VoIP use case

Most billing systems (BSS) consolidate user data only once a month for the issuance of their bill. In this use case, the customer wants to access his voice / voice over IP (VoIP) network billing tickets (CDR: https://en.wikipedia.org/wiki/Charging_data_record) in real time. The two main needs are:

  • consult, classify, analyze in real time, calls made and received by the network,
  • Avoid bill shock : detection of fraud and in particular the transmission of calls to premium-rate numbers.

Explore and visualize data

In order to consult your data, it is useless to be an expert in programming language. Thanks to the graphical interface of Superset, you can, in a few clicks build your report (slice in the Superset terminology).

In this case, you see a graph that counts the number of CDR ingested per 24h. You can adapt the duration that interests you but also the granularity of the graph (number of points).

Analyze your data

By using all the columns available in the database and associating them with adapted graphical views, you can create reports that are relevant to your trades.

  • View based on the geographical capabilities of the system.
  • View based on graphics capabilities such as the classic pie chart.

Building dashboards

After building a set of relevant views for a need, you can assemble all these views into a dashboard that can be positioned on a continuously refreshed screen. In this case, it is a dashboard for investigating security issues on the voice network.

Ingest layer

This use case required the development of a specific data ingestion software. The data from the voice equipment is in a binary format that requires the writing of a serializer / deserializer. The ingestion layer also provides additional services such as the enrichment of the records with geolocation data of the calling number, the identification of the type of number (mobile, fixed, over-taxed, …) and finally the cost of the call.

Special attention is given to the fields that define time. Our entire system must be consistent on the clock used and on the time information provided in the records.

The software module is seen as a Kafka producer by the message broker layer. It is deployed on as many VM’s or containers as needed to process CDR’s flows.

Software design

Based on the reference model that we propose for the PASS solution “Pragma Analytics Software Suite”, here is the model that Pragma Innovation has built to meet these specifications.