An 802.11 hub in every pot

Rohit Khare (rohit@uci.edu)
Tue, 31 Aug 1999 11:12:15 -0700


DanMKohn: Have you been following the new 802.11 DS 11 Mbps
standard? I saw a demo of the Lucent (Wavelan) equipment, and it
seems like its ready to be a real breakthrough technology, a key
component of munchkinware.
DanMKohn: The Lucent guys say people are installing it in
hotels, airports, etc. It seems to me there is a business to start in
creating a DHCP-extension to allow authentication and billing for
these sorts of roaming services.
rohitkhare: Yep. We're just installing it around the office here.
Farallon's promising a G3 Powerbook card soon

rohitkhare: Hopefully a better job than Ricochet's
nationally-centralized auth system

DanMKohn: The AirPort standard is just OEMed Lucent hardware.
DanMKohn: I'm serious. I think there is a business to start
here, if you could navigate the standards waters.
rohitkhare: Well, I figure it's a business similar to the
'fastlane' hourly-office concept

rohitkhare: I mean if it's walk-up access points, that's one business
rohitkhare: (advantage goes to those with physical presence,
airport lounges, etc)

rohitkhare: And if it's metro-scale, then it... well, it's not
802.11 per se
DanMKohn: I'm saying creating the equivalent of the NACN
network which is the settlements house for cellular.
rohitkhare: The business I'm thinking of is still a layer higher:
postage on message delivery.
rohitkhare: OK: re: settlement
DanMKohn: I don't think the propagation of 802.11 is that
different from Ricochet.
rohitkhare: Dial IP roaming is still a dicey proposition, but
that's the route to test this stuff.
DanMKohn: The issue is creating a business model that
encourages folks to build out as many hubs as possible.
rohitkhare: Question is, can you position an 802.11 business as a
premium component of a national dial isp?

rohitkhare: Ah -- I see, making little teeny profit centers

DanMKohn: Yeah, probably the infrastructure that iPass and GRIC
have set-up is the best jumping off point of 802.11 settlements.
rohitkhare: OK, so suppose I'm a national ISP, and I start
offering one cent per kilopacket for any traffic you deliver to me by
802.11 public-access circuit. Just play for a sec

rohitkhare: I mean, we're not in the peer-to-peer networking
model yet; it's a client-server network today
rohitkhare: And the key to rapid grass-roots deployment is
financing $1000 endpoints (short of issuing a few billion in bonds
and doing-it-yourself
DanMKohn: It wouldn't make sense for all traffic. Besides I'd
lie about the traffic I was sending. I'd keep sending etc/termcap
and say its from 802.11
rohitkhare: (and the line-of-sight on ricochet is far better, I'd bet)
rohitkhare: Waste, fraud, and abuse comes back in a sec -- I'm
assuming the ISP is also *charging* the actual end-users, like 2
cents per kilopacket
rohitkhare: Money off of the traveler, with a kickback to the FBO
DanMKohn: It's more that they would tell me that someone with
the following authenication creditials wants to log on. Will I pay
them 1 cent per kilopacket.
rohitkhare: Yep
rohitkhare: there has to be a session-based 'roaming'
authentication in there

DanMKohn: You could even do dynamic negotiation of price levels
by time of day, location, et al.
rohitkhare: But seriously, why not ask instead if we could
capitalize a $1B today...
DanMKohn: Right. I'm suggesting that iPass and GRIC have put
that infrastrucure in place for international roaming on dial-up ISPs.
DanMKohn: Can diameter do this?
rohitkhare: To me, it's still in the endpoints. The
802.11-equipped palm pilots and such aren't here just yet
rohitkhare: diameter is a joke - literally
DanMKohn: What about a user interface that let's me decide
whether I'll pay the current price.
DanMKohn: Radius then.
rohitkhare: Apple presold 140k ibooks so far

