put large amount of data in memory module for a long time. So a file being stored in memory should be processed instantly.
PAMSIS Business Intelligence
PAMSiS uses Business Intelligent Tools for its front-end view. The system provides near-real time summary in interactive graphical web-based user interface using BI Tools Dashboards.
BI Tools dashboard enriches user experience to analyze SMS Performance KPI by using rich-colored-clickable graphic. There are no queries needed, only point and click to the chart, and the other graph will change to the corresponding data.
The benefit of BI Tools Dashboard is that is very easy to use and eye catching. But the limitation is the dashboard only shows summary that have been defined at first. If the management wants another report, then the developer should create a new report page and flow using JSP scripting language and XML.
Another kind of reporting is using drill-down engine. It is Microsoft Excel Pivot-like reporting that allows user to analyze further detail of the data, create a necessary report on the fly, and export the report to Excel format. Users can drill down data
PAMSIS is suitable for Telecommunication industry that does not want to rely on machine vendors for performance evaluation, and gathers service summary not only from CDR statistics, but also directly from CDR files. Users will have detailed and summarized reporting through easy reporting engine and analytical tools that connect directly to a readable-optimized database.
Who has used this product?
One of Telecom Operator that has been using this product is Celcom (Malaysia) Bhd.
