On 15 Oct 2004 21:04:46 -0700,
(E-Mail Removed) (c hore) wrote:
>Are delay spread specs of interest, e.g., in designing point-to-point
>links? How does one use them, if at all?
Delay spread has to do with the ability of the receive to deal from
multipath. The idea is that the maximum difference between
sequentially sent symbols, that can be be bounced off something and
therefore delayed, before the delayed symbol tramples the *NEXT*
symbol to be sent. This is also called inter-symbol interference. If
the delay spread is larger than the inter-symbol time, you will have a
problem. Delay spread is sometimes called a measure of "robustness".
Google found this article:
http://www.eetimes.com/story/OEG20021008S0001
which sorta covers the topic.
>For example, the delay spread specs @11Mbps, @5.5Mbps, @2Mbps,
>and @1 Mbps, respectively, for two cards are are:
> 65ns / 225ns / 400ns / 500ns --- ORiNOCO Gold
> 140ns / 300ns / 400ns / 500ns --- Cisco 350 Aironet
>My vague understanding is that delay spread is the difference in time
>between the unreflected signal and the most delayed reflected version.
Yep. That's it.
>If so, it seems to me that delay spread would then really be a
>property of the environment, not the radio; and that the
>"delay spread" spec of a radio is really the maximum delay spread
>that the radio can tolerate for a given error performance.
Nope. It's a function of how long the radio waits before sending the
*NEXT* packet. If the reflected multipath signal is delayed longer
than the time to the next packet, then the delayed packet and the
directly sent next packet arrive simultaneously, clobber each other,
and neither gets decoded.
>Is that the correct interpretation? Does this mean that the Cisco
>is better than the ORiNOCO, in that it can tolerate larger delay
>spread (i.e., more amount of reflected signal) at 11Mbps?
I don't know. It means that Cisco 350 can tolerate a longer multipath
than than Orinoco. It doesn't affect performance, but is a good
measure of how well the unit will preform in a highly reflective
environment. As a rule of thumb, indoor is about 50nsec, office about
100nsec, and industrial (lots of metal boxes) about 250nsec. If you
run into anything over 250nsec delay spread, you're playing in a hall
of mirrors and probably won't find it to be a useable location.
>Are the terms "reflected signal" and "multipath" synonymous,
>such that the delay spread spec is an indication of radio
>susceptibility to multipath?
Yep. Note that multipath may not be a single reflected signal, but
might be a series of delayed transmissions. That's why it's called a
"spread". To the receiver, a reflected signal is the same as a
delayed signal.
Incidentally, the reason that OFDM is better for multipath than
802.11b is because it sends each carrier at a lower symbol rate. Note
the delay spread is better (larger) at 1Mbits/sec than at 11Mbits/sec.
That works fine until you slam into the guard interval (800nsec).
http://www.wireless-nets.com/papers/...s_lan_ofdm.htm
(See section on "benifits of OFDM".
--
Jeff Liebermann
(E-Mail Removed)
150 Felker St #D
http://www.LearnByDestroying.com
Santa Cruz CA 95060 AE6KS 831-336-2558
Similar Threads
Linear Mode
