[info] IBSS (adhoc-mode) does multihop?

Hey people,

I wanted to know whether, as specified, the IBSS (adhoc) mode of the 802.11b
specification does multi-hop. In the standard case if we have 4 stations
(called A,B,C,D) and they are all in range of each other then they should
all be able to talk to each other (assuming they all are setup with the
correct BSS id etc). Thats fine.

However, what if they are set up in a 'chain', meaning A-B-C-D like so, and
they are laid far enough apart that they can only talk to their neighbours.
Then obviously A can't directly talk to C or D. But as implemented, does
IBSS route the packets (frames?) from A to C via B? Or does it fail and
result in 4 overlapping IBSS's? But that sounds wrong to me.... not the
least because e.g. B will be in both the A-B IBSS and the B-C IBSS, both of
which will have the same BSS id.

Disclaimer: I ask because I am planning to implement this multi-hop
functionality as an undergraduate Senior Project using a fast-graphing
algorithm me and my mates came up with... sortof a showcase. But I don't
want to redo something that the spec already does. If you could cite your
answer in a spec or website where I can get further info, I'd be extremely
obliged.

I apologize if the question is poorly worded. I have just been digging
through the huge 802.11-1999 spec on ieee.org and am completely spaced out.

best, and thanks in advance,
/t.

(public posted replies fine but private replies to
tjaffri-please-no-spam-AT-cs.stanford.ee-dee-you)

I forgot my password for the IEEE Xplore website, but did you try over
there?
Probably in the Basic search you can type some thing like "802.11 ad-hoc
architecture" or "802.11 ad-hoc multi hop" to find relevant publications.

HTH

"Taqi Jaffri" <(E-Mail Removed)> wrote in message
news:c4glco$3pp$(E-Mail Removed)...
> Hey people,
>
> I wanted to know whether, as specified, the IBSS (adhoc) mode of the
802.11b
> specification does multi-hop. In the standard case if we have 4 stations
> (called A,B,C,D) and they are all in range of each other then they should
> all be able to talk to each other (assuming they all are setup with the
> correct BSS id etc). Thats fine.
>
> However, what if they are set up in a 'chain', meaning A-B-C-D like so,
and
> they are laid far enough apart that they can only talk to their
neighbours.
> Then obviously A can't directly talk to C or D. But as implemented, does
> IBSS route the packets (frames?) from A to C via B? Or does it fail and
> result in 4 overlapping IBSS's? But that sounds wrong to me.... not the
> least because e.g. B will be in both the A-B IBSS and the B-C IBSS, both
of
> which will have the same BSS id.
>
> Disclaimer: I ask because I am planning to implement this multi-hop
> functionality as an undergraduate Senior Project using a fast-graphing
> algorithm me and my mates came up with... sortof a showcase. But I don't
> want to redo something that the spec already does. If you could cite your
> answer in a spec or website where I can get further info, I'd be extremely
> obliged.
>
> I apologize if the question is poorly worded. I have just been digging
> through the huge 802.11-1999 spec on ieee.org and am completely spaced
out.
>
> best, and thanks in advance,
> /t.
>
> (public posted replies fine but private replies to
> tjaffri-please-no-spam-AT-cs.stanford.ee-dee-you)
>
>

An IBSS is an ad-hoc point-to-point network. There is no controlling station
taking the role of AP. There is no routing, no forwarding or packet
bridging, of any kind. All stations must be able to directly receive the
signals of all other stations. An ad-hoc network can be set up in which this
requirement is violated, which means not only that some stations won't be
able to talk to one another, but also that the so-called hidden node problem
occurs - some stations cannot detect frames sent between other stations, and
attempt to transmit at the same time, jamming the network and eventually
causing the overall bitrate to degrade.

It's always possible that I missed something in the standard, but that's my
understanding. Can you name the clauses in the spec that describe what
you're calling multihop functionlity?

