In article ,
Paul Keinanen wrote:

Running a 32 km LOS link is problematic in practice. At least the
first Fresnel zone should be clear of obstacles. At midpoint you
should have about 30 m free space _below_ the direct optical path
(i.e. you would have to add 30 m to the antenna towers), but still you
should have 10-20 dB fade margins for reliable operation.

I understand that there's another issue with long links which can
cause some problems. 802.11 cards are designed to automatically
acknowledge packets received, and retransmit them if they aren't
acknowledged promptly. This retransmission protocol is done by the
card firmware, and occurs below the level of any operating-system
packet acknowledgement or retransmission (e.g. TCP).

The automatic-retransmission timeouts in the cards are set to a fairly
short time-period, in order to improve throughput under typical usage
conditions.

A 32 km link is going to have a round-trip time of around 200
microseconds. If the transmitting card's timeout value is set to less
than this, it'll start retransmitting the "lost" packet before a
successful acknowledgement can come back. Lather, rinse, repeat.

Traffic will actually get through, I gather... the successfully-
received packets will be forwarded to their destination, even though
the sending card thinks that they haven't been received (and logs them
as "discarded, excessive retransmits"). Throughput suffers badly,
though.

Some cards and/or access points apparently allow the retransmission
timeout value to be changed, to allow for sufficient speed-of-light
travel time.

--
Dave Platt AE6EO
Friends of Jade Warrior home page: http://www.radagast.org/jade-warrior
I do _not_ wish to receive unsolicited commercial email, and I will
boycott any company which has the gall to send me such ads!

On Thu, 18 Oct 2007 11:22:20 -0700, (Dave Platt)
wrote:

In article ,
Paul Keinanen wrote:

Running a 32 km LOS link is problematic in practice. At least the
first Fresnel zone should be clear of obstacles. At midpoint you
should have about 30 m free space _below_ the direct optical path
(i.e. you would have to add 30 m to the antenna towers), but still you
should have 10-20 dB fade margins for reliable operation.

I understand that there's another issue with long links which can
cause some problems. 802.11 cards are designed to automatically
acknowledge packets received, and retransmit them if they aren't
acknowledged promptly. This retransmission protocol is done by the
card firmware, and occurs below the level of any operating-system
packet acknowledgement or retransmission (e.g. TCP).

Fade rates typically vary between a few times a second to several
days, thus resending a frame one millisecond after the initial frame
is not usually going to make a big difference, if the initial frame
was, say 10 dB below the threshold (i.e. deep in the noise).

While COFDM and various spread spectrum systems will help combating
(fast) frequency selective fading, but these do not help in slow fade
due to varying vertical atmospheric temperature and humidity profiles.

You still need the fade margin for slow fading, if reliable
communication is required.

The situation is different in typical amateur radio usage, in which
you are most interested in rare situations, in which the signal
strength increase above the average signal level, thus increasing the
communication range in rare occasions.

Paul OH3LWR