Wimax vs. WiFi: WiFi is the inheritor to Ethernet's Manifest Destiny
Robert Berger writes that WiMax and 802.16 may be eclipsed by near-term 802.11 development: Robert is a veteran of several industries, moving from digital video to Internet to wireless, having spent the last five years thinking about and building wireless systems and companies. He now consults in the industry through his firm Internet Bandwidth Development, LLC. Robert has deep thoughts that come from this experience, and I wanted to pass on (with his permission) an email he sent to Dave Farber's IP list today.
Robert writes:
I have been involved in these realms for the last 4+ years both in the hardware manufacturer and service provider realms. Here is my opinionated, but educated perspective on the WiMax vs Wi-Fi debate:
At this point in time, WiMax/802.16 is another Zero Billion Dollar industry. There are no WiMax Products today. There will be some WiMax products within the next 3 - 6 months, but they will be first generation and far from the promises that the WiMax forum has been promising. Wi-Fi/802.11 chipsets are already up the learning curve by several generations. Wi-Fi chipsets are already shipping in the high 10's of millions / year.
IMHO, 802.11 is recapitulating the evolution of Ethernet into the Wireless realms.
Ethernet was originally considered a "toy" technology by many of the industry leaders of the time. The manly technologies at that time were first Token Ring, then 802.12 AnyLAN VG, then ATM.
Wi-Fi is currently considered useful only for the home and some enterprise applications and a "toy" for outdoor Municipal Networks.
But Ethernet out evolved and kept delivering just enough functionality, at much lower cost than the too sophisticated QoS laden and expensive "heavyweights".
Wi-Fi/802.11 has taken on the mantle of Ethernet's Manifest Destiny (it uses almost exactly the same packet frame as Ethernet) and brings it into the wireless realms. There are many more companies, universities and hackers pushing the boundaries of what 802.11 can do and the volume is growing at an accelerating pace.
Today 802.11 is at a similar phase of evolution as early Ethernet was when there were only a shared contention medium via hubs and bridges. Ethernet really took off when switches became available and allowed the contention realm to be broken up to support parallel data flows. And that is what we can expect in the next stage of 802.11 evolution. This is what is needed to make mesh wireless networks viable with 802.11. There are already several companies developing mesh (though only a few are doing it in a way that will scale). There is also an 802.11s working group developing a standard for wireless mesh. And mesh is what will allow 802.11 to eventually cover municipal areas.
WiMax hype is extremely misleading. You hear that a WiMax basestation can create coverage of 35 - 70 miles, deliver 50 Mbps, will work in unlicensed and licensed frequencies, can deliver Non Line Of Sight (NLOS) through trees and buildings, will support mobility and CPE built into Laptops.
But this hype is misleading because they mush together all the claims for all the different frequencies from 2 GHz to 10 GHz, licensed and unlicensed, and projections of their roadmap for the next 8 years.
If you compare WiMax using the same 5.8 GHz unlicensed frequencies that 802.11a would use, there may be only 3 or 4 dB link budget advantage of WiMax over 802.11a. (I.E, the link budget is the total of receiver sensitivity and transmitter power, less losses between the two end points, thus it represents the distance that can be covered and/or penetration thru obstructions. So WiMax can deliver a link budget that is at most twice as good as 802.11a, and in the scope of things this is not very much compared to the total link budgets used in outdoor links).
If you say, ok, lets use licensed spectrum, then you can get long distances OR NLOS. If you really want to deliver multi MBps and be able to use laptops inside buildings as CPE, you'll still need microcell sites on the scale of 1- or 2-mile radius of coverage and use multiple WATTS of power. WiMax uses sophisticated base stations and relatively dump CPE. So each micro-cell basestation would be relatively expensive (compared to 802.11, but definitely cheaper than cellphone basestations).
AND you would have to buy the spectrum to create the coverage. At this point in time, in the US, the only spectrum that has half decent propagation characteristics and is available for this application in big enough chunks to be useful is the 2.5 GHz MMDS frequencies. These are already owned by primarily 3 corporations, plus a bunch of educational institutions (the later still holding on to it for "educational" distance learning TV).
So there is a customer base of maybe a handful of companies to buy and buildout licensed networks. Two of the license owners failed already in building out an MMDS network, the third is a "new" company, Clearwire, who bought spectrum from Worldcom. This does not represent a robust marketplace needed to drive a rapidly evolving technology. Its more like a legacy telco marketplace that will have to compete against DSL and Cable Modem in the urban/suburban markets that represent the bulk of the potential end user marketplace. It will not be subsidized by a parallel home / enterprise networking marketplace as will 802.11.
