Should Cellular be part of this converged IP network evolution? It already is part of it... especially in the Core and Transport parts of the network, and slowly in the Radio Access Network (although I have seen some interesting developments from Cisco in this area).
How can this be achieved? Well, I don't think there is enough space here to start the discussion in detail, but it starts with an internal discussion within your company about cost savings, innovation and what you are trying to achieve by this effort within the overall business strategy of the company.
For example, there are cost savings that can be obtained by consolidating various IP/MPLS back-bone networks in your organization, followed by the organizational operations and systems needed to support and maintain this consolidation. Why? Due to M&A growth in your company, it makes more sense to have one consolidated IP/MPLS back-bone and one consolidated operations than various separate ones. Of course this is a simplistic first start, but a first start towards many more objectives that would have been discussed internally within your company on "why" and "what" as mentioned earlier.
For most (successful) Carriers/Operators, it's about innovation through enhanced capabilities for the future, while obtaining cost savings through organisational consolidation followed by technology convergence.
I think people are too focused on the technical side of converged IP networks, when in my opinion this is not that important, but rather the business transformation needed to deal with converged IP network evolution is ignored or forgotten. This is the hard part where more questions should be asked because this is where the bulk of the issues and problems exist.
In reality there's two different IP convergences going on right now.
First there's the infrastructure convergence. Carriers are quickly realizing the value of a packet switched infrastructure over the traditional circuit switched architectures. For example pretty much every carrier is, has, or will be converting from circuit switched ss7, to ss7 over IP. But the driving factor isn't so much evolution of the design, as it is trying to find cost saving ways to implement the design. Similarly while cellular carriers are starting to block services such as Skype, they're implementing carrier class VoIP to connect their major switching centers to make better use of the circuits they already have.
Second, the other convergence going on is with the customer data. The reason I separate these, is the customer IP data for the most part is tunneled through the wireless infrastructure. While routing of that data is going on, that routing is not based on what data the customer is using. For a wireless device, http traffic and wap traffic traverse the network identically. So where is the convergence here? I think we're just starting to see carriers open up to evolving this part of the network. An example of this type of convergence is with T-Mobile's 'hot spot at home' offer, where your device can make voice calls over wifi from an access point at home, or through the cellular network. To be able to hand off from what is largely still a circuit switched call to a VoIP gateway through the internet is a big step forward.
In a world where carriers make a penny or two per minute of use, I believe it's the carriers who think outside the box who will lead the way. When they learn to leverage cheap outside IP resources (wifi hotspots like above) to bolster their expensive network core then everyone can benefit.
The simple answer... is that IP networks don't care about the transport. SIP (session initiation protocal) and IMS (IP Multimedia Subsystem) provide a way that transport providers such as telecoms who previously shunned application connectivity and API's can now offer integration with a large range of applications and network attached computers. In that context, the IMS layer knows who you are and what kind of network attached computer you're using at the moment, whether that be a PC or a mobile handheld. User experience can then be delivered relative to the capabilities of the device
3GPP and ETSI are working on an architecture called IP Multimedia Subsystem (IMS). This is an architecture to manage network infrastructure in a technology agnostic way to provide a standard way to provision services on top of IP and other network technologies.
Think of it as the SS7/IN standard which allowed the development sophisticated services (number portability, non-geographic numbers etc...) on top of the standard telephony infrastructure.
What IMS will do is effectively abstract out the network technology from the perspective of the user applications.
Though IMS was initially developed by 3GPP... which is a mobile/cellular standards group..... it was quickly adopted by ETSI TISPAN for wireline purposes with some tweaks. Both groups work closely now. IMS has gained a lot of interest from both wireless and wireline operators.
Consequently it's not necessary to worry so much about where IP convergence will go. The impetus to standardise on IP across the board was driven by the need to reduce complexity and gain consistent network services. IMS acts as an abstraction layer and thus to a certain extent deflects the drive for IP convergence. Having said that the cost savings achievable and capacity gains from a uniform network technology platform is still a powerful driver. Given that IP is here and works well, the convergence will continue.
Much of what we have in place now was developed as "art of the possible", meaning best with the technology available at the time.
Once we have invested large amounts of longterm infrastructure capital, there is a damping effect on change. The telephone system is a good example of this - synchronous, highly optimized (channelized and compressed) and "service-ized" (it costs more to print and send the phone bill than it does to provide the actual connection).
Having said this, it is clear that converged IP networks will occur slowly as benefits are realized by the operators. For example, lots of backbone voice traffic is carried on IP but done over a SONET infrastructure.
Following up on cellular - again we have huge investment, international road maps, and vested interests. If we were just starting cellular deployment in 2007 (given WiFi and related technology) would we build it the way it is? Probably not, but we have an existing infrastructure that must play.
The good side of this effect is that things like VoIP were well thrashed out before they were deployed (I'm not talking about Skype or Vonage here).
IP will generally rule. It is designed to work over just about anything (smoke signals anyone?) and has a proven track record. I think the improvements will come in the "routing" protocols rather then in the "routed" protocols ....so there is plenty of room for innovation during periods of wider convergence.