The sixty page document “Next-Generation Mobile Networks (NGMN): Beyond HSPA and EVDO? is the latest white paper of NGMN.org, an initiative by the CTO’s of China Mobile, KPN Mobile, NTT DoCoMo, Orange, Sprint Nextel, T-Mobile International and Vodafone Group. It provides a technical requirements framework to vendors for the next iteration of mobile networks.
To be clear, what’s defined is just a technology toolkit. Different carriers may deploy it in different ways with varying business models and services. Until we see the business models, jubilation or damnation is premature. Nonetheless, this is an extremely important document. The “walled gardens? of 3G are starting to look like weed patches, and this is a rare chance to define a truly new Telco 2.0 approach that takes the best of the Internet and traditional telecoms models.
The document avoids wild flights of fancy about sophisticated combinatorial services, and focuses on practical implementation concerns of mobile broadband. It rightly sees the mobile ecosystem as a co-evolution of devices, access and services. This offers a valid and viable parallel/alternative path to the fragmented and sometimes chaotic Internet approach. It’s clear about what generic classes of service are to be offered, and what tradeoffs are likely to be acceptable. The document also outlines a very much evolutionary approach: business-as-usual, only faster and cheaper.
And therein lie the big questions:
* Does it go far enough in addressing the forces tugging apart network access, services and devices?
* Does it react to the counter-forces that would push them back together in order to address deep architectural issues of IP and the Internet (such as weak security and low efficiency)?
Our answer based on our reading is “maybe, if deployed right? — but you need to be a bit of a Kremlinologist to read between the lines and think about what’s left unsaid.
We’ll start with the easy bit: things in the document that make sense about Making Money in an IP world. Then we can delve into the more philosophical and practical limits of that IP world and how a next-generation architecture might address them.
Plenty to praise
There are many positive improvements proposed. Some highlights might include:
- Self-configuring networks that cost less to run.
- Improved scheduling algorithms that focus on user “quality of experience? at the periphery of a cell site, rather than RFP-friendly numbers for maximum burst throughput standing under the cell tower at 3am on Christmas morning.
- Flexible and modular service-oriented architecture to accommodate future change.
Put simply, whatever NGMN turns out to be, operators want OSS and BSS thought through in advance, and for vendors to take responsibility for the operator and user experience post-installation. So far, so good.
Aligned with several Telco 2.0 trends
There are also some features which come with the “Telco 2.0 Approved��? stamp because of their reflection of the business trends we see:
- The ability to share equipment and do more slice-and-dice of the infrastructure similar to MVNOs, but better. We believe infrastructure sharing and new modes of financing/ownership as being a key Telco 2.0 trend (as we will discuss at our forthcoming Digital Town event workstream).
- Stronger device and end-to-end security to enable transactions of money or sensitive data. As telcos are already diversifying into the payments and identity business, this can only grow — and depends on such enabling infrastructure. DoCoMo are part of the consortium, and given their trailblazing in payments services, we’re hopeful of seeing diversification successes of operators elsewhere based on their learnings.
- Detection and mitigation of network traffic resulting from malware or attack. This we feel will be a growth area as the services become less controlled. A limitation of the “intelligence at the edge? concept is the ability of those edges to collaborate to detect and eliminate abuse. The experience of email spam and phishing tells us that not all is wonderful in Internetland.
Moving on, there are several things conspicuous by their absence.
The Internet elephant in the corner
Apart from some in-passing references in a few tables and diagrams, the word “Internet? is wholly absent from the document. It’s a bit like Skype, YouTube and BitTorrent never happened. In fact, you can only conclude this absence is deliberate.
It could very well be that the technology defined can be deployed in very different manners, and operators may take radically different approaches — such as the contrast between 3 and T-Mobile in the UK embracing open Internet access, O2 trying to keep people on-portal, and Vodafone outright banning many popular Internet services such as IM, VoIP and streaming. Will operators want to continue to ride the “Telco 1.0? command-and-control horse, or switch to a more open “Telco 2.0? Internet-centric approach? Will the point of a future mobile network to channel bits back at all costs to a cell tower where they can contend for expensive backhaul to be deep-packet-inspected. metered and accounted for? Or will it complement the other infrastructure that exists?
