[Olsr-users] Proposed OLSR plugin to adjust nodes' TX power

[Ben West]

Hi Henning and Juliusz,

Thanks for your responses.  Please see my follow-up comments in-line below.

On Sun, Jun 5, 2011 at 2:47 PM, Juliusz Chroboczek <(spam-protected)>wrote:

>
> > primarily low-power / battery-power devices
>
> Another advantage is that lowering the tx power, you reduce the amount
> of interference.  Think of a network where there's a lot of nodes that
> have access to wired internet, so most often you just want to reach the
> closest wired node.
>
> Besides optimizing for tx power vs. interference (which would be an
operational goal for normal meshes), another constraint is to reduce the
node's detectability via minimal tx power.  Remember, the targeted user may
be a pro-democracy protester in an unfriendly country.  That is, we'd
anticipate mesh users wanting to only operate their nodes at at some bare
minimum tx power level to enable the mesh to function, tho necessarily at
high throughput.

Further along these lines, and to preserve plausible deniability for
said protesters, it is actually ideal to encourage more hops to the wired
node(s), which again favors low tx power.  That is, in a star topology (i.e.
degenerate mesh), any packet passing thru a wireless node can plausibly be
associated with the node's owner/operator.  In a deeper mesh topology, this
ability to identify a packet's source becomes more difficult.

> - Gradually turn down TX power until reachability of 2-hop neighbor
> degrades
> > below specified LQ
>
> Wouldn't it be better to turn down TX power until the reachability of
> direct neighbours drops just below 1, and then raise it back until it
> reaches 1?
>
> (I assume you're measuging direct reachability here, not ETX.)
>
>
The suggestion to adjust tx power based on 2-hop neighbor LQ was actually a
suggestion from Jeremy Lakeman from the Serval, as a simple way to push tx
power down to a bare minimum level while still preserving mesh integrity.
 Do please note this metric for mesh integrity is still being conceived.
 Also, I would expect such a mesh to operate with LQ < 1 at all times
anyway, given the manifold needs to minimize tx power.


> > - Battery-powered (or even wire-powered?) node with no visible neighbors
> > should eventually turn down (or turn off?) its transmitter.
>
> What if two such nodes meet?  (Look up -i in the babeld man page.)
>
>
I don't yet have a course of action envisioned if two orphan nodes meet,
besides no action.  We are still anticipating a mesh with at least one wired
node.

On Tue, Jun 7, 2011 at 3:15 AM, Henning <(spam-protected)> wrote:

> > T ... we discussed a potential OLSR plugin that would gradually turn down
> a
> > node's TX power until a minimum specified LQ was reached by one of that
> > node's 2-hop neighbors.
>
> There was a presentation on the Wireless Community Weekend (last weekend)
> in
> Berlin about implementing Power Control in mac80211. I will have to see if
> I
> find a link for you.
>
> Doing power control in the driver allows the mesh to adjust the power
> separately for each neighbor.
>

I would certainly be interested in seeing that presentation, also for my own
mesh use.

For the power plugin as planned, I believe we would need to turn down tx
power for all neighbors.  A goal is to reduce a node's detectability by
surveillance (i.e. SpecAn, directional antenna), which won't care which
neighbor a particular node is talking to.

-- 
Ben West
(spam-protected)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.olsr.org/pipermail/olsr-users/attachments/20110610/322bf17b/attachment.html>