Today comes the well-trailed news that Nextel are deploying a high-speed wireless WAN network, presumably based on Flarion's FLASH-OFDM technology. (Nods to /. and Techdirt. More analysis also at Reiter's weblog.)
As I said last week, 802.20 will be really important. Why? Well, as one well-informed colleage reminds me, 802.16 (aka WiMax) doesn't have doppler tolerance built in from the start. That means it stops working when you start moving. It is essentially impossible to build this capability in later without starting again from scratch on new coding, silicon, FPGAs etc. Some limited hacks may enable a bit of motion tolerance. Flarion's technology is converging with the 802.20 standard, which does have doppler tolerance from the beginning.
I've had a demo of a Flarion network. I sat in the back of a van driving at 70mph down the I435 watching someone play networked Quake. I've never felt such intense motion sickness so quickly! But the demo didn't just make me want to reach for the barf-bag -- it blew my socks off too. They built a low-latency push-to-talk application in ... 2 days. Most carriers have taken years to achieve this. Nextel's 2G Direct Connect technology, which is wildly successful in the US, took years to perfect. New competitors still haven't reached that bar. With Flarion, it's a few hours playing with your SIP toolkit in your favorite IDE.
This announcement is interesting for other reasons. The acronym gives it away: FLASH-OFDM stands for Fast Low-latency Access with Seamless Handoff Orthogonal Frequency Division Multiplexing. Let's digest that piece at a time, and not eat the whole cow at once.
The "fast" bit is the easy one. Some modest knowledge of information theory and physics tell you how much you can pump through a slice of spectrum between two points. FLASH-OFDM is efficient in its encoding and handling of multi-path issues, so it comes out well on any speed test.
Traditional 3G cellular networks establish a virtual circuit to each handset. To communicate, a signalling channel is used to bring the circuit up and tear it down. This can take a long time -- 10 seconds or more under load. The channel is also typically torn down after a timeout following a period of inactivity. So there is a lot of latency in establishing and re-establishing a channel. Then even once you have a channel, a queueing algorithm determines who gets to speak when. The distance you lie from the tower and amount of interference can affect how often you get a chance to speak. If you're unlucky, it might be too long for you to win at Quake against wired competitors. FLASH-OFDM is pretty smart about queueing and also enables QoS for IP communications. (I'll write about QoS another day and when and why it matters.)
As the Flarion marketing department puts it:
Third generation (3G) mobile networks, on the other hand, retain a circuit-switched, hierarchical architecture. Consequently there is tension between the design objectives and the current environment of the wired Internet and mobile voice networks. The resulting design compromises of circuit-switched networks, which are optimized for voice, impair their ability to deliver high-speed, low-latency data cost effectively. The resulting high cost-per-megabyte of data delivery over circuit-switched based networks will prevent the emergence of mass-market wireless Internet access. An alternative approach, focusing directly on high speed, low cost and low latency wireless data delivery is required. Flarion, through its innovative FLASH-OFDM airlink, addresses the challenge of delivering affordable mobile broadband.
The worst-case latency for a packet transmission on a Flarion network is about 5ms. In other words, about 200 times better than a CDMA 1xRTT network, typically found in the US, Korea or Japan.
The usual figures for data throughput you see for a CDMA or GSM/EDGE/UMTS network are maximum burst or sustained throughput. But most people don't spend their lives doing FTPs. Instead they are using chatty, bursty applications. So the apparent spectral efficiency of 3G networks is a mirage. Your true throughput is terrible, because you spend all your life setting up and tearing down channels, and waiting for your one time-critical ACK packet to be sent. For instance, see this paper on page 5 for how throughput of a TCP/IP links varies with latency.
It's the throughput of the complete system loaded with users that counts, not an artificial network test with a single laptop under the tower at 3am. Peak throughput is irrelevant.
You don't have to be a genius to see that low-latency plus high-bandwidth looks like a tasty recipe for next-generation IP-based voice apps. Voice is still the "killer app". But it is evolving once freed from the clammy dead hand of circuit telephony. Should enterprises start demanding end-to-end encypted voice, then Flarion can deliver it using off-the-shelf technology. Integrate presence, IM and voice a-la Skype? No problem. Anyone left with a faux-circuit network will be left spluttering.
