Hi David,
Great story, however, I don’t think the analogy is quite right (unless I misunderstood it) about "the
fact that two, three or more little sheep eat as much as one big sheep." The impact on Bluetooth devices is not the same if one compares two 20MHz BSSs (say operating in channels 1 and 6) to that of 20/40 MHz BSS transmitting 40 MHz signals (again, channels
1 and 6). If a Bluetooth device is operating in close proximity to a STA that transmits in channel 1, it can still use the frequency channels occupied by the BSS that operates in channel 6 without much impact (of course, it depends on the proximity of the
closest STA/AP that operates in channel 6). On the other hand, if a Bluetooth device is operating in close proximity to a STA that transmits in channel 1&6 (thus, 40 MHz transmission), the chances are much higher that the Bluetooth device will see packet drops
due to interference in channel 6.
We have already seen that 40 MHz in 2.4 GHz band is a problem (either via simulations or lab measurements). Scanning is one option,
while the other option is to not have this option (40 MHz transmission) in 2.4 GHz at all.
Regards,
From: David Bagby [mailto:david.bagby@xxxxxxxx]
Sent: Thursday, October 23, 2008 5:04 PM
To: STDS-802-19@xxxxxxxxxxxxxxxxx
Subject: Re: Fw: [802.19] Possible 802.11n spectrum scanning requirements
May I extend your argument a bit? Let's use the grazing analogy....
Once upon a time there was a pasture called "the 900" - for a while it was a fine pretty green pasture, over time people started to
dislike it as they found it more and more crowded and they saw the grass get thin. Some ranchers adjusted by moving their livestock to a higher pasture they glimpsed in the hills - it was called "the 2.4". Time passed. It became hard to find any ranchers that
remember the days of "the 900".
More time passed and the 2.4 livestock multiplied.... now the ranchers started feeling the "the 2.4" was crowded, they worried about
the grass...
Some remembered that once some sheep ranchers moved some livestock to a new pasture ("the 5"), but for various reasons the flocks did
not prosper and the ranchers longed for the "good ol days of the 2.4". Some ranchers started breading sheep they feel will be more successful grazing in "the 5", while they also keep stock on the plains of "the 2.4".
One or two odd ranchers wondered, if it works, what they will do if "the 5" becomes crowded?.... there are vague stories of other pastures,
but it is said they are hard to get to and the feed is expensive. It's hard to leave home, the many 2.4 flocks are no longer migrant and many ranchers continue using "the 2.4" to grazing their sheep.
As time passed, the ranchers subdivided. Two prominent rancher groups now concentrate on raising different types of sheep (red sheep
and blue sheep), these ranchers also meet sometimes at a periodic gathering called "the 802 sheep rancher social club".
The red sheep ranchers want the blue sheep ranchers to limit the number of the blue sheep in the 2.4 pasture. The argument is that this
will preserve the pasture for all...... Oddly, they don't request limits on the numbers of blue sheep (the red sheep would not want sheep limits either). No one talks about the fact that two, three or more little sheep eat as much as one big sheep.
Both the red sheep and the blue sheep ranchers tend to think the only type of livestock are sheep of the colors they raise... they get
all upset about red and blue sheep and argue over the grass in the northwest corner of the pasture - they don't like to look around at the rest of the pasture... which amazingly enough is filled with not only with some sheep that seem to be colors other than
red or blue (there are stories of some of green sheep with a big "Z" brand on them) but also a whole heck of a lot of cows.
So the question I raise is: What is the point of an agreement between the red and blue sheep ranchers when they are only a part of the
overall pasture usage?
Suppose they do train the sheep to recognize sheep of another color (perhaps I should have called then Sneetches?) and they even learn
to do tricks when the see a sheep of a different color - what's the point? Would it matter?
(note the blue sheep ranchers have already trained their sheep to flee at a specific whistle signal, and they have let the red sheep
ranchers learn the whistle melody. Yet the red sheep ranchers want more...)
I'm not arguing in favor of either the TGn or the BT side of the entertaining 40MHz channel debate.
(While the sheep ranchers debate, the cows keep eating.)
Fundamentally, I do object to one class of 802 device having to scan for another and avoid "that device's spectrum". That is tantamount
to deciding that the one class of device is more worthy of spectrum use than another. I'd object to that no matter what device is deferring to what other device type. I find the idea that any type of 802 device has any "rights" to ISM spectrum absurd.
