Newbie question about 802.11 broadcast in ad hoc mode

Dear All:

Can anyone kindly explain how the MAC layer work when I do
broadcasting in Ad hoc mode? I went through the standard, but it is
still not very clear to me.

When I send out UDP packages to broadcast addresses in the IP layer,
such as 127.0.0.255, will the data frame in the MAC layer use MAC
layer broadcast address as the destination address or somehow it knows
who is in the BSS and do many unit casts?

If it uses broadcast address, does this mean that there is no
ACK/RTS/CTS frame exchanges and only media sensing will be used for
collision avoidance? Does this mean that the broadcast will only be in
1Mbps /2Mbps?

If the station is moving, will the moving speed affect the
communication? It is easy to see that speed will affect the time
duration of the station in coverage. Other than that, is there any
potential problems in the MAC/PHY layer?

Thanks in advance!


Jay.

"Jay" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed) m...
> Dear All:
>
> Can anyone kindly explain how the MAC layer work when I do
> broadcasting in Ad hoc mode? I went through the standard, but it is
> still not very clear to me.
>
> When I send out UDP packages to broadcast addresses in the IP layer,
> such as 127.0.0.255, will the data frame in the MAC layer use MAC
> layer broadcast address as the destination address or somehow it knows
> who is in the BSS and do many unit casts?

IP broadcast is distinct from MAC broadcast. Implementation is actually up
to the IP layer. IP broadcast to the local subnet *could* be implemented by
sending out a sequence of host-specifc IP packets. In physical networks that
do not support MAC-level broadcast, this is how it's done. When MAC-level
broadcast is available, it's generally used.

802.11 doesn't deal with IP issues at all. It does specify exactly what
happens with MAC broadcast, in clause 9.2.7, "Broadcast and multicast MPDU
transfer procedure". For an infrastructure network, a MAC broadcast packet
is sent to the AP using RTS/CTS, if enabled, and ACK is used. Effectively, a
broadcast MAC to the AP is not yet broadcast - it is a point-to-point packet
that will be rebroadcast by the AP itself. In ad-hoc networks, a station
transmits a broadcast MAC frame directly, without using RTS/CTS, and without
an ACK (MAC broadcast is inherently unreliable). RTS/CTS and ACK don't work
for the simple reason that multiple CTS and ACK responders can't be
supported - how many responders do you need to wait for? In any case,
RTS/CTS isn't meant to be used in ad-hoc networks.

>
> If it uses broadcast address, does this mean that there is no
> ACK/RTS/CTS frame exchanges and only media sensing will be used for
> collision avoidance? Does this mean that the broadcast will only be in
> 1Mbps /2Mbps?

I don't understand the question. I don't see what datarates of 1 and 2 Mbps
have to do with any of this.

>
> If the station is moving, will the moving speed affect the
> communication? It is easy to see that speed will affect the time
> duration of the station in coverage. Other than that, is there any
> potential problems in the MAC/PHY layer?

Motion might affect datarate, if you get multipath echo or move between
areas of differing opacity to 2.4 Mhz signals. 802.11 isn't designed to be a
truly mobile network, like cell phones. It's expected that the stations will
either stay in place, or move slowly within a fairly narrow range (for
example, someone walking around with a PDA).

>
> Thanks in advance!
>
>
> Jay.

Actually 127.0.0.255 isn't a broadcast address. 127 is a class A network,
so the subnet broadcast address is 127.255.255.255. But it's pointless
anyway, since all addresses in the 127 network are loopback addresses.

You can broadcast to all hosts in your own subnet in two ways. For example,
if your subnet is 192.168.1.0 with a subnet mask of 255.255.255.0, then you
can broadcast to your subnet using either 192.168.1.255 or 255.255.255.255.
The preferred way is to use the Limited Broadcast Address 255.255.255.255,
but both will result in an address resolution to broadcast MAC address
ff-ff-ff-ff-ff-ff.

Ron Bandes, CCNP, CTT+, etc.

"Jay" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed) m...
> ...
> When I send out UDP packages to broadcast addresses in the IP layer,
> such as 127.0.0.255,
> ...

One assumes - at least, this one assumes - that the poster had in mind a
class-C subnet of 127. And it is not the case that 127 is hardcoded to be a
loopback net - this is merely its typical use.

