Speaking Engagements & Private Workshops - Get Dean Bubley to present or chair your event

Need an experienced, provocative & influential telecoms keynote speaker, moderator/chair or workshop facilitator?
To discuss Dean Bubley's appearance at a specific event, contact information AT disruptive-analysis DOT com

Tuesday, November 11, 2014

5G standardisation requirement: The need to support multiple concurrent network service-providers

There are numerous 5G-requirements/use-cases documents being circulated by mobile vendors, industry bodies and governmental organisations. For example, 4G Americas recently put out a top-level white paper listing a variety of possible user- and network-functional options for consideration, based on the expected devices and user scenarios.

There are some very sensible conclusions – such as limiting the provision of complex mobility functions to those devices or contexts that actually need them. There is no point having complex, expensive and battery-consuming cellular mobility signalling protocols for IoT installations that are at fixed locations, for example. This type of context-smart design, built into the architecture upfront, could massively improve 5G efficiency and flexibility. This is a big shift from the one-size-fits-all approach seen in 3G and 4G design.

But, ultimately, many of the industry-issued papers are trying to reset the future balance of application/context-aware networks, vs. the growing trend towards network-aware applications. Coupled with hoped-for regulatory frameworks that allow more app-specific network operation, the intent is that mobile operators will retain a strategically important role with additional monetisation opportunities.

In an ideal world, both philosophies would co-exist. However, as the current concerns over net neutrality illustrate, there is considerable fear that network operators could abuse their position, especially in markets with limited access competition. There is a risk of oligopolistic operators demanding rents from application or content companies, which are both financially burdensome and act as a brake on innovation, by introducing extra business-process friction.

The huge success of the Internet model has largely been driven by the decoupling of application from network – and virtually no developers appear to wish to go back to the “bad old days” where they had to interact with the underlying infrastructure providers. Most would rather use a generic “vanilla” data connection. Perhaps others will emerge – especially in the IoT arena – that are prepared to pay for “priority” or “QoS”, but as yet there have been no enthusiastic potential customers. There is a notable absence of any lobbying group clamouring “please let us buy some extra QoS”, which suggests the entire concept is network-push rather than application-pull.

There are other stakeholders here too – notably end-users, and their employers, IoT infrastructure owners, families and other affiliated groups. They will wish 5G to be as responsive to their needs as possible – especially in offering flexibility of pricing/quality trade-offs. Furthermore, if networks are allowed to differentially treat (or even block) specific applications or content, then users will still want to be able to access those services through different channels instead.

Disruptive Analysis believes that a possible solution is to design 5G devices and networks with the explicit expectation that users should always have access to at least two completely independent network providers. This would allow users (or devices / OS’s acting as their agents) to arbitrage between different sources and styles of connectivity, as well as providing redundancy should one network experience a problem.

To an extent, this already happens. Most cellular 3G/4G devices also have user-controlled WiFi, which in many locations enables connectivity completely decoupled from the cellular operator’s control. In addition, short-range connections can be made over Bluetooth for certain use-cases, including device-to-device connections. And of course, there are some devices with dual SIMs or even dual radios. However, these are additions added by manufacturers, and work against the cellular industry’s polarised vision of “one device, one network operator”, rather than being designed-in as desirable models upfront.

It is imperative that the current level of user control over networking switching is maintained in the 5G era, and ideally extended considerably. Firstly, extreme care must be taken in any “multi-network” scenario to protect “seams”. There is a huge danger in trusting the concept of “seamlessness”, as often promoted by believers in carrier WiFi technologies like ANDSF, Hotspot 2.0 and Passpoint. While there are instances where automatic handover is useful, there must always remain a user-override and full visibility over which networks are used. This is especially true where one network is in unlicensed bands, as there will be a good chance that a cheaper/better/different network operator is available locally. This is often the case today with WiFi, where many locations are free at the point of use, and private access of utility WiFi (not “offload”) allows users to avoid using data plans, and perhaps avoid any onerous policy-management conditions their cellular provider imposes.

But ideally, this approach would go much further. Explicitly mandating devices to support at least two cellular network providers – where one might be virtual (e.g. with a Soft-SIM or second SIM or other network-authentication mechanism) would introduce extra competition, innovation capabilities and user/app choice at an architectural level. It would encourage networks to compete sincerely on quality and performance, as they would face easier switching. Applications and OS’s would be able to avoid “bad actor” rent-seeking operators by choosing an alternative path – and this in turn would make concepts such as prioritisation more palatable, as it would act to “keep the networks honest”. It would also make the distorting effects of handset subsidies much less tenable – which fits in with a broader move towards cellphone financing plans, rather than direct subsidy and locking.

A related approach is to make sure that device-to-device connections are decoupled from the cellular operator(s) entirely, and not locked-in to the user’s normal data plan. That way, users and their software agents will be able to “route around” over-restrictive or expensive operator controls, by using others’ connections instead. We already see this with portable-hotspot WiFi tethering, but more generalised and user-friendly forms of “social tethering” can be devised. It is quite plausible to imagine “sharing my connection with my Facebook friends”, or perhaps having separate, independent, D2D service providers.

Apple’s SIM card, with its selectable mobile networks, is a good step in the right direction here, but care must be taken to avoid anointing Apple or Google as another single point of network control as well.

A further possibility here is to apply strict Net Neutrality to just one of the connections. That way, the user always has open access to all legal apps and services via the public Internet, but the other connection(s) could support alternative business models, if they prove technically and commercially viable. Indeed, the user might choose to purchase both Internet and non-Internet service connections, and thus allow non-neutral connections to prove their “more innovation and investment” credentials over time.

Overall, 5G requirements and standards should support switchable, multi-network capabilities, using independent service providers, with at least one neutral Internet connection. By mandating this:

  • Users and applications can always exercise choice over network connections and pick “the best tool for the job at hand”
  • Network connections can offer optimisations for specific applications or contexts, potentially getting revenue premiums from end-users, or (conceivably) application providers.
  • Full and open Internet access is always available, but so are other potential “managed” services which can compete directly and fairly. 
  • Innovative solutions involving network bonding, roaming, D2D sharing and other models will be encouraged. We may see the emergence of new classes of secondary network service-providers.
  • WiFi or other unlicensed technologies remain valuable as an “arbitrage layer”. We may see the emergence of unlicensed-band LTE/5G networks run by third parties.
  • Regulators can manage to “square the circle” of guaranteeing both consumer choice, open Internet access and the potential for new mobile broadband business models. 
  • There can be good fits with new MVNO or liberalised-MNC models for network service provision.

All that said, it seems unlikely that the traditional mobile industry will suggest this approach itself, as the usual model tends to view lock-in as preferable to loyalty and choice. It will be important for regulators and other stakeholders (eg Apple which is now a member of NGMN) to push for multi-network capabilities to be recognised as a 5G requirement. This is a critical area, which seems to be overlooked at the moment. 5G needs not just “seams”, but “zips” that allow it to be more open, more versatile, and of greater benefit and value to the user.

1 comment:

D.B. said...

Dean, I like the idea but I'm not sure it can be made to work without regulation. Remember the good/bad old days in the UK when the tracks, trains and marketing all belonged to a vertically integrated British Rail? What you're proposing sounds like how the UK trains run now (or the Texas utilities), where the infrastructure is dislocated from the service financially, with the marketing, billing and sometimes the final part of delivery separate from the underlying "pipes". I don't think the market tends to drive that, it's usually lawmakers and regulation.