So, I'm am asking people to think objectively about this idea of 802 devices scanning for other 802 devices.
But would it really help address the underlying core concerns?
Would it alter the overall usage of the pasture?
What good for the pasture, at what cost to the red/blue ranchers?
I am pretty certain it will sour the mood of the "802 sheep rancher social club" gatherings. It might even make one set of sheep ranchers
feel lofty and politically correct - for a while.
But would it really, actually, make any noticeable difference over time?
I personally suspect the ROI here is not good.
What the sheep ranchers really want is to own some land they can fence in and make their own....
..and that's a different tale (buy me a drink some evening and I might spin the historical fable of "King WinForum and the Apple" )
-----Original Message-----
From: Bill Shvodian [mailto:bill.shvodian@xxxxxxxx]
Sent: Thursday, October 23, 2008 11:16 AM
To: STDS-802-19@xxxxxxxxxxxxxxxxx
Subject: Re: [802.19] Fw: [802.19] Possible 802.11n spectrum scanning requirements
Is everyone familiar with the concept of The Tragedy of the Commons?
http://en.wikipedia.org/wiki/Tragedy_of_the_commons
The idea is that if there is a public resource (like common grazing land) that is free for everyone to use, then individuals will decide it is in their best interest to use more and more of it until eventually it is so overused that it is useless (like overgrazed
land). The analogy could be applied to the ISM band. It is great spectrum and it is free and everyone can use as much as they like as long as they are willing to accept interference. However, as devices gradually are built to use more and more of this resource
at one time, it becomes more and more crowded and eventually unusable for everybody.
We have heard anecdotal evidence that a device transmitting on 40 MHz will transmit for a shorter time and so the net interference will be the same. However, the simulations presented by TI show that the aggregate throughput for the resource is decreased when
40 MHz channels are used. Likewise other technologies are potentially crowed out when 802.11n uses wider bands. And the argument we have heard that 40 MHz is required to enable new applications says that individual devices would be using more of the resource
themselves. This push for using more and more spectrum is analogous to overgrazing making the land less useful for all users.
Limiting 802.11n devices to 20 MHz in the ISM band would go a long way to keeping the ISM ban useful to all users, and not befalling the fate of the Tragedy of the Commons. Alternatively, the TAG has been discussing sensing other technologies to avoid degrading
their performance when they are present. Is it ideal for the 802.11n manufacturer? Of course not. Limiting 802.11n to 20 MHz in the ISM band would be much easier for device manufacturers to implement, but that has so far been rejected. Sensing appears
to be the next best option.
Bill
On Thu, Oct 23, 2008 at 1:59 PM, Ivan Reede <i_reede@xxxxxxxxxxxx> wrote:
I think your comment addresses a fundamental question.
I think you are correct in some ways in your assertion that "nobody owns this spectrum" and "everybody may expect intereference" and "everybody has to live
with it".
However, I remember some time back, not so long ago, where 802.11 was really worried about 802.15.1 devices not following CSMA/CD rules and interfereing with
802.11 operations... worry which from my perception ultmiatly led to the EC froming 802.19 ... the intent being that although "noboy owns the spectrum" we also have "and 802 devices should play nice to each other".
So the question is, eihter "we elect to play nice to each other and really work at implementing mechanisms enhancing playing nice to each other within 802
devices" or we drop all that and just start having fun clobering each-other. Personally, I rather work together.
----- Original Message -----
Sent: Wednesday, October 22, 2008 8:36 PM
Subject: Re: [802.19] Possible 802.11n spectrum
scanning requirements
Dave,
Thanks for your comments.
I would just like to point out that 802.19 is just a TAG and as such is a forum for having technical discussions. The decision on the 802.11n
draft is in the hands of the WG and the WG/Sponsor ballot voters, the EC, RevCom and the SB, not the 802.19 TAG.
The TAG is not setting policy it's a forum for technical debate.
There was a request for such a discussion. Are you suggesting that I tell those who requested such a discussion that the TAG will not allow
for such a technical discussion? I think not.
Regards,
Steve
I honestly do not think that .19 should go down that road.
