As founder and chief proponent of the IP exchange (IPX), the GSM Association (GSMA) represents the interests of 800 mobile operators and 200 industry-related companies. IPX provides a commercial and technical framework for IP interconnection between any type of communications service provider – including fixed network operators (FNOs), mobile network operators (MNOs), application service providers (ASPs) and internet service providers (ISPs) – with the appropriate standard of endto- end quality and security guaranteed. Key components of the IPX include: traffic with a managed quality of service (QoS) at various levels of performance; payments which are identified and settled between operators; bilateral or multilateral modes of interconnection which can be selected for different types of traffic; protection of IP traffic – IPX is a private IP network separate from the public internet; and services which can use less demanding QoS when premium quality is not required.
IPX will be operated under competitive market conditions, to allow carriers to differentiate on price and service. So it is up to operators and carriers to thrash out commercial terms. “The GSMA has gone as far as we can in providing the tools for the market to go away and create their own products and commercial and technical relationships with each other,” says Dan Warren, senior director of technology at the GSMA. “It is up to the market to use those tools as they see fit.”
The one-to-many model
Under IPX, a communications service provider will be able to sign one contract with an IPX provider to gain IP interconnectivity with any other service provider connected to an IPX. This “hubbing” approach can be traced back to the GPRS Roaming Exchange (GRX), established by the GSMA in 2000 to transport GPRS data from roaming subscribers back to a home network. GRX specifications were based on the use of simple domain name server (DNS) queries to allow the visited network to identify and tunnel traffic back to the home network over a private IP connection. Previously, interconnect for voice and SMS had been conducted almost entirely on a bilateral basis.
“There was an awful lot of redundancy in the process,” says Warren. “Each roaming agreement started from scratch and, in many cases, although all looked very similar, they would be subtly different. With 800 GSM operators in the world, this meant that in theory all mobile operators needed around 799 TDM interconnect agreements that had to be maintained individually.” The GSMA’s solution was to introduce a common framework that meant a mobile network operator could sign a contract with a GRX provider, and then exchange traffic with all other MNOs connected to a GRX. Today, the GRX cloud comprises multiple providers peered via three neutral facilities in Asia, Europe and north America, and serves 400 MNOs worldwide.
The aim with IPX is to make everything achieved with GRX applicable to interconnect in the IP world. However, there are expected to be more peering points for IPX than the three used for GRX, to achieve a higher QoS. IPX extends the GRX architecture by interconnecting any IP service between any pair of operators, and allows the addition of end-to-end QoS with a variety of charging models over and above that of volume. These are: transport-only – the provision of a large pipe from one operator to another, which is not service aware, so that any protocol can be transported, but with guaranteed QoS; bilateral service transit, with QoS-based transport and cascading interconnect payments; and multilateral service hub – so a single agreement between the service provider and the IPX provider delivers QoS transport and cascading interconnect payments between a number of interconnect partners.
“Functionally, the latter two are pretty much the same,” says Warren. “Both allow the carrier to offer additional services, such as interworking functionality in the bearer and signalling planes, and the support of QoS end-to-end at a much more granular level.” Carriers connected to the IPX under the bilateral/multilateral model are able to “see” the services and applications passing across their networks, allowing them to apply functions to improve performance, such as different classes of service (CoS), traffic shaping, transcoding and signalling interworking.
Warren believes IPX providers will be able to sell these functions on to the operator community as value-adds. “Larger carriers using IPX will likely adopt transport-only because they will have sufficient scale to offer QoS and interworking functions at the edge of their own networks, and smaller operators without the capex to do so will ‘outsource’ these functions to an IPX provider under the bilateral/multilateral model.”
Aligning industry efforts
The first IPX specifications were published in 2006, prompting a wave of packet voice service pilots during 2007/8. Trials have included 22 service providers (fixed and mobile), 14 IPX providers and eight SIP-I platforms (softswitches), with all principal IPX features successfully demonstrated in 12 end-to-end service trials.
But carriers and equipment vendors initially struggled with IPX and the general feedback was that it had been over-specified to the point that it was difficult to implement – either due to technical complexity or the cost of developing bespoke implementations. So the specifications were revised in 2008 to allow carriers to implement IPX services without the need for R&D from equipment vendors. “There was no point in a technical description of something that carriers couldn’t go away and build, let alone sell,” says Warren.
Fragmented industry development and confusion surrounding the commercial model have further hampered the progress of IPX. Initially, deployment of IP Multimedia Subsystem (IMS) in service-provider core networks was expected to be a key driver of IPX. However, migration to softswitching has subsequently occurred well in advance of IMS, a trend that pushed the industry in the direction of Session Initiated Protocol (SIP). Fixed-network operators were already using SIP-I (SIP with encapsulated ISUP, a part of the SS7 protocol suite) for the interconnection of their softswitch implementations, while moving forward, interworking will be necessary between IMS domain voice and circuit-switched domain voice. “With SIP-I, everybody had roughly the same architecture and signalling plane protocols in mind, and this made TDM-based interconnect with IP softswitches much easier than it would have been had there been a lot of diversity,” says Warren.
Meanwhile, market fragmentation became a distinct possibility following the creation of the i3 Forum in 2006 by a group of carriers aiming to develop business recommendations that would guide interconnection strategy. “At one point, the i3 Forum was positioning itself as being the counter-voice of IPX discussion. But the good news is that we have converged and aligned our IP interconnect principles, and a lot of the i3 Forum recommendations are being worked into GSMA documents,” says Warren.
Standards - renewed interest
The current IPX core technical specification is “IR.34”, otherwise known as “Inter-Service Provider IP Backbone Guidelines”. It contains the technical definitions of the three commercial models for IPX connectivity, identifies the functionality that IPX providers should support, and forms the basis of accreditation. The latter is overseen by the IP Interworking Alliance (IPIA), a forum of the GSMA.
The commercial framework for IPX is laid out within the AA.80 specification. This provides a template contract between a service provider and the IPX provider, with general terms and conditions, SLAs, specifications and connectivity options. There are currently three services with technical and commercial frameworks – AA.81, Packet Voice Interconnection Service Schedule; AA.82, SMS Service Schedule; and AA.83, IP Packet Exchange Service Schedule for multimedia messaging interworking (MMS IW).
According to Warren, the restructuring of IP-based interconnect around a common set of principles makes both commercial and technical sense from the perspective of carriers acting as a community to ease the deployment of IP. Outside this, he concedes that there is little difference between an IPX product and a conventional IP interconnect. Nevertheless, he sees the true value of IPX in providing a product that is marketable to the entire carrier community. Provided that it gains traction, he believes it will result in a fundamental shift in the way the carrier’s carrier business will be conducted. “The carriers are getting on-page right now with the interconnection of circuit switched domain, particularly as they move towards softswitching, and I know there are pairs of operators and carriers piloting that service with a view to it becoming commercial very soon. Meanwhile, the transport-only model is already operational. Beyond that, the real big piece will be IMS-related.”
The GSMA is now focussing its IPX efforts on the principles of voice over LTE (VoLTE) based on IMS implementation, plus another for (Rich Communications Suite (RCS). The latter will specify standards for delivery of an “Enhanced Address Book” and richer calling and messaging features for both mobile and PC RCS clients. GSMA launched VoLTE in February this year, with backing from more than 40 companies and industry bodies. “Defining how voice would work on LTE was the initial requirement, but the real principle is that you can implement it over any access as long as you have IMS as the core platform,” concludes Warren.
.