The IMS mouse in the cupboard
Equally conspicuous by its general absence is reference to IMS. Our take is that there could be a polarisation here between “service-centric? operators trying to define interoperable new services and compete against Internet players; and “connectivity-centric? operators who create “smart dumb pipes? and enabling platforms for a wide ecosystem of players. You could deploy NGMN and completely ignore IMS if you chose to do so.
Local connectivity, globally interoperable
At the other extreme of connectivity, another thing not given much ink is the explosion of highly local connectivity. For example, we’ve just passed the billionth Bluetooth-enabled device. Motorola’s Chief Software Architect, John Waclawsky, described this at the last Telco 2.0 event in October in his presentation “From POTS [telephony] to PANS [Personal Area Networks]?. The mobile network itself can still play a part in this, such as offering directories of resources. If you’re sat in Starbucks today and want to print out a document, you’re out of luck — the network can’t help you locate or pay for such services.
Given that this is an integrated vision of handset, network and service evolution, we think it may be gold-plating the longhaul connectivity vision, and underspecified the local connectivity one. The business model will also need to evolve, since there may be no billable event. It has to anyway: products like Truphone will make it ever easier for users to bypass or arbitrage network access.
What’s the commercial vision?
Naturally, the operators can’t write down a collective commercial vision (because of anti-trust), nor an individual one (due to commercial confidentiality). So you have to impute the commercial vision from the technology roadmap.
The stated requirement is for a network that’s low-latency, efficient, high-throughput, more symmetrical, good at unicast, multicast and broadcast, cheap, and interoperates seamlessly with everything that went before it. It’s a bit like low-calorie cream-topped chocolate cake. Sounds like a good idea, until you try making one.
The inevitable billion-dollar question is what are the services and the business model that will pay for all this? The experience from 3G was that “faster? isn’t itself a user benefit of significance (particularly when it doesn’t work indoors!) In fact, given that battery technology follows a curve well below that of Moore’s Law (or its transmission equivalent), there’s the “oven mitt? problem of early 3G handsets still lurking: how to create hand-held devices that are physically capable of sourcing and sinking data at such speeds and over such distances (and high power) — and that create services users care about in the process.
Or, to put in another way, why sync my iPod over the air slowly when I can plug this USB cable into my laptop and do it at 400Mbps for free?
What is a mobile network for, exactly?
There’s a significant difference of expert opinion here that’s worth noting. There isn’t universal agreement on what wireless networks are best used for compared to wireline. For example, Peter Cochrane, the former CTO and head of research at BT has long been keen on forgetting DSL and copper and going all-wireless. NGMN’s ambitions to match and exceed the technical and cost capabilities of DSL suggest a commercial vision of competing against fixed access.
Our take is that success is most likely to come from intelligently blending the best of fixed, mobile and media-based delivery of data, rather than an absolutist approach to any one of these. Furthermore, the unsolved user problems are more to do with identity, provisioning, security and “seamlessness? than speed or even price. Finally, users don’t generally see the up-front value in metered or fixed buckets of IP connectivity, particularly given the anxiety it causes over cost or overage. True unlimited use isn’t technically possible, so the network has to allow connectivity to be bundled into the sale of specific device or application types, where traffic is more predictable.
Stop looking for the platinum bit
The hypothesis seems to be that some bits will be blessed with “End-to-end QoS? and continue to gather super-premium pricing (by many orders of magnitude). The need for this QoS capability is repeatedly stated. At the same time as the network capacity, latency and cost improve to near-wireline levels. I think you can spot the problem. I’ve made a successful Skype call to someone 35,000 feet up on a 747 somewhere over central Asia, and there wasn’t any QoS involved.