"Taqi Jaffri" <(E-Mail Removed)> wrote in message
news:c4glco$3pp$(E-Mail Removed)...
> Hey people,
>
> I wanted to know whether, as specified, the IBSS (adhoc) mode of the
802.11b
> specification does multi-hop. In the standard case if we have 4 stations
> (called A,B,C,D) and they are all in range of each other then they should
> all be able to talk to each other (assuming they all are setup with the
> correct BSS id etc). Thats fine.
>
> However, what if they are set up in a 'chain', meaning A-B-C-D like so,
and
> they are laid far enough apart that they can only talk to their
neighbours.
> Then obviously A can't directly talk to C or D. But as implemented, does
> IBSS route the packets (frames?) from A to C via B? Or does it fail and
> result in 4 overlapping IBSS's? But that sounds wrong to me.... not the
> least because e.g. B will be in both the A-B IBSS and the B-C IBSS, both
of
> which will have the same BSS id.
>
> Disclaimer: I ask because I am planning to implement this multi-hop
> functionality as an undergraduate Senior Project using a fast-graphing
> algorithm me and my mates came up with... sortof a showcase. But I don't
> want to redo something that the spec already does. If you could cite your
> answer in a spec or website where I can get further info, I'd be extremely
> obliged.
>
> I apologize if the question is poorly worded. I have just been digging
> through the huge 802.11-1999 spec on ieee.org and am completely spaced
out.
>
> best, and thanks in advance,
> /t.
>
> (public posted replies fine but private replies to
> tjaffri-please-no-spam-AT-cs.stanford.ee-dee-you)
>
>

Gary,

> It's always possible that I missed something in the standard, but that's
my
> understanding. Can you name the clauses in the spec that describe what
> you're calling multihop functionlity?
>

You are probably right; and I didn't see anything that said this in the
spec. However, I do believe that the term 'adhoc network' is overloaded and
other than the 802.11 mode it might also mean a self-configuring multi-hop
'mesh-type' network. Thats why I wanted to make sure what the 802.11 mode
exactly does since I would look pretty silly implementing something on top
of IBSS that the standard already did.

Kalyan, thanks for the idea. I will try to get into IEEE XPlore today.

Thanks to both of you. Other comments always welcome.

/t.

Well, actually the term ad-hoc is Latin for "concerning the thing". It's a
generic term originally used to refer to any attempt to address a single
specific issue. For example, an ad-hoc committee is usally set up in
response to a a single problem, exists only to solve it, then dissolves. It
usually connotes an informal, relatively unstrucutured approach.

In networking, the usage has these properties. It doesn't precisely refer to
a "self-configuring" network that can be joined by any station without
reconfiguring the rest of the network, because this is true both of an IBSS,
and a BSS supported by an AP. Joining an infrastructure network has no more
effect on the rest of the stations than joining an ad-hoc net. The main
difference is that an IBSS exists only to support the fundamental
requirements of sharing a channel - sensing other transmitters, avoiding
collisions, and sharing timing information to synchronize logical clocks. It
doesn't support a backend distribution system, roaming, solving the hidden
node problem, or any of the other value-adds that come with an
infrastructure nework. It exists to solve a single, specific problem in a
relatively unstructured way - hence, it is an ad-hoc network.


"Taqi Jaffri" <(E-Mail Removed)> wrote in message
news:c4hnsi$3ua$(E-Mail Removed)...
> Gary,
>
> > It's always possible that I missed something in the standard, but that's
> my
> > understanding. Can you name the clauses in the spec that describe what
> > you're calling multihop functionlity?
> >
>
> You are probably right; and I didn't see anything that said this in the
> spec. However, I do believe that the term 'adhoc network' is overloaded
and
> other than the 802.11 mode it might also mean a self-configuring multi-hop
> 'mesh-type' network. Thats why I wanted to make sure what the 802.11 mode
> exactly does since I would look pretty silly implementing something on top
> of IBSS that the standard already did.
>
> Kalyan, thanks for the idea. I will try to get into IEEE XPlore today.
>
> Thanks to both of you. Other comments always welcome.
>
> /t.
>
>