WiMax: How Far? How Fast?

One of the big questions about WiMax is exactly what kind of data transfer speeds and range the products -- based on the 802.16d fixed wireless standard -- will actually offer.

WiMax is a metropolitan area networking technology, which is seen by some as the savior of the fixed wireless industry, because it will allow them to reduce costs, by sourcing standardized components from vendors, and offer carriers interoperable infrastructure kit.

The initial fixed wireless 802.16d WiMax spec has just been ratified (see WiMax Spec Ratified). A mobile variant, called (surprise!) 802.16e, is expected to be ratified in 2005.

According to the coming edition of our monthly paid research report, Unstrung Insider, the answer is still a moving target, but performance may in fact lag behind the proprietary systems currently offered by vendors of broadband wireless access (BWA -- a.k.a. fixed wireless).

"The term WiMax is a catch-all phrase that hides a huge range of applications and variations in how the spec is implemented," says Gabriel Brown, chief analyst for Unstrung Insider. "So asking a vendor how their WiMax system will perform is like asking, 'How long is a piece of string?' " Nevertheless, Brown has dug up a chart from Intel Corp. (Nasdaq: INTC) that attempts to offer some of the speeds-and-feeds questions about WiMax performance:

Table 1: WiMax System Performance
Range < 4 miles 4-6 miles > 6 miles
Base-station cost (�04 pricing) $5k - $20k for WISP class $20k+ for carrier same same
CPE price < $300 same same
Adaptive modulation scheme 64 QAM 16 QAM � QPSK up to 16 QAM
Data throughput (20 MHz channel*) 75 Mbit/s 50 Mbit/s 17 Mbit/s to 50 Mbit/s depending on link quality
No. of business users (T1 level) 1 206 138 46 to 138
No. of residential users (512 kbit/s) 2 1,552 1,035 345 to 1,035
Source: Intel
Assumes two 10MHz bands in the base station as benchmark for comparison purposes. Over-subscription rate is 5x for business and 12.5x for residential. Also takes into account overhead (efficiency), which for 802.16 is 85% independent of number of users.

But -- even as the general industry is backing WiMax -- some vendors are claiming that the standard won't be able to offer the kind of performance that their proprietary systems can.

Allan Klein, vice president of technology at SR Telecom Inc. (Toronto: SRX), says in the Insider report: "“We’ve been saying for sometime that basic WiMax won’t do the trick. To get the performance in non-line-of-site that we get today, you’ll have to implement the optional extensions in WiMax that aren’t part of the basic profiles." — Dan Jones, Site Editor, Unstrung

WiMax: Going the Distance will soon be available as part of an annual subscription (12 monthly issues) to Unstrung Insider, priced at $1,350. Individual reports are available for $900.
ZoomB 12/5/2012 | 1:21:29 AM
re: WiMax: How Far? How Fast? Can anyone here back the numbers quoted in this article and the unstrung insider report. I have been doing some research and some interesting papers have popped up regarding 802.16 NLOS performance i.e. max distance is only around ~100 Meters.

In a LOS system the distance and rates are comparable to ones shown in the article.

My understanding is that 802.16d was supposed to solve the NLOS problem. does it? if so, can anyone point to some data / simulations etc.

The IEEE paper I am looking at is

"Coverage Prediction of New Elements of Systems beyond 3G: The IEEE 802.16 System as a Case study" By G. Plitsis
aschuh 12/5/2012 | 1:21:13 AM
re: WiMax: How Far? How Fast? Thanks for the reference. This paper is fine, but the bottom line -- 2.1 Km top cell radius when allowing for maximum throughput -- even if hardly surprising, is a far cry from vendor's own estimates.

As it happens, your question is currently a major concern of mine, since I'm preparing an opportunity assessment of WiMAX. Not surpringly, capital expenditures are the critical variable, and cell radius/capacity is main component here. Vendor estimates, besides being typically vaguish, are also scarce. By and large, only the "4 to 6 miles typical radius" is mentioned in product specs. In addition to the estimate by Intel published by Unstrung, the other major references I've been able to dig up are:

- Redline: (http://www.dailywireless.org/m..., suggesting a typical bandwidth of 24-56 Mbps per sector at distances between 3 and 12 Km, NLOS, under a broad range of propagation conditions at 2.7 GHz.

- Swisscom: a report of their trials, by Daniel Rodellar, available upon request, that shows that a pre-standard 802.16a at 3.5 GHz would behave a lot like the older WLL systems in the same frequency.

It is still quite uncertain, and we are all fog navigating. However, a lot can and will change, so we cannot afford to throw away the baby with the bathwater. A lot of implementation-specific elements -- MIMO high-gain antennas, error correction schemes, sub-channel aggregation and QoS management -- will also impact end user perception of performance, with consequent ability to get market share. At the end of the day, trials with real equipment under actual conditions, controlling for vegetation/construction loss, humidity, tower placement will be the only reliable source of data.

Hoping we can continue sharing information, best regards.

apaterra 12/5/2012 | 1:21:02 AM
re: WiMax: How Far? How Fast? I have seen the pre-802.16 gear (infrastructure variant for ad-hoc comms) do 5-10 miles complete NLOS (using omni antennas too). My memory tells me that a(infrastructure/base station), d, and e(mobility) all got re-wrapped under 802.16-2004 to make the alphabet soup less confusing though don't quote me on that.

Also it depends on the environment for the distance performance. Longer NLOS scenarios are going to come from downtown metropolitan areas vs. standing on different sides of a mountain in the dessert or in a forest of trees.
rtfm 12/5/2012 | 1:20:51 AM
re: WiMax: How Far? How Fast? The NLOS is key to profitability (along with CPE costs), since truck rolls and external antennae are a non-starter, except in niche markets.

Non-Line of sight (NLOS), not to be confused with near-line of sight (NLOS!), is not well measured. There are formulae that modify the free-space loss (x^n, where n=2 in free space) to n=2.5-4. OFDM can help with certain kinds of interference, esp. multi-path. BUT, everyone I've talked to says hard reflections (urban buildings) are much easier to deal with than forests or trees (soft reflections/interference).

If someone can give me a product that will do 20 km NLOS (non-LOS), at high bitrates within a 20 MHz band (and I am not too fussy on power levels, within reasonable international limits) then we might have something...


p.s. I have repeatedly heard proprietary solutions to beat standard ones thus far. But standards will give us volume = lower prices? What I don't mind paying for from vendors is extra money for extra goodies not in the standard, e.g., MIMO.

Sign In