-----Original Message-----
From: Joe Polastre [mailto:polastre@cs.berkeley.edu] 
Sent: Tuesday, July 15, 2003 1:04 PM
To: 'awoo@cs.berkeley.edu'; 'ttong@intel-research.net';
'culler@cs.berkeley.edu'
Cc: 'gsk-devel'
Subject: Island Multihop

Alec and Terrence are owed many props for their work on the multihop
routing integration with the GDI2Soft application. They worked their
asses off to help me get multihop integrated, reliable, and adhering to
power management goals.

On 7/9/2003 we deployed a test network of 14 multihop motes in the
field. We felt we weren't ready to do it yet because we didn't have
enough experience with the system, but due to the pressures of Wired, we
decided to go ahead with it anyway. The system worked tremendously
better than expected. Over the 6 days that it ran, we saw nodes switch
their parent from one that provided 87% reliable to one that provided
94% reliability as shown by the packet reception rates logged to the
database. The self-organization of the network using multihop routing
left a significant impression on the Wired reporter (I would argue it
was the coolest thing she saw while she was here).

Today we deployed 42 motes. In the first 20 minutes of the deployment,
37 motes joined the network creating a multihop network of over 4 hops
spanning approximately 1/5 of a mile. We expect the remaining 5 nodes
to slowly integrate themselves into the network (as I write this, I just
noticed that we now have 38 nodes). We expect that the network will
operate for a minimum of 58 days on 850mAh capacity ignoring the
presence of node failures due to environmental conditions.

With the pressure of the NEST demo and other activities, these guys
really put in a phenomenal amount of effort that I wanted to make the
rest of the GSK group aware of. We certainly owe them some lobster :)

-Joe