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

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

Tuesday, April 03, 2012

Operator WiFi: Seamless is the wrong approach

*Sign up for this blog's email list*     *Attend #TelcoOTT / Future of Voice workshops*

As well as #TelcoOTT services and voice & messaging, many of my meetings over the last month or two have been about WiFi - specifically, its evolution as a carrier-driven technology for offload and other purposes. I've been speaking with various operators, vendors and industry associations. I also participated in a recent webinar (sponsored by iPass, a client of mine) about this theme.

There appear to be two broad trends:

1) There is massive operator interest in large-scale WiFi deployments - for example, the Chinese operators, KDDI in Japan and various players in North America
2) However, there is also massive confusion and naivety about exactly how WiFi is going to be controlled, especially in terms of the user experience from the handset

There seem to be various underlying causes of this confusion:

1) Market participants from the cellular side of the industry are mistakenly assuming that "offload" is the only, or most important, use of WiFi on a handset - often without a decent definition of what offload actually means.
2) Some market participants know that there are multiple use-cases for WiFi, but are actually engaged in trying to subvert this so that their needs and requirements are better-served than those of end-users.

The problems can be largely summed up with a single word: "seamless". The cellular industry has an unhealthy obsession with getting rid of "seams", not understanding that sometimes they are valuable and worth keeping (or even worth adding deliberately).

Sometimes, seamless is good - for example, when a mobile phone user is moving and moves from one cell-tower to the next mid-call, without noticing. Having your phone automatically hook up to WiFi when you get home is good too.

But in other cases, seamlessness is a mixed blessing - for example in the case of international roaming. While it undeniably convenient at one level for handsets to transparently connect to a visited network, the downside is that this can lead to "bill shock", anticompetitive pricing or conditions - especially for the dark art of data roaming. This is why the European Commission is currently trying to re-introduce seams, potentially allowing users to select a separate roaming provider to their home operator. It is also why so many users create their own seam - switching off data roaming entirely.

And finally, there are times when seamless connection is outright bad - for example, if a device is "forced" to use a specific network, when the user or perhaps an app would prefer a different one. This is especially relevant for WiFi, where there are frequently various options for connection, with different ownership, speed, price, security and features. Most WiFi use is private connectivity (it's WLAN - ie wireless ethernet), not offload, and operators have no business becoming involved in it.
Think about an iPhone versus an iPod Touch. Any WiFi use on the iPod is by definition private - it can't be offload as it doesn't have a cellular radio to offload from. Therefore the same use on an iPhone should also be considered as private WiFi access, not offload, and outside of telco visibility and control.

Some of the proposed standards even suggest switching on the device WiFi non-consensually by the network (see this post of mine about OMA), while others such as the 3GPP's ANDSF tries to push or enforce preferences for network selection.

While for certain use-cases this might be beneficial (eg a Kindle-type device connecting to WiFi in the background, with no user intervention), in other cases it will lead to a significant restriction in user freedoms, and may directly inhibit some innovative business models. For example, O2 UK is exploring an onload model for WiFi, aiming to capture users from other network operators, rather than offloading its own subscribers' cellular traffic. There is plenty of scope for conflict where the device (and its SIM-driven seamless connection policies) contradict an app the user has empowered to make its own WiFi decisions.

Another angle on seamlessness comes from a venue-owners' perspective. Imagine that you run a cafe, with WiFi available for customers as perk (pun, apologies!) for their patronage. Your customers like it, some of them connect their phones or laptops, and come back regularly for coffee. Others ignore the WiFi sign for various reasons - perhaps because they are chatting rather than browsing Facebook or doing Skype video calls. Now imagine the same cafe with seamless WiFi. All customers' phones connect automatically. The net result is congestion - users get a poor experience, the cafe owner needs to upgrade the broadband connection, and meanwhile the atmosphere of the venue changes as people spend more time on their phones than with their friends. That is not a positive outcome for seamlessness and automated WiFi log-on. In some cases a little friction is a *positive* for loyalty and promotion - it gives the marketeer a better image of customer-friendliness, while limiting their extra capex/opex. The same is true of airline frequent flyer schemes, which gain loyalty despite the fact it can be hard to actually redeem points for flights.

There are also other insidious aspects here - various vendors I've spoken to have products that can monitor a user's entire handset WiFi experience, tracking which access points you connect to, perhaps enforcing policies even where those APs are not operator-owned or -affiliated. While there are some corner-case exceptions here (eg the perennial content-control for kids use case), this is not acceptable for massmarket use. Indeed, for many enterprises, an attempt by a carrier to check up on private WiFi use could constitute an unacceptable security breach.

Overall, many of the operators and standards bodies involved in carrier WiFi need to go back to the drawing board and start again. Their connection-management designs need to recognise that seams must sometimes remain visible to users, or applications acting on their behalf.

Another way to think about it is that seam=border . And borders can be crossed in many ways - a simple signpost by the roadside (eg in Schengen Europe), basic passport checks, more involved arrival cards, paid visas, deliberate illegal entry (smuggling). Or you can be taken across against your will, blindfolded in the back of a van.

