Rooftop mesh almost ready to fly

The first telephone exchanges in the US were built by fire-alarm companies. They had central offices with radiating networks of wires to carry the press-button alarm signals over distances of a couple of kilometers, and they only needed to add a patch-panel.

http://www.telecomasia.net/telecomasia/article/articleDetail.jsp?id=80303

This star configuration has persisted with us today, even though radio is often used in the local loop, and even though CDMA allows users to share wide bands of radio frequencies without needing call-by-call channel allocation.

Central hubs are always potentially single-points of failure, but from the viewpoint of the service provider, there’s one supreme advantage in having all calls pass through his office – they are also central points of billing. The star-hub configuration allows them to bill customers on a per-call and/or per-minute basis.

But with CDMA and intelligent radio controls, it doesn’t make much technical sense to rely on a central hub. The base-station might be many kilometers away and hidden behind a mountain or a building, while the party being called is nearby. Hub systems therefore need extraordinarily complex power-output and timing controls.

A few years go the US tank corps decided that it could no longer trust any star network. When calls between tank commanders passed through a headquarters’ base-station, the network could be wiped out during pitched battle by one stray shell.

So, in association with California’s Santa Cruz University, they developed the concept of a CDMA-based, peer radio network known as GloMo. This technique relies on low-power radio modems in each tank which double as intelligent relay-routers when not carrying their own conversation, thus creating a mesh of potential connections across the full extent of the battlefield.

Each modem maintains routing tables of all other nearby modems, and with fairly standard Internet routing protocols they are able to calculate the most efficient path across the network and divert calls around any node destroyed or occupied in a call itself.

There are many advantages to this approach – these networks are self-scaling and self-healing. The more users there are, the greater the total packet bandwidth and the greater the number of redundant pathways. And long distances can be spanned with multiple hops using low-power transceivers, which also equate with reliability.

The only substantial problem with meshed networks comes from signal latency – router delays accumulated. However with GloMo, the voice latency was still considerably less than VoIP on the Internet.

Management of such systems is also simple – it consists mainly from checking for blocked signal links, and meshed relay-based networks are assembled like Lego blocks, using standardized, off-the-shelf transistorized transceivers which are virtually plug-and-play.

Solving the billing problem

What stymied the development of meshed networks was that, without a central hub, there’s a single point of billing failure. The telcos generally see little value in promoting a free/cooperative service in competition to their own for-profit networks, and they’ve done their utmost to restrict these devices to the higher ISM spectrum with very-limited transmitter powers.

However the Santa Cruz team commercialized the idea a few years ago for suburban Internet use under the name Rooftop, and this company was taken over in 1999 by Nokia who tentatively extended their trials to Canada … and then abandoned the project.

Rooftop’s data-only radio modem generally spanned only a kilometer or so, but when used from a fixed location it was possible to extend the range with highly directional antennas.

Three different Australian groups looked at the possibility of modifying Rooftop technology in the late 1990s, with the idea of using it to link rural and remote outback communities with both voice and data. Australia’s distances and low population densities, make copper wire networks vulnerable to lightning, and copper is costly to replace.

The problems the Australians found with Rooftop came from the power, bandwidth and frequencies employed, and the lack of a large-enough market to produce suitably modified equipment in Australia. But clearly the potential for such a technology is immense in developing countries.

Now a British company, Rural Radio Systems, has come up with its DaRT (Digital Access Rural Telecom) solution, which solves most of these problems. They are offering a range of frequencies and powers, and their units can be run from 12-volt solar cells.

They aren’t using CDMA, but have relied instead on frequency agile time-division duplex (TDD) with what appears to be GSM voice-coding and a variation on the European TDMA cellular signal modulation.

DaRT offers a two-line subscriber service for voice/fax, with data rates at 9.6 kbps. They’ve managed to add billing to the system by incorporating a pre-paid credit cache within the subscriber unit, which is triggered by metering pulses. The company also offers a larger “base-station” gateway into the wider telephone network.

So now that the billing obstacle is overcome, these technologies could well take off.

Stewart Fist ([email protected]) is an Australia-based, award-winning journalist and columnist, and author of The Informatics Handbook