The Bay Area Wireless Research Network White Paper

The Bay Area Research Wireless Network (BARWN), as its name suggests, is a Bay Area Wide wireless network dedicated to research into the development of very low cost, high bandwidth network infrastructure and applications that can use it.

With the recent commoditization of high speed (>10Mb/s) licensed-exempt wireless radios such as 802.11b and 802.11a, wireless networking has dropped several magnitudes in price. No longer does it cost tens or even hundreds of thousands of dollars to move data via high speed wireless links tens of miles. Also, the monopoly strangled "last mile" link is now opened to anyone.

We feel that the this technology has great potential to tie communities together, help with emergencies and erase the "digital divide". At this point, the technology is in its infancy to reach these goals. A number of problems need to be solved to get there. BARWN is designed to address them.

The first objective is looking to develop and document cheap long distance networking. We would like to see alternatives to the high cost of high speed long distance links. Currently low-end "carrier class" equipment such as Western Multiplex's Tsunami radios have list prices about $12,000 per radio. This does not include low loss wave-guide, and high gain dishes. An installation of this class of gear can mean costs of up to $50,000 per backbone link. These costs can be out of the range for many low income communities such as Native American reservations or underdeveloped countries. We believe we can get this price down below $2,500 per link with some sacrifices in uptime that can be partially solved with redundant links or more frequent installations on a path. the Customer Premises Equipment (CPE) costs for licensed-exempt radios for BARWN will be under $300.

A number of third-world countries are will benefit and can build on this development as this hardware has low power consumption and in the case of limited power infrastructure can be use with solar arrays and hydroelectric systems such as the type used by the EcoPartners Project¹ directed by Jon Katz.

The second objective is designed to increase the speed of which public safety can access data. Current technology such as Motorola's Private DataTAC² network is limited to 19.2Kb/s. There is an increasing demand in public safety for more bandwidth. Graphics intensive data such as identification photos for police or weather maps for fire, can seriously tax a 19.2Kb/s network. Networks that are shared over a large area and also used for other communication such as vehicle positioning will magnify this problem.

Once stopped, a police car or fire truck equipped with a laptop with a high-speed radio and a directional antenna pointed to a mountain top access point, could actually stream back live video of the incident with plenty of bandwidth to spare on a 11Mb/s network. Other examples of "broadband" applications for public safety could include streaming web cams, Voice over IP (VoIP), GIS mapping such as differential GPS, etc.

Also, many of these public safety networks rely on proprietary equipment that are expensive to deploy and replace. It is possible that with the low cost of licensed-exempt radios, that a network nearly 600 times faster than 19.2Kb/s and with far more security, as any level of well tested, openly-published encryption can be used, can be deployed at below 1/10 the cost.

The third objective understands that with the lower cost of equipment there is a trade off of shorter distance and uptime. Redundant, self-healing networks will be needed to addresses this problem. Dynamic routing protocols will be needed. As most ISO layer-2 (ie. spanning-tree) and layer-3 (ie. OSPF) protocols were developed for wired networks, additional tuning of the protocols will be needed to address the profile of failure for wireless networks. We expect to do this with the help of other non-profit groups such as the research center Packet Clearing House³.

BARWN is working with equipment manufactures on other products that offer support for non-802.11 protocols⁴. These new chipsets allow new protocols to be developed to extend the range of 802.11b/g/a and increase the speed for our backbone network. These new protocols can be developed and distributed in the same open-source/open-API manner as the high-level routing protocols that we would develop.

The final objective is meant to address the increasing asymmetrical access that the Internet is going. Low cost broadband access to the home and small office is provided through providers using cable modems or DSL technology. Typically this will mean that the client has access to hundreds to thousands of kilobits of bandwidth down to them but will be limited to just 128 Kb/s up. With more multimedia rich demands on publishing, 128 Kb/s will limit how many uses can access a server or the kind of data that can be published. On top of that most broadband providers will use dynamic IP address assignment so it would be difficult to find any servers that a client would use to publish materials. In some cases such as cable-modem broadband providers, they will even filter content (through port filtering) so it would be impossible to publish from a client.

This problem has become so significant even non-Internet organizations like the ACLU are taking action to stop it⁵.

In order to encourage the true democratic communication tool that the Internet was and should be, we would like to start to deploy high speed bandwidth to the home that does not have the restrictions commonly encounter by current broadband providers as outlined above. We are interested in seeing the Internet back to how it was originally designed as a resource sharing and publishing tool that provides equal access to all.

BARWN feels that many of the problems identified here may be solved by obtaining access to strategically located communications sites. Traditionally mountain tops and towers have been seen as having the advantage of clear line-of-sight to other mountain tops and can view significant terrain such as populations areas. By anchoring our backbone on these sites we can provide connections to the network to other communities and provide "last mile" connections to end users.

This differs from what other community networks are doing through mesh networks as we do not have to wait for the critical mass of random "technically advantaged" individuals to show up and help. We see this network design as supporting these organic networks through bandwidth and expertise. The network would have a more stable infrastructure and organization.