"Ron Bandes" <RunderscoreBandes @yah00.com> wrote in message
news:uASoc.47609$(E-Mail Removed). net...
> Actually 127.0.0.255 isn't a broadcast address. 127 is a class A network,
> so the subnet broadcast address is 127.255.255.255. But it's pointless
> anyway, since all addresses in the 127 network are loopback addresses.
>
> You can broadcast to all hosts in your own subnet in two ways. For
example,
> if your subnet is 192.168.1.0 with a subnet mask of 255.255.255.0, then
you
> can broadcast to your subnet using either 192.168.1.255 or
255.255.255.255.
> The preferred way is to use the Limited Broadcast Address 255.255.255.255,
> but both will result in an address resolution to broadcast MAC address
> ff-ff-ff-ff-ff-ff.
>
> Ron Bandes, CCNP, CTT+, etc.
>
> "Jay" <(E-Mail Removed)> wrote in message
> news:(E-Mail Removed) m...
> > ...
> > When I send out UDP packages to broadcast addresses in the IP layer,
> > such as 127.0.0.255,
> > ...
>
>

Thanks a lot for replying.

"gary" <(E-Mail Removed)> wrote in message >
> In ad-hoc networks, a station
> transmits a broadcast MAC frame directly, without using RTS/CTS, and without
> an ACK (MAC broadcast is inherently unreliable).

As there is no ACT in broadcasting, do you somehow think media sencing
is used in the PHY layer ( MAC layer? I am not sure )? Or there is no
effort at all to reduce the collision.

> Motion might affect datarate, if you get multipath echo or move between
> areas of differing opacity to 2.4 Mhz signals.

Could you please elaborate this a little bit. What could happy if I
use 802.11 with broadcast when I drive, say 60 miles per hour.

Thanks a lot!

Jay

"Jay" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed) om...
> Thanks a lot for replying.
>
> "gary" <(E-Mail Removed)> wrote in message >
> > In ad-hoc networks, a station
> > transmits a broadcast MAC frame directly, without using RTS/CTS, and
without
> > an ACK (MAC broadcast is inherently unreliable).
>
> As there is no ACT in broadcasting, do you somehow think media sencing
> is used in the PHY layer ( MAC layer? I am not sure )? Or there is no
> effort at all to reduce the collision.

ACK is used for reliability, not collision avoidance. DCF basic access is
used for ad-hoc broadcast. This does carrier sense, collision avoidance,
and honors other sations' RTS requests.

The only thing it doesn't do is send RTS, or wait for ACK Since there is no
ACK, broadcast is inherently unreliable. Even in infrastructure mode, where
the broadcast packet is first sent point-to-point to the AP, the AP
broadcasts it to all stations without expecting ACK or using RTS.

Broadcast is disruptive to both ad-hoc and infrastructure networks, but it
affects ad-hoc more. As I mentioned, in an infrastructure net, clients
wishing to broadcast send point-to-point to the AP first, and this is ACK'ed
and RTS/CTS is used if enabled. The AP processes the clients' broadcasts
from a queue it keeps, which allows it to coordinate broadcasting with its
own normal outbound frames. For networks with an ISP link, where much of the
traffic is headed toward the wifi net from the ISP, this is much more
efficient.

There are lots of proposals out there for reliable multicast (broadcast is
just a special case). Maybe 802.11e will address it.

>
> > Motion might affect datarate, if you get multipath echo or move between
> > areas of differing opacity to 2.4 Mhz signals.
>
> Could you please elaborate this a little bit. What could happy if I
> use 802.11 with broadcast when I drive, say 60 miles per hour.

I would not count on this working. None of the 802.11 standards were
designed for it. The assumption is that an 802.11 network is analogous to an
Ethernet implemented using radio, and the nodes are expected to be
relatively stationary. When you have a node moving in a vehicle at 60 mph,
it implies that you will be moving very far away from the AP very rapidly.
To get ranges beyond 300 feet, you need high-gain antennas - which are
inherently more directional, and require line-of-sight. How do you guarantee
that from a moving car? Plus, the kind of reflections you get as you move
rapidly between rows of buildings with metal frames are simply beyond the
design capabilities of 802.11. It's simply not cell phone technology.

I don't know much about antennas, so I hope someone will correct me if the
above is misleading or contains incorrect statements. But I'm quite sure it
won't work well, and maybe not at all.

>
> Thanks a lot!
>
> Jay
>