Cutting the Wires... (Part 1)

Copyright (c) 2000-2001 by Rich Morin
published in Silicon Carny, May 2000

Long-haul networks keep increasing in capacity, but the "local loop" continues to be a bottleneck. Wireless networking may offer a solution, at least in some locations.

Fiber is being run alongside every new freeway, railroad line, and bicycle path. Networking companies are spending billions of dollars to connect these links together with high-speed hubs. As a result, long-haul connections have ceased to be a major bottleneck for most of us.

The "local loop" (service to the subscriber), however, is still a problem in many areas. Cable TV companies may provide service, but it generally lacks bandwidth guarantees and all too often has unacceptable usage restrictions.

My own cable company, for instance, specifically disallows use by "servers". Although I can understand this (cable modem systems are optimized for inbound traffic), it is still a show-stopper for me...

Telephone companies are hyping DSL quite loudly, but they ofen promise more than they can provide. My own attempt to get DSL was stymied for over a month by a combination of distance limitations, political problems, and general lack of organization.

Wireless Networking

Fortunately, as I was starting to get desperate, I found out that wireless network access was available in my locale. Although my first few months were problematic (the server kept "losing" my machines), that problem has long since been resolved.

For the past year, in fact, my Internet connection has been supplied by a Yagi antenna, mounted to a mast on my roof. The connection gives me a guaranteed 384 Kbps (in both directions!). Typical performance ranges from 700 to 2000 Kbps; the best performance, understandably, is in the wee hours of the morning.

My "WavePath" connection employs a Multipoint Microwave Distribution System (MMDS). The antenna on my roof points at San Bruno Mountain, where the nearest WavePath base station is located (another station covers the San Jose area).

Each base station services a number of clients, electronically "pointing" its antenna in different directions. The base station isn't looking my way most of the time, but it checks in with my site frequently enough to provide smooth service.

Unfortunately, the WavePath connection isn't exactly cheap. At $150/month, it is about twice as expensive as the local cable modem service. It does, however, allow me to operate a server, as well as providing bandwidth guarantees. In short, it's a deal I can live with...

Hacker Alternatives

I am always on the lookout for alternatives, however. Ideally, these would be the same sorts of "toaster nets" (such as our local Little Garden Network) that opened up economical, unrestricted Internet access a few years ago.

Until recently, the cost of equipment (modems, transceivers, antennas) has been prohibitive, keeping wireless networking out of the reach of most private citizens.

Fortunately, the IEEE 802.11 standard, which employs 2.4 GHz spread-spectrum technology, is starting to become economically available. Cards such as the Farallon SkyLINE and Lucent Wavelan (used in the Apple "Airport") seem to be well suited for creating low-cost radio-based WANs. IEEE 802.11 has reasonable bandwidth (~10 Mbps) and (with an external antenna) acceptable range.

A recent discussion on a MacInTouch bulletin board covers users' experiences with adding antennas to Wavelan-based transceivers. 10 mile distances are quite possible using a pair of directional (Yagi) antennas; under nearly ideal circumstances (to an island), in fact, a 35-mile range has been achieved!

Cushcraft makes a "Yagi Directional" antenna (PC2415N, 15.9 dBi, $157), which is well suited for long-distance, point-to-point links. Add about $250 for a transceiver card, and a few bucks for masts and cables, and you have a base station for a mere $500.

Some Linux hackers down in Australia have been working up even cheaper ways to go. Using pre-802.11 cards and home-brew antennas, the Canberra Wireless Network has brought the station cost down to $250. At $500 per two-station link, this stuff starts to sound pretty reasonable!

Xnet Wireless Network, a more formal effort, is trying to promote the wide use wireless links as part of a country-wide network. Their web page is both intriguing and informative.

There are also some efforts going on here in the US. The "Bay Area Wireless Internet" web page lists folks all around (and above!) the US that are exploring ways to use wireless technology.

Legal Issues

Each country has its own rules regarding radio, but most of the rules tend to be very similar. Most democratic countries, in any case, allow unlicensed use of the 2.4 GHz band by spread-spectrum transceivers.

Unfortunately, there are two gotchas. The first is that power levels aren't allowed to go beyond set limits, based on the band, etc. The second is that arbitrary combinations of transmitters and antennas may produce unacceptable levels of harmonics.

Power Levels

In the US (and many other countries), the Effective Isotropic Radiated Power (EIRP) limit for spread-spectrum use of the 2.4 GHz band is 36 dBm (36 decibels above a milliwatt), or 4000 milliwatts.

The raw power level of commercial 802.11 cards is quite low, but if you add enough antenna gain, you could exceed this limit. The Wavelan card is rated at 15 dBm. Adding the 16 dBi Yagi antenna above, for instance, would yield a system with 31 dBm of EIRP, which is still under the FCC's 36 dBm limit.

Put another way, the card's 30 mW, when multiplied by a gain of 40 (10^1.6), has an EIRP of 1200 mW (40 * 30 mW). Again, this is under the 4000 mW limit set for unlicensed use on this band.

Home brew antennas don't come with tested dBi ratings, however, so you're on your own if you venture in that direction. Although you don't need a license to design antennas, you are still responsible for keeping your transmissions within the legal limits for your country. If you don't know what you're doing, get qualified help!

Harmonics, etc.

Because arbitrary combinations of transmitters and antennas may produce unacceptable levels of harmonics, transmitter/antenna combinations that are used in the US must be tested and certified by the FCC to meet Section 15 rules. Specifically, Section 15.204 requires that the specific combination of antenna, transmitter and external amplifier, if any, must be certified as a system and can only be sold and used in that combination.

This isn't just bureaucratic silliness. A transmitter which generates entirely acceptable amounts of harmonics with a small, built-in antenna may be a real problem with an external, high-gain antenna. These harmonics, produced by resonances within the system, may get in the way of other types of radio use.

Harmonics from a 2.4 GHz system, for instance, could show up in radio astronomy and satellite downlink bands, both of which use low signal levels and could suffer greatly from such interference.

In short, if you want to stay legal (at least in the US), you should find a manufacturer which can supply a system which has been certified by the FCC. Next month's column will discuss some manufacturers and systems which have been through this process.



American Radio Relay League (ARRL)

  The ARRL has been serving the needs of radio amateurs
  (aka "ham radio operators") for years.
  Their "Radio Amateur's Handbook" (a yearly publication)
  should be on every electronics experimenter's bookshelf.
  Their "UHF/Microwave Experimenter's Manual" is quite relevant
  to antenna design and other issues of the 2.4 GHz band.

Bay Area Wireless Internet

Canberra Wireless Network

Cushcraft Communications Antennas

Farallon Communications

Federal Communications Commission (FCC)

IEEE 802.11

Lucent Technologies



Xnet Wireless

About the author

Rich Morin ( operates Prime Time Freeware (, a publisher of books about Open Source software. Rich lives in San Bruno, on the San Francisco peninsula.