If it does, I strongly suggest that requirements to scan for other 802 family wireless devices would have to apply to ALL 802 wireless devices.
1) The ISM bands are what they are: ISM band devices are required by law to accept any interference received from other ISM devices.
2) Another factoid is that 802 devices constitute a small portion of the things found in ISM bands. (ref the tutorial given by the hospital guys re what
they saw in ISM bands when they scanned).
3) Independent of if we like it or not, the mixture of signals in the ISM bands is not static - what one accounted for in yesterday's design may or may
not be good enough tomorrow - the band signal content is dynamic.
4) The law has no analogy of a homestead act for spectrum in the ISM bands - Ownership of spectrum is not conveyed by sales of devices.
5) The source of what one's device perceives as "interference" is not really relevant (in that it does not matter if it is from another 802 device or
a non-802 device).
6) Channel width used by a device is also not relevant (100 1MHZ adjacent channels fill the same amount of spectrum as 1 100MHz channel).
My pragmatic conclusion, developed over many years, is that if one wants to play in the ISM bands, one had better be able to operate in the ISM environment,
including accepting the interference one is likely to receive. If one's device can't handle that, don't expect to have a successful product.
SO then I ask why scan for 802 devices?
Presumably because "someone" wants "something" to happen to make their operation "better" when the "other" devices are found....
Who is to say what use of the ISM band is more important that another?
The only consistent answer to those questions I would expect to hear is "mine is more important than yours"; a rat hole argument that can never be "won".
I've observed that people tend to react emotionally along the lines of "just don't interfere with me"....
ISM band reality is that if ISM product operation depends on assumptions that can not be guaranteed in the ISM band, one may not have made a good choice
of spectrum for the product design.
When I consider these points, I wonder what is the benefit of having 802 devices looking for only other 802 devices?
What will they do when they find them?
Who gets out of the way of whom?
based on what objective or criteria?
Having found 802 devices, does it matter given the full extent of devices operating in the ISM band?
Seems to me like a lot of work to address a rather small percentage of the ISM "interference sources".
Suppose 802 did eventually require that all 802 devices look for other 802 devices....
how would 802 keep that updated as new devices are invented?
Cross coupling operational aspects of different 802 standards in that manner would seem to be an enormous complication; and one that I don't see a payback
for. The pace of the 802 standards process pretty much tells me that by the time that "802 family scanning" were standardized, the assumed mixture of devices would be obsolete. And what would old 802 devices do wrt to new ones? they would have no way to know
how to scan for them...
I think it very unwise for .19 to attempt to extend simple "coexistence" (which is not and has never been a synonym for "zero interference interaction")
into "cross 802 wireless standard awareness" or (even more complicated) "dynamic spectrum management between 802 devices".
I suspect that anyone which sticks a toe in that tar pit is unlikely to ever see their toe (or foot or...) again...
-----Original Message-----
From: Shellhammer, Steve [mailto:sshellha@xxxxxxxxxxxx]
Sent: Wednesday, October 22, 2008 3:05 PM
To:
STDS-802-19@xxxxxxxxxxxxxxxxx
Subject: [802.19] Possible 802.11n spectrum scanning requirements
All,
On the 40MHz 11n coexistence conference call it was suggested that the 802.19 TAG start to look at possible Spectrum Scanning Requirements. A proposal was made to
add include an option for spectrum scanning in the standard. The proposal was made by John Barr and can be found at,
https://mentor.ieee.org/802.11/file/08/11-08-1101-04-000n-additional-40-mhz-scanning-proposal.ppt
While the 802.11 WG discusses the merits of this proposal, the 802.19 TAG is a good forum for having technical discussions on possible requirements and technical feasibility.
This information may be useful to 802.11 in making its decision on how to address this proposal. The text on Slide 6 of John's presentation beings to discuss possible requirements. That may be a good place to look at to stimulate thinking on this topic.
During the conference call it was suggested that the primary non-802.11 systems that are of concern are 802.15.1 (Bluetooth) and 802.15.4 (Zigbee) since they both operate
in the 2.4 GHz band.
Anyone who would like to prepare a presentation on possible spectrum scanning requirements for these non-802.11 systems please notify me. We can discuss any such presentations
on the next conference call on November 3 or at the Plenary meeting in Dallas.
Thanks,
Steve
|