City-wide WiFi can be tough to build


The Internet wants to be ubiquitous. The Internet wants to go everywhere we go as we go about our lives. There's a dream many share of being able to tap the seeminly infinite resources of the Internet wherever one goes. One manifestation of that dream is for cities to install a ubiquitous WiFi network throughout their borders.

The citywide Wi-Fi reality check (Published: May 27, 2005, 4:00 AM PDT, By Marguerite Reardon, Staff Writer, CNET

The article discusses the effort in Philadelphia, and some naysayers who are saying they're going to run into problems. Let me offer my own set of possible problems, rooted in the experience of watching a previous effort to build ubiquitous wireless Internet access.

See, this isn't the first time this has been tried. There is a service named Ricochet originally designed by Metricom. Metricom spent 15 years developing the technology, and then blew it through an agressive rollout plan.

The Ricochet service was built using boxes attached to streetlights. You can see these boxes on streetlights all through the San Francisco Bay Area, and the service was installed in several other cities, probably including Philadelphia. The boxes were known as "poletop radios", and contained the ability to send/receive TCP/IP packets and route them over a mesh network. One would attach a Ricochet modem to their computer, the Ricochet modem finds the nearest poletop radio(s), and sends the TCP/IP communication to the closest ones. It was able to hand off the Ricochet connections at highway speeds, and the throughput was over 120 kbps in the year 2000.

Additionally, the poletop radios were installed at 1/4 mile radius intervals. The poletop radios would hand the packets they received upstream to what was called a "Wired Access Point" (WAP). The WAP had a T1 connection leading to it, sending packets to a centralized location per city where the T1 lines were congregated and the packets injected into the Internet.

This was a lot of infrastructure that had to be built, and the most difficult portion was negotiating access rights to install that infrastructure. You can't just go to the streetlight poles and attach boxes to them, you had to get permission from the people who owned the streetlights. This meant negotiations and lease agreements with cities and utility companies all through each metripolitan area. And then the WAP sites had to be spread through the city with an even distribution, meaning negotiating with building owners to have access to install equipment etc.

For example one of the WAP sites was installed in a historic old church. They took apart the church steeple, installed the antenna and other equipment there, and put the steeple back together. In a church that's been there for a couple hundred years. And now that Metricom went out of business, what is that WAP equipment useful for? Good question.

Back to Philadelphia and the problems of city-wide-WiFi.

My point is that such a WiFi system would have to follow a similar architecture. Instead of installing the "poletop" every 1/4 mile, it's got to be at an even closer spacing because useful WiFi range is around 100-300 yards. The WiFi equipment has to know how to handle what's called "mesh routing" so that a computer can connect to the closest WiFi hub and have their packets routed across neighboring hubs until it reaches a wire. And, then, before it can go onto the Internet, there have to be wires allowing the packets to be injected to the Internet.

This is an interesting project and I wish them luck. I also wish them to heed the history of previous attempts to do the same thing.