Carrier Models For Interconnection - Exchange And Settlement

Few doubt that IP will replace circuit-switched (TDM) architecture in national and international long-distance interconnect. Yet the way in which current commercial wholesale models will evolve in support of this transition is still fiercely debated. In one camp are the internet purists (drawn largely from hardware and software vendors), who argue that the “bill and keep” (BAK) settlement-free model should apply across all IP-interconnect scenarios. In the other camp are those (predominantly carriers) who argue that wholesale settlement will always be a necessity, because network providers add value and are therefore more than just bit-pipes.

“The IP interconnect dream has always been to allow any user to communicate with any other user via any application over the internet, and that the service itself will be free because the user pays for access to the network,” says Jim Dalton, founder and CEO of BSS provider Transnexus. BAK may work perfectly well for SMS and email, he says, but for voice and other applications, it will not be sufficient. “The argument I make, as do most carriers, is that because network providers add value – whether that be in terms of transcoding packets, or providing an air interface (which is a major value-add to an IP connection) – they should be able to charge other networks for carrying that traffic.”

Under the conventional inter-carrier model, settlement is tied to calling party’s network pays (CPNP), where the originating network of a call makes a wholesale payment to the terminating network. CPNP for interconnect can be made under either bilateral interconnect, negotiated directly between two service providers, or multilateral interconnect, allowing for an intermediary operator to provide one commercial and technical interconnect that gives reach to multiple destination networks.

With legacy interconnects, the accounting rate for wholesale voice has traditionally been time-based although a capacity-based fee is gaining traction. The latter allows operators to request a specific capacity for interconnect and then pay a flat-rate fee to the terminating party. “I think, from talking with a number of players in the market, voice will move closer to cost-based charging,” says David James, principal analyst, wholesale telecoms, at Ovum. “If you bill by cost, then that is more related to the amount of traffic and the complexity of service you are providing – whether voice or video. However, the transition will be painful, because current billing systems are built on the assumption that voice is charged by the minute. Although network elements can report all kinds of events that can be used in terms of billing, the greater change is in providing this as a business model.”

In the IP domain, settlement is complicated, not only because of the nature of packet-switching, but also because there is no standardised mechanism such as the call data record (CDR) for telephony. Under conventional IP interconnects (ie that of the public internet) there are two types of commercial agreement: peering – when IP networks of a similar size choose to interconnect directly to save costs, and operated as BAK if there are no interconnect fees; and transit – where smaller IP networks pay to interconnect with larger IP networks, with fees based on the cost of access rather than usage.

The process by which an IP network qualifies for peering is usually privately negotiated based on market position, network coverage, volume of traffic, range of services provided and network reliability. According to analyst firm Telegeography, only a handful of the world’s largest ISPs are able to exchange all of their traffic via unpaid peering relationships. All other service providers must rely on wholesale internet connectivity (IP transit) from other backbone providers for at least a portion of their traffic. Tier 2 and 3 providers pay for both upstream (specifically traffic generated from their retail base) and downstream traffic, while Tier 1s might also adopt settlement-based interconnection.

Similarly, carriers currently operate on a settlement basis with full charging per minute. Often this is packaged in a bilateral commercial contract, and sometimes with net-off/balance traffic parameters. It is unusual to find BAK relationships between operators outside north America. “We remain in a world of bilateral agreements for the exchange of pretty much any type of traffic,” says Dalton, who has direct experience of what happens when challenging the status quo of access and transit fees. He founded Transnexus in 1997 with the aim of being a multilateral services and rating provider, but had to refocus just two years later as a software provider for routing, rating and billing of interconnects. 

There are those that believe the industry is on the cusp of significant change, driven by: the co-existence of TDM and IP infrastructures; the advent of IP telephony and VoIP peering under “federated” network models, supporting both bilateral and multilateral arrangements; and the fact that regulators in Europe and the US are looking closely at the potential of ex-ante regulation surrounding the BAK/settlement-free model for IP interconnects (see the diagram opposite).

“We are talking about a seismic shift in interconnect models in moving from old to new, and it is much closer to a revolution than an evolution,” says Eli Katz, founder and CEO at Xconnect. “There won’t be an implosion of bilateral, but new models are evolving that suggest a change in telecoms life as we know it. We see a variety of additional commercial models emerging, especially from the newer operators, including cascade payment, multilateral relationships and also managed bill-and-keep relationships.”

According to Katz, the federated network hub delivers on the primary drivers in the modern interconnect environment: end-to-end feature transparency (protocol support, interworking, QoS), the most cost-effective route for calls, avoiding mediation and multiple hops, and with the highest quality. However, he also stresses that this is not a case of “hippy-based, free love”. Rather, he sees it as a viable alternative in the longer term. He also points out that Xconnect supports a variety of commercial models, including the classic bilateral, the GSMA’s cascading payments model specified within IPX, and a multilateral approach that can be managed on a peer-by-peer basis. 

With IPX, the GSMA has adopted a more tightly-defined approach to multilateral hubbing than other peering models, since only cascade payment arrangements may be applied. With cascading payments, each party involved in the transport of traffic is compensated for its efforts in providing services to the end user. The underlying principle for this is that the party who perceives a value in offering a service should pay for the network capacity used. That is, the party who generates revenue from the transaction pays for the required transfer of data across the IPX domain.

“The issue – and IPX kind of hits this – is that if you want interconnect among carriers, service providers and networks to flourish openly, you need to give them an economic incentive,” says Transnexus’ Dalton. “In the PSTN that is settlement. But in the IP network, people think settlement is a business model that doesn’t make sense. Talk to people that run the IP networks and that pay billions of dollars for the devices to handle new services and they need a return. So why should it not come from other service providers and their subscribers who pay for some of that?”

Yet carriers have so far been reluctant to enter into multilateral peering relationships for VoIP. “There has been a lot of buzz in terms of supporting the next wave of global interconnect of calls on a settlement-free basis, and a lot of emphasis on multilateral relationships where groups of carriers exchange traffic,” says Aaron Blazar, vice president at market watcher Atlantic-ACM. “But one of the things interconnect providers have realised as these relationships and models have evolved is that carriers are very hesitant to get into multilateral relationships because they appreciate the control and the security a bilateral or direct interconnect supports.”

According to Blazar, only a handful of third-party interconnect providers continue to support multilateral peering, given that MNOs are still in the process of moving to IP, whilst IP-originated calls via 4G LTE and Wimax are still some way off. He also highlights the necessity of having an anchor tenant. “In the first five years of VoIP peering, we haven’t really seen a major player coming to market with a scaled VoIP peering product. The key inhibitor is getting the large carriers who control large bodies of end users to buy in and realise they need to evolve to an IP interconnect model,” he says. “Carriers realise that as soon as they do buy into an IP interconnection model, they are disintermediating their embedded base of legacy wholesale revenue. So we are still early in the game in terms of how this plays out.”