Re: Long-Distance Wi-Fi

[David Farber <dave () farber net>]

________________________________________
From: Jim Thompson [jim () netgate com]
Sent: Tuesday, March 25, 2008 7:32 PM
To: David Farber
Subject: Re: [IP] Re:    Long-Distance Wi-Fi

Dave, for IP if you wish:

On Mar 25, 2008, at 1:16 AM, David Farber wrote:
> ________________________________________
> From: Lars Poulsen [lars () beagle-ears com]
> Sent: Tuesday, March 25, 2008 1:25 AM
> To: David Farber
> Subject: Re: [IP] Re:   Long-Distance Wi-Fi
>
> From: Brett Glass [brett () lariat net]
> > What the article fails to mention is that this technology would not
> > be legal to use in the United States due to the FCC's restrictions
> > on effective isotropic radiated power. Our wireless ISP has long
> > advocated that the FCC allow broadband providers to use additional
> > power to reach outlying areas, subject to the use of spectrum
> > etiquettes to avoid interference.
>
> What the article describes should be feasible under FCC's ISM rules
> for
> the 2.4GHz band. You may have overlooked the rule that allows an
> increase in EIRP for point-to-point links with directional antennas
> at both ends.

Pedantically, there is a 2.4GHz ISM band (as well as 900Mhz and 5.7GHz
bands), but there are no "ISM rules for the 2.4GHz band".

> With 24dBi antennas at both ends and 20dBm transmit power, you
> can get 69 miles at -90 dBm sensitivity. (Try the RF link budget
> calculator at http://www.afar.net/rf-link-budget-calculator/ ).

Several problems here:

1) -90dBm sensitivity (@ 8% PER (*)) is unlikely to exist at data
rates above 5.5Mbs (11b, CCK modulation) or 6Mbps (11g OFDM), no
matter how big and power-hungry your LNA is.
(Yes, I understand that many manufacturers quote more.)   I covered
the math on this before on-list in reference to claims made by Tropos:
<http://www.interesting-people.org/archives/interesting-people/200510/msg00264.html
>

2) As the posters have noted, the FCC limits operation using ISM rules
(commonly 15.247) in the 2.4GHz band to an EIRP of 30dBm into a 6dBi
antenna, unless the application is point-to-point.  If you have a
point-to-point application, then you have to back-off transmit power
1dBm for every 3 dBi of antenna gain above the 6 dBi allowed for
multipoint operation.   Thus, for a 24 dBi antenna gain, you're
limited to 30 - (( 24 - 6 ) / 3), or 24 dBm into that 24 dBi antenna,
for an EIRP of 48 dBm.

2) 69 miles is 111 km by my calculation.   Using a line-of-site path
loss model, one encounters "path losses" (in dB) of  32.44 + 20 log
(dist in km) + 20 log (freq in MHz)

range       "path loss":
     1 m     40.04 dB
   10 m     60.04 dB
100 m     80.04 dB
      1 km 100.04 dB
    10 km  120.04 dB
  100 km 140.04 dB

(hopefully your readers will see a pattern here)

and finally,  111 km -- 140.95 dB

So that 48 dBm of EIRP is subject to 141 dBm (minimum!, but we'll get
to that in a moment) of path loss, yielding a signal level of -92 dBm
entering the remote antenna.   The gain of the remote antenna 'raises'
this signal level 24 dB, to -68.95, and perhaps another dB or two is
lost due to coax and connector losses so it enters the radio front-end
at around -70 to -71 dBm.   Sufficient to decode by good margin.

The first problem is that very few "line of site" shots exist, and the
"path loss co-efficient" therefore can't be 2.0, which is reflected in
the "20 log (dist in km) in the above equation.   The reality is that
the "path loss coefficient" is much more likely to be 2.8 - 3.2 in sub-
urban environments, and well above this in urban environments.   When
this happens, the "69 mile" link will no longer work, as the signal
level entering the radio won't be above -98 dBm after coax / connector
losses.

Another problem is that there are many other devices operating in the
2.4GHz ISM band (under ISM rules) and these serve to raise the noise
floor, decaying the ability of a receiver to demodulate signals.

Of course, Intel is targeting rural areas, so they're not fighting
this decay of the noise floor (yet).  Brett and others who attempt to
use WiFi gear to erect blanket coverage of an area are challenged by
the many other devices operating in the same spectrum.   (Heck, my
phone has WiFi now.)

Jim

Obligatory disclosure, my company sells exactly the same hardware as
used in the Intel RCP efforts, (in fact, we supplied Intel with most
of the boards and cases for their development of same), with largely
similar software.  Intel has chosen to keep the changes to the FreeBSD
"ath" driver (removing the 802.11 MAC layer in favor of a TDM
approach) proprietary.

(Terms in quotes above are there to indicate to Dr. David Reed and
others that I understand that they're poor models of the actual
physics.)

(*) 8% PER (1024 byte packets) is the tested limit for IEEE 802.11's
standard on sensitivity.


-------------------------------------------
Archives: http://www.listbox.com/member/archive/247/=now
RSS Feed: http://www.listbox.com/member/archive/rss/247/
Powered by Listbox: http://www.listbox.com