[VOIPSEC] Starting a 4G Issue Map

[Geoff Devine]

Mobile VoIP:
If you look at today's networks, pretty much all phone calls are now VoIP.  It merely depends on where in the architecture you're converting from circuit to packet.  On today's 2G networks, you convert at the output of the mobile switching center and haul traffic between offices as VoIP.  There are other architectures like UT Starcom where you convert at the base station.  On licensed spectrum, I think CDMA and GSM traditional methods to transport voice will continue to be used since they make the most efficient use of bandwidth.  With IP header suppression on the access network, VoIP will start being phased in on some networks but the migration will be slow because of the cost of replacing all the access network equipment.

On unlicensed spectrum (using WiFi and WiMAX technologies), there are still QoS and authorization/authentication issues to be resolved.  The standards now (mostly) exist but you can't go to Best Buy and get a home WiFi router that will do VoIP properly.

Fixed Mobile Convergence:
There are 100 solutions out there today.  3GPP is arguing about how to build it.  The reality is that business issues drive the implementation.  Does the cellular operator own the customer?  Does the wireline operator own the customer?  Does the ISP own the customer?  All three could be different.  The business relationship between the various operators will end up driving the architecture.  It's easy if it's Verizon and Verizon Wireless or the AT&T wireline telephone and internet product and the AT&T/Cingular cellular product.  It's much less clear if it's spread between different corporate entities.  In my opinion, the cable operators will end up buying spectrum (or merge with one of the 2nd tier cellular companies) to compete against the Verizon and AT&T combined product offerings.  

Triple Play:
The MSOs are already selling bundled services.  I happen to live in this world and it's pretty clear that it's a winning formula.  You can afford to discount the bundle to the point where you end up virtually giving away the telephone part of the service for free to sign up customers to your premium video products and internet product.  It gives you new customers and prevents churn in existing customers.  

I'm dubious that the wireline players are going to be significant players in this market for many years.  Video is a "buy content wholesale, sell it retail" business.  The cable operators give a significant fraction of your CATV bill to the content providers.  The margins aren't all that great and satellite has kept them honest.  I keep hearing that it costs over $1,000 to run fiber to a home and that's assuming something close to a 100% take rate.  I don't see how you can make the business case for a product like FIOS since the entry of the TELCOs in the video business will obviously drive down prices.  The MSOs already have their plant in place and COAX is a much more cost-effective solution than fiber to the home.  

Quadruple Play:
The truism in the cellular industry is that the most important thing is coverage.  The 2nd most important thing is coverage.  The 3rd most important thing is coverage.  Verizon Wireless and AT&T/Cingular have roughly twice the antennas of their 2nd tier competitors.  Personally, I'm not going to swap my cellular provider to get a bundled service from my cable operator where the coverage is inferior.  Until the MSOs come up with a technical solution to this problem, the two dominant cellular operators in the US aren't going to have any incentive to lower their prices by bundling service.  Cellular is where their profit is these days.  Wireline just isn't all that profitable.

Technical issues:

SIP: 
In my opinion, this protocol is proving to be a disaster.  If you look at the work being done in the various standards bodies, it's exploding in complexity.  Messages are now incredibly bloated.  You have to support gigantic recursive history headers to detect and prevent forwarding loops.  The notion of GRUU is being introduced to cope with multiple devices sharing one public identity.  I'd prefer to throw the whole mess away and start over using a suite of type-length-value protocols and have everyone agree on a common architecture.  

QoS:
QoS is a business issue.  You can't just give it away or everybody will always request the highest possible priority.  In walled garden architectures like PacketCable and 3GPP, QoS works just fine.  When you introduce unlicensed spectrum and access via the public internet, we need to sort out the business issues to be able to authorize and charge for QoS.  I view this as a business issue rather than a technical issue since there are already several possible technical solutions that would work.

SPAM:
As long as your telephone identity is an E.164 number and everyone uses trusted service providers, you don't get SPAM phone calls.  As soon as you allow phone calls to a non-E.164 URI, you've opened up a huge can of worms that we all know and love in our email environnment.  Personally, I'll pay good money to ensure that my phone doesn't get flooded with garbage calls from Nigeria.  I will cancel service if this ever starts happening.  Somehow, I keep hearing from IETF people that this is a _GOOD_ thing.  I don't see it.

Re-inventing the wheel:
I have yet to see a feature I'd want that can't be done already on legacy networks.  I remember seeing GSM, PBX, Wireline, and DECT integration at Ericsson a decade ago.  1 phone number worked at home, at the office, and on your cell phone.  Somehow, all of this became brand-new again a decade later.  

Technology enabling disfunctional behavior:
VideoPhone has been the next big thing since Bell Labs demonstrated it in the 1960's.  It's still not here yet and may never achieve market acceptance.  The one big change we've seen is that cellular has taken over the world.  I'm still not convinced that I'm willing to go blind watching microscopic videos on my cell phone but we all seem to accept that the Blackberry has replaced the pager as the most personally intrusive device in the known universe.  When they tried to give me a pager 15 years ago, I removed the battery and placed it safely in my desk drawer.  I threaten the same whenever they threaten to issue me a Blackberry.  There's such a thing as being too connected.  I think all restaurants, movie theaters, concert halls, trains, buses, conference rooms, ... should be outfitted with radio scramblers so we can have civilized behavior again rather than listen to someone shout into their cell phone or ignore you when their blackberry rattles.

Geoff
-------------------------------------------------------
I'd like to invite the community discussion on the following:

1) Mobile VoIP
2) Fixed Mobile Convergence
3) Triple and Quadruple Play

Reply on something your curious or passionate about.

If possible, please post with comments on:

- top line issues
- positioning in the technology stack: PHY, MAC, media stream, control, etc.
- your sense of significance of the issue
- grist for others to join you in comment as a thread