Personally, I don't want my WiFi "trafficked" across borders without my consent. Seamless WiFi has its uses, but it should not be viewed as a universal target or "obvious truth" by the cellular industry.

*Sign up for this blog's email list*     *Attend #TelcoOTT / Future of Voice workshops*


wifidave said...


I've really appreciated the analysis you are developing around Carrier Wi-Fi. I'd like to "harmonize" our separate work in this area to use the term the WBA and Wi-Fi Alliance chose in merging their hotspot initiatives.

I agree with the main points you are making here, especially in regards to the mistake the MNOs will be making if they attempt to wrest the Wi-Fi connection experience/decisions away from the users. I was encouraged by some info I just received from the Wi-Fi Alliance regarding this. Apparently the Hotspot 2.0 spec will not stipulate how the handset Connection Mgr should prioritize the 4 network selection characteristics (user pref, operator policy, app requirements and hotspot conditions). It will be up to the handset vendors to determine how those are factored into the network selection decision. The person I was corresponding with went on to say that he thought it would normally be prioritized user pref > operator policy > etc... I hope he is right.

So I strongly agree that public wi-fi usage should not be subject to operator control. However, I'm fine with operators having high level visibility regarding my use of their wi-fi hotspots, just like I'd be fine with this visibility for any other non-MNO hotspot operator. I think this will be the implicit deal users will be making to obtain access at little to no-cost. The venue owners and operators will be able to gather consumer analytics/intelligence which they will monetize via push marketing and other targeted services. (That's also one reason I believe Apple, Google and maybe Amazon will end up playing in this space as SSPN identity providers - it fits their whole consumer analytics model so well)

As for automatic connectivity leading to hotspot saturation/congestion, the Hotspot 2.0 spec does include mechanisms for the handset to query for WAN metric information, including downlink and uplink load. I'd like to see how this will be practically implemented, but in theory this could address the issues in the scenario you outlined. And there are other mechanisms for the wi-fi infrastructure to "recommend" the handset connect to another AP on a different channel within the ESS. The connection mgr could also look at the WAN load pre-association and decide on its own to connect to a different wi-fi network or fallback to cellular. Lots of possibilities.

Not sure we need to start over on all of this. For my part, I think 802.11u and the Hotspot 2.0 ANQP extensions are pretty solid and will usher in a new era of simple and secure public Wi-Fi. I'm a big fan of these.

I'm much less optimistic about the backend auth/roaming services architecture that I've seen so far. It looks like someone took a cellular roaming architecture, replaced HLR and VLR with AAA server, appended "AAA" in front of roaming hub and called it a day. This is where I see the MNOs trying to control the experience by forcing the cellular subscription model onto public wi-fi. And this is where I see opportunity for better models to be put forward.

I like your smuggling analogy if the operator controls the network selection. I still think you need to extend it along the lines, "Get in the back of the van! We'll take you where you need to go. Trust us." because they would be able to connect you to whatever network they chose. Totally "counter-Internet".

Keep up the great work.


Unknown said...

The operators control which plmn your radio connects to when you are roaming. With the usim and "steering of roaming" centralized control systems from companies like Smarttrust /G&D the operators have total control of the network your phone roams on, and therefore they are the ones responsible for the bill shock.

Should probably discuss how the sim evolved from a authentication token to a Trojan horse used by the carrier.

paulindo said...

Great discussion on the operator love affair with "seamless" (perhaps "carrier grade" is another increasingly vaccuous claim of the Telcos).

In watching the evolution of Voice Call Continuity it always amazes me how the focus is on the technical aspects. Ok, we all get the "seamless handover" bit but dig a bit deeper and I've never seen a satistactory explanation at the user experience level especially around cost minimisation. For why else is offload interesting to operators if not to reduce costs? So, these should be passed onto the consumer - right? And consumers also have a right to know the cost of any good or service before they decide to purchase. Ok, we have published rate plans, but if rates depend on access technology, and access networks make "decisions" on seamless handover without user consent, there is:

(a) The question as to what motiviates a handover "decision" (operator advantage or consumer advantage - I wonder???)
(b) No way for a user to understand service charges before consumption
(c) No way for a user to check or challenge a bill. Even if every handover and rate change was itemised (please no) there is still no way for a user to check that the "service delivered" is as per itemised bill
(d) Possibility of abuse or just as bad, accusations of abuse and complaints like "why did you choose that access network when cheaper alternatives were available"

Its just a consumer minefield.....

So in many cases the removal of technical seams is a noisy distraction that fails to address issues of accompanying commercial incentives in selecting access and how the overall infrastructure cost reduction (and/or re-distribution of costs) results in a more optimal pricing outcome for the consumer.

Any device, anywhere, anytime, ok...but any price?

Ram Krishnan said...

A really inconvenient 'seamless' use case is my iPhone seamlessly connecting to WiFi networks (that are not free and are either ad-sponsored or paid models). Typically, I have no intention of paying for the WiFi so I need to go into 'Settings' and disable WiFi access. And of course, I forget to enable WiFi when I get home and end up using the 3G network