I would urge readers to drop the baby, turn off the oven, sit down and read this MIT paper on viral networking.
In a nutshell, it describes the future of mesh networks. There are two core results:
This is the E=mc2 of communications. It means that fibre to the home and so on are just icing on the cake. The lower bound for the future of connectivity is going to be damned high wherever humans or their powered objects congregate.
It means the end game is already pre-determined. Centralised telecom won’t exist in its current form. Don’t hold long-dated bonds in network operators or their equipment suppliers.
The caveat? Getting this into reality is, as they say, non-trivial. You have to make it scale in a world where bad actors may be at play. You have to get all the non-functional stuff right, like battery life. We could be talking anything from years to decades. It’s as big a jump as E=mc2 to the atom bomb — the Manhattan Project of communications.
But the theory is rock solid, and the future inevitable. You’ve been warned.
Posted by Martin Geddes at 11:23 PMTrackBack URL for this entry:
http://www.telepocalypse.net/cgi-sys/cgiwrap/mgeddes/MT/mt-tb.cgi/387.
Listed below are links to weblogs that reference A fatal infection:
»
Lippman and Reed on Viral Networking from McGee's Musings
Let me add my recommendation to Martin's and add that the paper is authored by David Reed and Andy Lippman of MIT.
[Read more]
»
What's an Access Provider to do? from Cox Crow
Jim McGee points out Martin Geddes's pointing out a paper from 2003 by Andrew Lippman and David Reed on "Viral Communications." Mr. Geddes says,
It means the end game is already pre-determined. Centralised telecom won’t exist in its current fo...
[Read more]
»
Today's Jots posts from Joe's Space
»
Telepocalypse by Martin Geddes: A fatal infection from GungHaggisFatChoy
From Telepocalypse by Martin Geddes: A fatal infection.:
QUOTE
I would urge readers to drop the baby, turn off ...[Read more] Tracked on January 31, 2005 10:02 PM
»
MIT viral networking paper is E=mc2 of communications - Martin Geddes from North American Bandwidth News
[Read more]
Getting this into reality is non-trivial... ?
One of the key stumbling blocks just fell - Atheros, the most flexible of all Wi-Fi chipset vendors, is now shipping a single-chip dual-band DSSS/OFDM chipset. Add a dash of MIMO (now shipping) for "intelligent antennas" that will largely fix the 2.4 GHz situation(s) and 500+ MHz of 5 GHz spectrum that can be used for backhaul and redundant interconnections on the mesh, that's enough to ignite this revolution, and the addition of license-exempt use of television spectrum will only accelerate what is already a trend.
Martin,
I have been reading for a while now, and I really appreciate the insights.
I did want to point something out I've been noticing for a while, although it may not be relevant to the article above. I've noticed an increasing trend at work and play of people taking advantage of web clients. Both Lotus and Microsoft offer web clients for their software when connecting, and I see an increasing number of people using them. (almost in preference to the real thing!) My friends increasingly use web storage, over half now use web book marking, while archiving photos for viewing on the web is common.
My point is that I am seeing successful applications of the webclient, and I wouldn't be surprised if it's pushing intelligence away from the edge . Firefox is a very small download, but it can do the majority of what's needed. The IMac Mini is an example of the internet toaster to me, it and products like it commoditize the edge, while services like the webclient may start to rise in value.
Just my 2 cents. Btw... good paper to post. (to get slightly back on topic :) )
John.
Posted by: at January 29, 2005 12:46 PMto john:
Having a mesh doesn't mean you can't have centralized client-server services. The mesh will make the distribution of the data form at the edges, while the actual data can be both an edge and a 'center'.
For instance, now you have say an ASP connected to their ISP offer a web-based app and keep all the data on their servers; you launch the app in your browser while you're connected to your ISP.
In a mesh-based future, the ASP wouldn't be connected to an ISP and neither would you, they would just make their service available and both of you will be connected through some path over the mesh. The actual service would be the same but how it gets from one place to the other shifts to the edges.
Posted by: at January 29, 2005 09:13 PMconfused: i don't see anything about latency or anything particularly explicit about thru-put in the paper linked.
Posted by: at January 31, 2005 01:11 AMMy apologies; the latency bit was in some surrounding private communications and I forgot it wasn't in the paper itself. Just take it from me, the same horse says that latency isn't forced upon us in a mesh network as a result of physics. We can build layer 1/2 repeaters that do the job. It's a technology implementation problem, not an electromagnetic physics problem.
Posted by: at January 31, 2005 03:43 PMMartin,
I wanted to provide my hypothesis on Mesh networking overcoming the "non-trivial" pursuit of reality. My contention is that our current military actions will lead to a commercial boom of Mesh technology. As a member of the 1996 commercial launch of Sprint's CDMA systems, I am well aware that the primary source of technical talent were veterans of the first Gulf War.
The mesh veterans of the current war will rotate back into a market that does not require expensive spectrum, regulation or significant G&A investment to prop up a technology-based business. I predict an impact on communications geometrically larger than we saw with CDMA 10 years ago.
Posted by: at March 22, 2005 04:00 AM