Our post on Paris Metro Pricing attempts to challenge some of the assumptions that drive this requirement. It sounds esoteric to those from the commercial side of the business, but ignoring this small technical detail is telecom’s equivalent of the frozen O-ring. Set the price high, and at some point all the valuable bits flow around the “premium pipe? and not through it, and the commercial model fails.
NGMN could be part of the solution here, not the problem. If operators can switch to a congestion-based mode of pricing, rather than pure capacity, they could offer users a far better deal.
What are the real sources value?
Here are some examples of requirements in the document, and how NGMN provides opportunities for product and business innovation:
- Making user data more seamlessly accessible, blurring the line between online and offline. The specification includes
Standardised APIs (i.e. not operator or handset-specific) to sync online and offline data like address books, so the user doesn’t have to care so much about network connection state. This whole process could be taken much further to cover all content. This lecture video by Van Jacobson, former Chief Scientist at Cisco, points to a very different future network architecture based around diffusion of data rather than today’s packet-only networks where you have to know where every pieve of data is located to find it. (Hat tip: Gordon Cook.) This isn’t a theoretical concern: wireless networks readily become congested. Maybe it’s time to reward your neighbours for delivery you the content, rather than backhauling everything across the globe. The Internet’s address space is flat, but its cost structure is not.
- Deeper coverage, richer business models. The document talks about hub terminals (e.g. femtocells). Deep in-building and local coverage is a clear user desire. The first step is outlined, but there’s no corresponding economic model being included. Companies like FON and Iliad are doing innovative things with user-premises equipment and roaming. We nope NGMN doesn’t repeat the experience of Wi-Fi, where hooks for payment weren’t included (causing a mess of splash screens), and the social aspects neglected (am I sharing this access point deliberately?). The existence of bottom-up network deployment is an interesting possibility. You need to create new security and payment mechanisms so that local entrepreneurs can extend networks based on local knowledge and marketing. Top-down is becoming top-heavy.
- Support for a diverse array of charging models. It’s in there, but could get lost in the deep-packet-inspection swamps. The genius of telephony and SMS is to sell connectivity bundled with service in little incremental slices. We’d like to see richer, better and simpler ways of device makers and service providers bundling in connectivity. (See out earlier artlce on this for more details.) For example, the manifest of a download application could say that Acme Corp. is going to pay for the resulting traffic — and the secure handset will ensure it’s not abused to tunnel unrelated data at Acme’s expense. NGMN could enable this.
- Uplinks vs. downlinks. Users create as much content as they consume. Devices are equipped with multi-megapixel cameras and video capture, which will be uploaded for online storage and sharing. That media is then often down-sized for viewing (if it is ever viewed at all). Yet the standards continue to emphasise downlink performance. We’ll acknowledge that from a technology perspective uplink engineering is like trying to fire bullets back into the gun barrel from a distance. Somehow this issue needs to be looked at. NGMN takes us closer, at least.
- Peer-to-peer. A great requirement in the specification is “better support for ‘always on’ devices, with improved battery performance and network resource usage.?. We’d second that. But given this requirement, where’s the peer-to-peer specification of the services those devices should host? Or do operators still believe that the purpose of the network remains distribution of professionally authored media entertainment from “central them to “edge us?
- Building an identity-centric business. Another good requirement is for more advanced modes of device authentication, such as sharing a SIM among multiple devices. In some ways it defines an “identity network that is independent of the NGMN, and potentially fixes some serious problems with the Internet. Mobile networks may happen to use those identities, but they’re equals with other uses. We’d encourage more creative thinking in this area.
Summary thoughts
Overall, it’s a good piece of work. Change doesn’t happen overnight, and given a 3-5 year time horizon, the world will not be beyond recognition. Nonetheless, without a parallel vision of business model evolution, much of the investment in NGMN could become as equally stranded as that in 3G. With the right vision, it could make the “mobile Internet really work, since the “real Internet continues to be a polluted, expensive and frustrating experience for users.