Diameter in IMS: Why real world IP service testing is here to stay

by Tom Maufer, NGN/IMS Forum

The IMS Forum has been, since 2007, hosting 2-3 Plugfest events per year at the University of New Hampshire Interoperability Lab (UNH-IOL). The Forum just finished their eighth NGN IMS Forum Plugfest in October, and the degree of difficulty has been steadily increasing. Initially, the challenges were around now-passé interoperability and functional testing requirements such as SIP interoperability, specifically the IMS flavor of SIP. Yes, this is still SIP, it's just that it has some specific types of proxies (known as Call Session Control Function (CSCFs)) and each does specific jobs to facilitate call flows. The Proxy Call Session Control Function (P-CSCF) is the first-hop proxy and all SIP calls go through it. The Serving Call Session Control Function (S-CSCF) is where the User Entity's REGISTER messages are sent, and the Interrogating Call Session Control Function (I-CSCF) facilitates inter-domain SIP traffic.

However, as operator voice services require more than just a signaling protocol, larger issues loom. There are operator databases of users, billing information associated with those users, call detail records requiring storage for each call, etc. These functions are collectively known as Authentication, Authorization and Accounting (AAA) by the standards bodies (e.g., IETF and 3GPP). The protocol that has emerged to do these AAA tasks is Diameter. The IMS Forum Plugfest VIII was about integrating real Diameter-based AAA services into the SIP-based-CSCF infrastructure so that there could be realistic Diameter exchanges to facilitate realistic SIP call flows.

For instance, when a UE registers, there is a Diameter exchange between the S-CSCF and the HSS using the "Cx" interface. This verifies that the user has an account in the system and that the account is valid, etc. When the UE initiates a call, the -CSCF uses the "Rf" interface to communicate with the Application Server (AS) to establish whether the user is authorized to use the application, whether their account has enough credit to pay for the usage, etc. For instance, perhaps the SIP session is carrying an SMS message, which is billed differently than phone calls.

As you can see, the Diameter protocol is an important integration point for carrier tracking of their customers' usage and as the basis for monthly billing. For IMS to be useful, it needs to demonstrate an end-to-end, multi-faceted solution that involves all the key subsystems: CSCF for call processing, HSS for the user database, Application Servers, and real charging interfaces so the network feels like a carrier network, where calls and usage are tracked as if some customer would ultimately need to be billed.

Carriers need to test networks and real IP services--not just components--in just these kind of real environments, where there are multiple interdependent systems communicating via open standards. What the hands-on NGN IMS Forum Plugfests demonstrate is that it's difficult (but not impossible) to make the technology work. The Plugfests prove that for testing an IP Service like IMS, you need to test the network as a service, not as a collection of isolated components. The components definitely behave differently when they are interacting dynamically with other components, compared to when they are being tested in a standalone mode. The more realistic the test environment is to a customer's network, the more applicable the results are, and the more indicative of the likely ability of the service to perform as required once it's rolled out in the real world.

 Tom Maufer is the Director of Solutions Architecture at Mu Dynamics, which serves as the Vice Chair of the NGN IMS Forum's Billing/OSS & Security Working Group.

Suggested Articles

lphabet and Google CEO Sundar Pichai announced on Monday that Google plans to invest around $10 billion in India over the next five to seven years.

Hewlett Packard Enterprise (HPE) has juiced its SD-WAN platform by buying Silver Peak for $925 million.

Comcast's investment in the development of artificial intelligence and machine learning paid off in spades during the coronavirus pandemic.