DanMKohn: WinCE devices support PC Cards, and 802.11 PC Cards
will soon be available for $100.
rohitkhare: Yeah, I have no doubt the dial-roam solutions will
work here too; MCI was hot on the trail, and iPass does already kinda
work. No question there
rohitkhare: Oh I got ya on the early-adopter $99 cards angle.
DanMKohn: Lucent thinks they have a two year window to make
their PC Card-based product as ubiquitous as possible. After that,
they believe that 802.11 WLANs will be built in standard on laptops,
as V.90 modems are today. They want to win the battle now to ensure
that their chipsets are used going forward.
rohitkhare: I'm thinking more devices like, say, hotmailmate, a
dedicated email device.
rohitkhare: Good point on Lucent's part.
rohitkhare: Q: Can 802.11 chipsets ride the price curve fast
enough to destabilize bluetooth, seeing how delayed they are?
DanMKohn: So, you don't want to start a competitive settlements house?
DanMKohn: What's bluetooth's range?
DanMKohn: Has it shipped?
rohitkhare: Perhaps I'm naive, but I don't see the settlement
business as viable without massive scale -- i.e. lots of your own
traffic

rohitkhare: Bluetooth is 10 feet to 20 feet. No, still prototyping
DanMKohn: Right, 802.11 is one hub per floor.
rohitkhare: Hmm. What about an 802.11 based IP telephony 'handset'? 

DanMKohn: I really think 802.11 is going to win and kill
bluetooth. And I'm so relieved that FH can't go past 2 Mbps, so DS
wins by default.
DanMKohn: Exactly.
DanMKohn: I really think this is THE technology for making
munchkins a reality.
rohitkhare: Yes, I do see them munchkin link.

rohitkhare: A wince/palm/psion 'computer' with an 802.11 link in
sufficient quantity that you can do the settlement business at the
application layer -- actual emails and vmails and web pages

rohitkhare: The 802.11 hub business has to go away too.
DanMKohn: Using what standard?
rohitkhare: The future $100 chipset also has to be able to
function as a hub if none's present -- peer to peer is killer

rohitkhare: The app-layer settlement business will have to be
based on a new hybrid, or meta protocol (*TP)
DanMKohn: You'd like the Lucent hub. It's a box with a 10baseT
slot and a PC card slot where you insert their card. Strictly
speaking, it's a bridge.
DanMKohn: I'm suggesting that some of those boxes need to have
the accounting and authorixation function built in.
rohitkhare: I've recently been envisioning the paper I want to
write for Sigcomm, my real 'debut'.
DanMKohn: I don't want to pass your traffic if you won't
reimburse me or aren't willing to pay my prices.
rohitkhare: The Internet Transfer Layer (7.5)

DanMKohn: I don't get it.
rohitkhare: OK: so what if you have an 802.11 hub with an
auctioning authorization network, I see
DanMKohn: Bridge, but yeah.
rohitkhare: The point of that paper would be to define the
transfer layer in one place for the first time. Just as the cerf/kahn
TCP/IP paper established packets and streams, I'd like to define
messages and paid-queues
rohitkhare: Kind of a 'summa'
rohitkhare: Now, walking the cat back and asking if a bridge can
make money at a packet-flow level...
rohitkhare: Well, I'm still thinking a little further down the
pike, when there's a few hundred gigs of storage in that hub, and it
really acts like a store-and-forward point rather than an IP
streaming point
DanMKohn: Hmmm, you've spent too long in academia.
rohitkhare: That's why I like it here 

DanMKohn: You're interested in the status of a breakthrough
paper rather than the status (and money) of a breakthrough company.
rohitkhare: Oh, that should be terribly clear by now. The number
of company ideas passed up...
rohitkhare: No, I really want the fame and the respect. Money is
nice, but I suppose I'll have 'enough'

rohitkhare: Seriously, what do you envision in this space? Who's
the target user today and in one year?

...