Seamless hand-off matters for voice calls. WiFi is weak in this respect (today) because authentication and hand-off is slow. Plenty of people are working on fixing this. Dropped calls (or the perception of a dropped call) are the #1 driver of customer (dis)satisfaction with wireless networks. This is a "must-have" non-trivial feature.
So that was low latency and seamless hand-off. There was a bit more to the acronym soup, though. That OFDM stuff -- what does it mean? Well, CDMA (the basis for 3G) has some unfortunate disadvantages. It needs complex and expensive power modulation circuitry to keep everyone shouting at the same volume. The cell site coverage shrinks as the system becomes loaded (called "cell breathing"). The coding isn't built for IP communications -- it was all done for voice, with packet as an afterthought.
ODFM is generally a more tolerant technology, even if it is a bit heavy on the signal processing (lots of matrix algebra, folks!). For instance, those orthogonal frequencies make power management of adjacent multi-path signals less of an issue. To quote the Flarion marketroids again:
Since all conventional cellular wireless systems, including 3G, were fundamentally designed for circuit switched voice, they were designed and optimized primarily at the physical layer. The choice of CDMA as the physical layer multiple access technology was also dictated by voice requirements. FLASH-OFDM, on the other hand, is a packet switched designed for data and is optimized across the physical, MAC, link and network layers. The choice of OFDM as the multiple access technology is based not just on physical layer considerations but also on MAC, link and network layer requirements.
So we have here a co-ordinated set of changes at layers 1 and 2 of the protocol stack. What's fascinating is that this destroys the last refuge of the circuit carriers from business model change. Changes at the lowest technology levels ripple all the way up to the business model for the application layer. Wireless is not a safe harbor.
Thus far we have seen gradual but unstoppable substitution of wireless minutes for wired calling. Within a decade, Flarion and its competitors could potentially drive carriers out of the voice service business entirely. Cable companies and wireless ISPs will be able to give away voice to get you to buy into their pipe. Why? Because an ever-larger proportion of calls become VoIP, and the only significant direct cost is a remote directory look-up for a SIP routing function. Off-net PSTN interconnect shrinks to zero.
If Flarion tweaked their technology to lower the peak throughput and instead give a predictable low-latency 100-200kbps to more users at a lower unit cost, you would have the killer transport for a wireless successor to Skype. Now that would be interesting. Any VCs out there wanting to fund some market disruption?
Posted by Martin Geddes at 5:01 PMTrackBack URL for this entry:
http://www.telepocalypse.net/cgi-sys/cgiwrap/mgeddes/MT/mt-tb.cgi/142
Listed below are links to weblogs that reference The end of wireless as we know it:
»
Blog-Power from Om Malik on Broadband
There has been a lot of coverage of Nextel’s Broadband service and how Flarion was powering the trial networks. I find it amusing that the headlines are coming three days after the news became all the rage in the Blogsphere....
[Read more]
»
802.20 is the end of wireless as we know it? from North American Bandwidth News
[Read more]
»
802.20 is the end of wireless as we know it? from Lance Tracey
The cool thing about 802.20 is that it has "doppler tolerance" which means that unlike the currrent version 802.16, 802.20 ...
[Read more]
»
Wireless Broadband from Von Freud
A Vodafone anunciou o lançamento em Portugal do seu primeiro serviço de UMTS baseado numa placa PCMCIA, que mistura GPRS/UMTS para acesso em portáteis (não sei é quantos macacos conseguem pendurar-se por UMTS na mesma...
[Read more]
»
Skype's Mobile Conference Potential? from Unbound Spiral
Last night I had a mind bender from a friend who said: "With our solution each Skype conferencee (each on a different continent) could be on a cell phone - and each one could be on a different continent to...
[Read more]
»
phones and Telepocalypse by Martin Geddes: The end of wireless as we know it from
FCC steps in as cell phones disconnected - Hurricane katrina noaa hurricane katrina swirls toward the gulf coast.
[Read more]