Finally, the WiMax industry has (in terms of active, as opposed to paper members) one giant company, Intel, and scores of small, mostly barely surviving wireless equipment companies that had already spent most of their efforts on proprietary LMDS or MMDS technology and then threw their hats into WiMax as a way to try to keep going. Most of these companies plan to offer proprietary enhancements to their WiMax products to "differentiate" from the competitors. So there are already way too many companies involved in WiMax than there will be demand for their products. So we can expect that when the hype dies down most of the companies will fail.
Sometime in the near future, I would expect that Intel will drop out of most activity with WiMax. They will realize that they need to get back to their "knitting" as AMD is challenging their core business and that there is never going to be the kind of volume in WiMax chipsets that is needed to keep Intel's interest.
There are a few WiMax companies, that will do very well for themselves. Companies such as Alvarian, who are already a leader in the outdoor, wide area wireless network equipment even before WiMax, who understand the market and have the distribution channel / customer base. This niche will grow with the lower costs for this style of rural and Multiple Business Unit (MDU) type network buildouts that can afford the price points that WiMax will end up with. But it will not be a mass market.
In conclusion, Wi-Fi will out evolve and deliver connectivity at costs dramatically lower than WiMax. WiMax / 802.16 is just starting on its path to evolution, has a much smaller base of innovators and chipset growth volume. Wi-Fi is already far along on its core learning curve, has an easy order of magnitude larger base of innovators / investors and chipset growth volume. WiMax hype will sputter out to reality of a niche backhaul and rural marketplace, Wi-Fi/802.11 will evolve and grow into many more realms and dominate the Local Area Network (LAN) / Neighborhood Area Network (NAN) / Metro Area Network (MAN).
Mr. Berger's piece offers many interesting points, and is well-argued given the information to which he has access. I would offer for consideration a handful of facts we've gathered through our inside-the-industry work in broadband wireless that will make his picture less clearly black and white:
(1) Operators who have and/or are building brands the mass market trusts are uncomfortable with large-scale investments in unlicensed spectrum, so the broadband wireless market is more likely to evolve in the licensed bands worldwide. Municipal governments without quality-of-service constraints from demanding enterprise customers may feel differently, but we don't see the commercial operator community buying into the unlicensed story just yet.
(2) In markets where licensed-spectrum, ubiquitious broadband services that support seamless user mobility are now available, public WiFi network efforts are scaled back (witness Optus comments on their hotspot strategy after the launch of Personal Broadband Australia in Sydney)
(3) The WiMAX supply side includes not just the small players who got their starts in fixed wireless, but a veritable who's who of the world's largest telecom equipment manufacturers, and I assure you the pace of innovation in this community is intense
(4) The WiMAX demand side will include (even if not publicly evident yet) a veritable who's who of the world's wired and wireless telecom carriers, who are all looking at ways to create true strategic advantage now that it's clear 3G technologies cannot support true broadband services economically, and in some markets (e.g. Korea) now that fixed wired access has reached high penetration and revenue growth has tapered off.
(5) Meanwhile, there are actually quite viable "pre-WiMAX" solutions that will help operators build macrocellular businesses offering fixed and mobile wide-area broadband service in the next couple years and that will give mesh a solid run for its money. These technologies have a more than 10x advantage in site density that has huge operational advantages well in excess of benefits perceived to accrue to WiFi through its cheaper access point and modem gear (anyone remember Cometa?)
I could go on. In the end, I'd stop short of saying anything like "Wide-area WiFi is dead" -- but I would suggest that scenario-based thinking about the future in this case is worth considering -- there are lots of possible outcomes, the most likely of which is a healthy mix of technologies in the market much as has been the case in the wide-area wireless world since it got its start 30 years ago. It may make for less exciting tribal-warfare headlines, but that's the nature of this industry...
Cheers,
Steven Glapa
Director of Marketing
ArrayComm, Inc.
This is an issue of perspective and the scope of WiMAX/802.16.
From one perspective, WBB is a bottoms up phenomena with characteristics of the "PC Revolution". 'Toy' PCs begot the client-server environment that dominates much of computing today. And that was because of the underlying revolution in ICs and software. This was successful was because it increased personal and group productivity... the same corollary can be seen in WiFi. WiFi was not able to occur until GHz silicon ICs were available.
From the other perspective, WiFi has been a failure in many regards: WiFi has been fast road to commodity pricing. While cheap devices and ideas of nearly free Internet access and ptp networking are exciting for much the same reasons cheap PCs were exciting, the need is for wider spread and ore reliable access more like cellular service than patch work WiFi model.