BARWN is concerned about proprietary protocols that bolt a user into one manufacture. So much as been accomplished in standardizing high-speed wireless networks so they are not vendor dependent. BARWN is committed to taking this further, where the the equipment we will deploy on will not be dependent on a signal manufacture or brand of device. Software will be written so it can be run on any practically any x86 PC motherboard and with the effort that almost any radio card can be used. We gain so much in scaling and cost by basing the BARWN architecture on common interfaces and open APIs such as Ethernet, MiniPCI & PCMCIA along with open platforms such as PC clones.

BARWN understands that each of these objectives will encounter limitations based on current regulations and policies that surround them. BARWN is interested in making this network replaceable for others to deploy. Designs will be in accordance with local, state and federal regulations and laws encountered in the US with foresight for deployment outside of the US.

During the last 12 months of the planning and development phase, it was identified, through the feature specifications, that current off-the-shelf low cost licensed-exempt equipment would not support the high level (layer 2 and above) dynamic routing protocols that BARWN would expect to use and provide the flexibility needed for development and upgrade. During the months of software and hardware development BARWN created a flexible low cost system that can be scaled to as may radio links as needed for a site.

The first attempt for BARWN at long distance networking was the installation of a 20 mile path between Hayward and South San Francisco with experimentation to increase the reliability of this link. A number of lessons were learned in the deployment of this link such as reliability, configuration, new tools, that have been documented in the BARWN mailing list.

A second link was completed February 19^th when we deployed an 11Mb/s link from the United Layer colocation center at Sixth and Mission⁶ in San Francisco to an Access Point on top of San Bruno Mountain. This site has the ability to supply up to 11Mb/s of bandwidth to the Internet to anyone in an 8 mile radius that points a 2.4GHz high gain antenna to it.

BARWN has also secured access or currently negotiating access to a number of mountain tops around the Bay Area for links including KPFA's transmitter site in Berkeley, Sign Hill in South San Francisco and the County of Santa Clara communications site at Carol Drive in San Jose. The next steps are to provide backbones to these locations and additional connections to the Internet to increase bandwidth and reliability.

Funding for the project is currently limited to the donations of the members of BARWN and the mountain top and tower managers and owners. We are in the process of setting up a California non-profit corporation and then applying for a 501(c)3 status from the IRS. We anticipate no problems in this process. Until we're granted non-profit status from the IRS, we will be working under the umbrella of the Internet Archive⁷. Afterwards, we will, on our own and with other non-profits, apply for a number of grants that are specifically designed for this type of effort.

Internet bandwidth is or will be provided through at least four transit providers: Internet Archive (IA - San Francisco), San Mateo Regional Network, Packet Clearing House (PCH - Berkeley) and Layer 42 (San Jose). We expect the (wireless) connection to IA Peering will eventually be provided through the Palo Alto Internet Exchange (PAIX) via a wireless link.

Tim Pozar is a communications consulting engineer specializing in commercial microwave path and tower engineering for government. Pozar is a founding member of the Bay Area Wireless User Group. He was an early entrepreneur and developer in the Internet startup area, by co-founding a number of companies such as TLGnet (San Francisco's first ISP), Brightmail (first commercial anti-spam company) and Omniva (digital rights management). Previous to this, for 25 years, Pozar was a radio broadcast engineer for commercial and non-commercial radio stations.

Pozar is currently leading an effort, called BARWN, to deploy of high speed Internet access through out the San Francisco area. The infrastructure is based on low-cost licensed-exempt equipment.

Pozar has also published a number of papers covering the regulatory issues in the United States of licensed-exempt radios.

Matt Peterson is the founder of the Bay Area Wireless User Group. BAWUG began as outgrowth of Peterson's work on PlayaNET; a temporary wireless intranet for the weeklong desert art expression event Burning Man. The group provides an informational clearinghouse of wireless knowledge, monthly networking events & features an active 2000+ subscriber mailing list. Mr. Peterson has been involved in ISP and related technologies since his middle school age.

Matt is currently consulting for various commercial and non-for-profit wireless organizations. He is also a foundation board member of FreeNetworks.org and co-chair of the first dedicated community wireless networking (CWN) conference, FNcon. Peterson has spoken on CWN at various industry events, including: 802 Planet, EyeForWireless, Emerging Technology, and APRICOT.

Western Multiplex and Tsunami are trademarks of Proxim Corporation. Other product names used are trademarks of the companies that produce them. Mention of their names should not be construed as a challenge to such status.

3 - Packet Clearing House - http://www.pch.net

4 - Chipsets from Atheros and Broadcom allow the generation of non-standard network framing

5 - "ACLU Warns of Threat to Online Free Speech From Cable Monopolies. A Technical Report Showing How Cable Operators Can Interfere With Internet Access" -

http://archive.aclu.org/news/2002/n071002a.html

6 - Bandwidth donated by Internet Archive.