Project Decision Log

This area attempts to document the project decision ladder so that as conditions change, decision history can be reviewed and revised as appropriate.

4/2009 Q = Where should we test/show our demonstration links

A = Pursue sites with the easiest access so as to facilitate experimentation. Strategic sites can be better pursued after viability has been demonstrated. Current favorite is LAH.

4/2009 Q = Should we buy a reflector dish antenna for the Nanostation 5's – 32 dB for $142 from Streakwave

A = No. Modification for external antenna is cheaper and should happen within the month. For contingency if that doesn't happen or doesn't work, let's get some 2 GHz equipment to continue with.

Radio spectrum and licensing

Part 15 vs Part 97

See the FAQ: Part 15 vs Part 97

Geographical Topology

Our goal in using the geographic topology of the Silicon Valley is to establish a very well interconnected set of sites situated around the valley on top of the first rank of hills. This would form the core of our network. From this core we intend to have "sector antennas" (at least 20 degree beamwidth antennas) pointing from these hilltop sites down into the valley to provide access to sites on the valley floor.

Network Topology

Using OpenWRT flashed radios

SVWUX relies heavily on OpenWRT platform routers.

Using the Linksys WRT54G boxes is very inexpensive but limiting. The primary problem we have run into is their use of Broadcom chips. Broadcom is very closed and proprietary and the OpenWRT distro uses an encapsulated binary driver from Broadcom, which severely limits our ability to use all the capabilities we need to - primarily the ability to use IPv6 at all.

On a positive note, Atheros has recently opened their driver source code and are now contributing to OpenWRT. As a matter of policy, SVWUX will strongly prefer vendors using Atheros chipsets for future purchases.

SVWUX makes efforts to contribute our changes and additions back to the OpenWRT project.

Offloading routing services from WRT devices to full machines

We turned the WRT boxes into dumb bridges (just a radio modem) because they would not have enough horse power to handle the OSPF and/or BGP routing, network services, VLANS/Qos/Traffic Shaping, and applications we thought would be installed. If we off-loaded some functionality we should off-load it all and make the WRT boxes drop-in field replaceable units. This was our hope but that will not remain pure as the WRT boxes need to be aware of VLANS and other stuff.

Reliability and Redundancy

We want to provide increased reliability and redundancy through having a number of connections via a number of different technologies. This will be managed by Layer 3 routing technology rather than overly complex and fragile hardware failover schemes. Sort of a "Redundant Array of Simple Connections".

-- Main.randomandy - 19 Jan 2009