IEEE 802.16/WiMAX standards started out with carrier class systems in mind. It solves problems needed to take WBB from use in local networks and from congested spectrum applications and provide a reliable, cost effective framework upon which widespread use of compatible systems can evolve. WiFi is limited in technology but also in scope of application... there is far too little commercial motivation to make it ubiquitous and reliable. WiFi dreamers should check their own 'hype meters'.
Some of the criticism of WiMAX is justified. Particularly that aimed at debunking the hype that insinuates that it will be capable of 30 or more miles of coverage with each user having 10 to 70 Mbps bandwidth. Touche!
But misunderstanding and hype is part of the landscape and detractors should not be so quick to throw the baby out with the bath water. WiMAX is not a stake in the ground but a framework for the evolution of technologies. Conceived as such, it stands apart from 802.11 that was conceived to solve more immediate goals with much less regard for how technologies would likely evolve.. and how to plumb for them.
WiMAX won't be fully up to speed until 2006-2007. Before then it will be introduced as fixed and then nomadic (self-contained antenna) CPEs (customer premises units) before it is available in mobile CPE and embedded products. No, WiMAX isn't here in widespread mass markets today. But remember that it took 802.11 over three years from the date the standard was approved to when it became a mass market success. 802.16-2004 was approved last April and the mobile version, 802.16e is scheduled to be approved next March. Both the hype and the criticism are premature.
But why squabble? Both are part of larger enabling trends.
I do agree that 802.16 will have a roll in building out networks, but
I expect it to be small compared to 802.11. I don't expect it to ever
reach the scale and scope that the WiMax Hype portends and I don't
expect it to ever reach the later milestones of the WiMax roadmap
before the rapidly evolving 802.11 based tech fills out the market
niches targeted by 802.16.
Most of the comments supporting 802.16 sounds just like the comments
made by the "Enterprise" proponents of 802.12 or the "carrier grade"
proponents of ATM against Ethernet.
802.12 VG-AnyLan, for those of you who might not have been around in
the early years of physical networks (1994), was promoted as the
replacement for Ethernet with VG-AnyLan offering "advanced QoS
features making it more suitable for Enterprise applications"
The types of claims for 802.12 besting Ethernet even sound similar to the
types of claims for 802.16 over 802.11.
Here is a historical FAQ on 802.12 VG-AnyLAN that was supposed to blow
away Ethernet:
-----------
The 802.12 standard for 100 VG-AnyLAN allows for a backbone supporting both
the 802.3 frames and the 802.5 frames. This means that an existing
enterprise network with both token ring, ethernet, and some central backbone
can easily migrate to the 100 VG-AnyLAN environment. This is due to the
diverse media architecture this new technology can utilize: Cat. 3,4,&5 four
pair UTP, Cat. 2 two pair STP, and single/multimode optical fiber. Meaning
that if there is an existing FDDI, token ring, or 10baseT backbone in place
all that need be done is simply replace the endpoints (router or HUB
blades), connect the 100 VG-AnyLAN repeaters together, and voila a network
structure based on a high speed new technology.
Highlights
# Support for those applications demanding a not only high bandwidth, but
that are also time sensitive (this is due to the media access method called
demand priority)
# Adapt legacy ethernet and tokenring networks to a high speed backbone with
great ease because nodes with 100 VG adapters can be configured to transmit
either tokenring or ethernet
# Extremely expandable when compared to tokenring, and all forms of ethernet
# Maximum network diameter 8000 meters
# Cascading up to five levels
------------------
(from http://www.eece.unm.edu/faculty/rjordan/440/ether/100VG.html)
Here's an obituary from a 100VG AnyLan FAQ
(http://www.io.com/~richardr/vg/)
Hi! Welcome to V1.2 of Richard's Unofficial 100VG AnyLan Web FAQ! This
substance of this FAQ was last updated on Sunday, January 28, 1997.
January, 2001: At one time, 100VG AnyLan was a very promising technology.
However, due to market forces (Fast Ethernet slaughtered it in the market),
VG is a dead technology. To my knowledge, there no currently no VG products
for sale.
--------------
I suspect we will be seeing something similar for 802.16 (maybe not
the "no products for sale", but definately that 802.16 gets relegated to a
very small niche) when 802.11 will have new PHYs
with 100's of Mbps throughput and a mesh "switch fabric in the air"
in chipsets that cost under $10 or $20.
The article gives some good points. However, in emerging markets outside of the US I think WiMAX has a much greater opportunity. For instance many South American and African countries do not have a good infrastructure for telecom yet. Since WiMAX has the capability to be used for both wireless VoIP and "last mile" data, there is room for WiMAX to take the place of POTS and cellphone infrastructure as well as provide data backhaul.