DanMKohn: The thing lacking to fund infrastructure buildout
seems to be the AAA capability.

rohitkhare: And let's go with the control hypothesis -- why not
assume AAA will succeed by then. We're back to the better question of
*who's* incentivized to put in a public 802.11 infrastructure for
such a smallish initial user base
rohitkhare: Is the public infra partner going to do it because 1)
there's a real, wealthy early adopter market 2) there's a brand
cachet in leading the market or 3) there's easy money to be had to
build it out anyway in the bond market

DanMKohn: I'd try to get financial backing from Cisco, Lucent,
and maybe MSFT, and then quickly take the company public and lever up
on high-yield. It is the kind of thing Craig could do, or a KP-like
VC. Unfortunately, we're a little busy.
rohitkhare: So, a financial shot in the dark
DanMKohn: That's exactly the problem. But, I do see the
argument that this is an essential exercise in branding, something
I've never really liked or understood.
DanMKohn: No, get investment from the companies that would get
the advantage from success in order to break through the chicken and
egg problem.
rohitkhare: See, one reason I'm still in academia is a dose of
timidity. I'd keep punting this proposition back to a big planetary
commco. I don't see it as an independent company.
rohitkhare: What I'm looking for as a consumer is a big, warm,
happy, IP bear-hug. Guaranteed access, anywhere, anytimne, anyhow.
Whether it's telephone, cable, 802.11, ricochet -- someone to roll
all my access possibilities into one
rohitkhare: Tangent: I wonder if an all-in-one 802.11 telephony
handset is any more or less silly than the first portable MP3 player?

DanMKohn: My view of "late capitalism" (a phrase I saw in the
NYT which I just love for it's faux historicism), is that big
companies (e.g., Cisco) have decided it's more economic to outsource
most of their R&D and new venture development to VC, and then just
buy the winners.
rohitkhare: On the other hand, any kind of campus use might
saturate the link?

DanMKohn: Hard to imagine that Lucent doesn't already have a prototype.
rohitkhare: I definitely buy the outsource-the-R&D angle.
rohitkhare: Better question: Nokia?
DanMKohn: No, I think it works like cellular that you just add
hubs to get more capacity.
DanMKohn: Right. I bet KP would fund this sort of thing.
rohitkhare: Well, and as far as pure voice goes, there's still a
heck of a lot of 32-64kbps channels there
rohitkhare: But still, let's at least name a pilot market scenario: where?
DanMKohn: 12 kbps gives you PCM-like quality with modern
codecs. That's even more.
rohitkhare: Although, given your R&D angle, here's another arc:
we never actually build it, just demonstrate and do the system
integration so some national commco *can* just buy out the sw AAA
solution and roll out a few thousand access points.
rohitkhare: Co-brand it with Apple's Airport 
DanMKohn: At SeaTac airport, you can rent a portable DVD player
and return it at the next airport. You could similarly rent out
and/or sell the PC Cards after installing the hubs at the top 10
airports.
DanMKohn: ;=)
rohitkhare: Yeah, I saw their dvd kiosk in MSP.
DanMKohn: I wonder if the airports would let you install the
stuff for no charge if you let them use it for free.
rohitkhare: Eventually, 802.11 could be like those empower kits
rohitkhare: BUT -- the techsupport hell! -- eek
rohitkhare: Dealing with airports is another good reason to
partner with someone who knows that real estate -- quickAID, Sodexho
Marriott, an airline.
DanMKohn: Actually, it's much better with Win98 and W2K. It's
only NT 4.0 on laptops that is true hell.
rohitkhare: Phase 1 is the prototype; if we can line up funding
O(10^8)$, then all we need is shoe leather to install them "free" for
now.
rohitkhare: Of course, we don't even need to solve the AAA
problem at the point of presence.
DanMKohn: Right, which is why my original suggestion was to
create a technology company rather than an infrastructure company
(sell denim jeans to the miners rather than pan for gold)
rohitkhare: After all, we're not really doing a munch-net,
decentralized and all; we're doing another Ricochet or ISP play,
backhauling all the traffic home to be accounted, retunneled, etc
rohitkhare: But then, the paradox is: there ISN'T any R&D here.
It's just execution

DanMKohn: That's D.
rohitkhare: Well, ok. Us academics spell D with an 'R' 
rohitkhare: And as soon as it's an execution play, well, it's branding.
rohitkhare: So, seriously, which national or wannabe ISP/commco
would benefit most from this $50M 'advertising campaign'?