Introduction to Mediation

By | April 9, 2010

Mediation devices collect usage data in the form of xDRs (CDRs, IPDRs and EDRs etc.) from several data sources and deliver this data to specified locations in a specified format.

The devices (switches, routers, IP PDUs) and applications (IN etc.) in a telecommunication infrastructure report their usage statistics for different purposes. The main purpose is billing, so, every mediation device generally interfaces with one or more billing (or charging) systems. We may also see other integrations with Fraud Management Systems (FMS), Revenue Assurance Systems (RA), IN systems, Interconnect systems etc.

Different types of devices on the network output different types of CDRs in different formats. They also deliver those via different protocols (ftp, sftp, snmp, xfer, ftamip, s12ftp, ftmxot, netflow, odbc to name a few). Most of the CDRs are in ASCII format and some of them are in binary format (like ASN.1 and Radius)

Usage dependent applications on the other hand, want to have the usage data in a specific format. For example a billing system’s specs may say that the input data must be in ASCII format. It may also dictate the orientation of the expected file (i.e. columns within it)

Mediation devices does this protocol and file conversion. They deal with the polling logic, (network errors, retries, bad data etc.) and prepare an error-free world for the usage based applications. They detect the file errors and save them in a place where they are further analyzed and corrected. They do CDR normalization (adding prefixes, suffixes etc). They aggregate partial CDRs. They do duplication checks. You can run some rules on them, such as “discard the CDR if the usage duration is less than 5 seconds.”

Mediation devices act as a “mediator” between the systems that deal with the usage data. They act as a single point of contact. Mediation process reduces the load of other applications by taking over the collection, aggregation, normalization and conversion tasks. Considering the number of data sources and data pollers, it may require multiple load-balanced servers and dedicated organizational departments.