Jump to content

Talk:IEEE 802.11 RTS/CTS

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

== Edit Rational == NTABWO ARIBYO

Deleted the following from the article:

Another situation where terminal A and B trying to send C, where C and B are very close while A is far but still in the range, Near/Far Terminal problem will occur. Since power decay propotional to distance square, B drowns out A's signal (at physical layer), so C cannot receive A. Thus power control is required.

This specific issue is not a problem for 802.11, since even if the power levels were adapted to be equal at the receiver (C), a collision would occur since "C" can only listen to one transmission at a time. Actually, if B could really "drown out" A, then at least B would be able to communicate! I believe the author is thinking of the way Code Division Multiple Access (CDMA) schemes can work, where multiple senders can transmit and be heard at the same time, provided their power levels are closely controlled. This is a common scheme in the cellular world, search for "cdma" or "IS-95" for more information. However in 802.11 a given receiver can only listen to one sender at a time. --DW1492 03:39, 18 December 2006 (UTC)[reply]


Exposed Station Problem solved?

[edit]

So does RTS/CTS solve the exposed station problem or not? i.e. Does a node start sending data if it overhears an RTS (but not CTS) from another node? The current revision says it does, but previous revisions say it's a "misconception" to believe it does. - 69.197.128.53 04:32, 22 April 2006 (UTC)[reply]

I hadn't heard of this behavior before, and I don't think it would be a good idea. Suppose there are 2 links, s1→d1 and s2→d2, and s2 hears the RTS from s1 but not the CTS from d1. If s2 went ahead and started transmitting at the same time as s1 it seems like it may (or may not) still cause interference with s1's packets to d1. Whether or not this happens depends on the SNR at d1, which could be low, even if s2 didn't hear d1's CTS. Of course this behavior would be prevented by physical carrier sensing. So how does RTS/CTS work with physical carrier sensing? My understanding was that PCS was always on.

--Ms whosit 06:11, 18 August 2006 (UTC)[reply]

802.11 Radios only listen OR transmit - they can't do both at the same time (at least, they're not expected to). Therefore, they listen for a clear channel before attempting to transmit. Regarding causing interference with d1 - yes s2 might increase the noise level, but since it couldn't hear d1 in the first place the noise should be negligible (presuming all radios are using the same transmit power levels). --DW1492 03:23, 18 December 2006 (UTC)


Empirical Data?

[edit]

http://aqua.comptek.ru/test/HiddenNode/hidden_node_en.html says RTS/CTS does not solve the hidden station problem at all in certain circumstances. From my experience, enabling RTS/CTS for a hidden node in a network that is under heavy load does not help. I don't know the exact reasons; however, I suspect nodes with high upload and high SNR prevent RTS requests from the other node to reach the access point. Are there any other tests apart from the comptek ones? - NotInventedHere 12:27, 28 May 2006 (UTC)[reply]

My 2 cents is that the comptek "Empirical Data" is slanted in favor of their product. First, their graphs show bursty variation in throughput for CSMA/CA, but they failed to report the actual throughput achieved, or packets dropped (if any). Also, the 17 second throughput periods may have forced some flows to appear "stopped" when in fact they were not. Granted, the latency will be highly variable, but bursty access is not necessarily a problem for throughput. This is not to say that CSMA/CA doesn't have it's problems, and it will perform less than optimal under high load situations, just not sure they demonstrated that really well. --DW1492 03:23, 18 December 2006 (UTC)

Regarding your specific issue with RTS/CTS under heavy load - from your wording it appears you only enabled RTS/CTS on a single node? It would need to be enabled on all nodes. If you did enable on all nodes, then the hidden node may still be getting "blocked out" by multiple nodes physically closer to the access point who who are constantly making requests that are received by the access point before the request from the hidden node arrives. But to consistently have this issue you must be running quite a loaded network. How to resolve? Add more capacity, or use a polling scheme similar to the Comptek solution. --DW1492 03:23, 18 December 2006 (UTC)

there is an area of medium access research known as spatial reuse which addresses the problem of overactive/underactive RTS/CTS exchange. jfgi. 128.128.98.46 (talk) 18:39, 20 October 2008 (UTC)[reply]

Broadcast address

[edit]

there should be mention of what is done in the case of broadcasted frames; when a node receives an RTS not addressed to itself, but instead the destination address is a reserved broadcast address, how does it decide whether/when to return a CTS frame? i cannot find this in the literature. 128.128.98.46 (talk) 18:42, 20 October 2008 (UTC)[reply]

Explanation of terminology choice

[edit]

Exposed node problem redirects to Exposed terminal problem, but Hidden terminal problem goes to Hidden node problem. I've noted on both pages that the pattern should be consistent, but until there is a decision I am standardizing this page with ``terminal because that is what I see most in my reading. Feel free to edit. 129.93.154.192 (talk) 19:41, 10 September 2009 (UTC)[reply]