I’m falling hopelessly behind. That means I’m posting something closer to my rough notes. Filling in the blanks is left as an exercise for the reader.
Panelists: John Atkinson (Wireless Ghana), David Young (CUWiN), Michael Peralta (Tribal Digital Village), Arun Mehta (Radiophony), Bogdan Tancic (BGWireless)
Bogdan
- Need to improve technology or wireless networks at the physical layer to increase throughputs.
- BGwireless community network infrastructure
- Node consists of one AP, omni-directional antenna, point-to-point link to another node, can handle about 20 users/node and coverage is approx 1km.
- Use HG24 (HyperGain) to link hops together.
- Get 40-50Mbps
- Can actually run high-bandwidth applications over this.
- This is a serious wireless network. Very different from the urban groups that are looking at providing connectivity everywhere to support a connected lifestyle.
- BGWireless has a workshop where they make their own antennas. They have information about their antennas on the website.
Arun
Teaches programming to blind people – to the blind, the computer is as important a development as written language was for non-blind. But screen readers are not available in every language. So a computer is advanced technology for many people
What we consider to be advanced today we may consider primitive in two years.
We are at a critical juncture (like microcomputers were in 70’s). There’s a difference between what we had then and what we have now – then the big companies thought micros were a joke, they figured they’d disappear, no one considered them a threat. Today, the big telecoms are worried (and fighting). They do see the potential. And telecom is regulated, and regulations are different in different parts of the world.
How do we advance wireless networks? One of the most interesting developments is the hack of the Altheros chip set; we’re now able to take commercially available hardware and hack it to do completely different things. This is particularly important in developing countries; it’s not easy for the police to see that you’ve gone in and changed code to do something that would be illegal. Even more we could use steer-able antennas to allow for better use of spectrum (highly directional antennas while working in omni field)
GNU Radio – FBGA integrated circuit that can be programmed/changed. You can do anything at all. You can make advanced telecom devices – it’s like the IBM PC compatible. Open source, fully programmable mobile phone on steroids. And you don’t have to have advanced telecommunications training to program them thanks to middleware. Great device for teaching students how to do telecom programming, how to make stuff. If you’re interested in rural communications in developing countries, and there are many different standards (CDMA, GPRS…) – if GNU Radio can detect what type of phone that’s come into range and provide service.
Enables working in frequencies that are much lower frequency and allow for connections without line of sight requirements.
Free Space optics (wireless optic communications) No regulations issues, efficient, easier to focus. You can shorter masts. If you use infrared, no one can see what you’re doing. Optics is not black magic, we’ve been dealing with it for centuries, one of the most expensive things in broadcasting is the pre-amp—in the case of light, your pre-amp is simple, it’s a passive device, it’s called a lens. We’re used to optics. (low power, spectrum is unregulated) Can you imagine the Internet today with its 500% annual growth without optic technology? We’re going to be wanting this capability in wireless. We can’t run fibre to every house in a third world country, but we could do optical wireless.
What’s holding up the optical wireless uptake? Us, we’re just not doing any research; people have not been interested. We need to start some real research into this in order to overcome the problems that will come with. This is not a new technology; the point is building critical mass of interest. RF is good, but there are limits to what RF can provide for bandwidth. We need to look to the future.
Print disabled – also interested in this problem; software that works for visually impaired also works for illiterate (if the computer’s reading to you, you don’t need to worry about not being able to read). So, things that can do audio are interesting…Asterisk server (OpenSource PBX). Asterisk was also mentioned by Gabe as the tool being use to serve out the stories in his [murmur] project.
David
Problems with mesh networks and some technologies that might help resolve them
1/ hops/distance from Internet
As you get farther away from the Internet gateway, your bandwidth is going to be dropping off because of interference. Can use more than one channel to retransmit to avoid collision. We advance the networks by using >1 channel on each node.
2/ Marginal links; a link where errors occur often enough that you usually have to retransmit in order to get past that link.
Have nodes collaborate to rebuild damaged packets : multi-radio diversity. I’m not sure if this is the reference Doug was thinking of, but there’s a paper by researchers at MIT here.
3/ Adaptive antenna systems at 900MHz band (new ubiquity.com cheap radio)
900 MHz helps to get around the problem of rain on trees…There is less interference with signal than 2.4GHz; 900MHz can get through more things. But it’s hard to build on small scale – it has really high gain; you might need have an antenna 3* as large in order to get the signal. It doesn’t allow you to really focus the signal.
4/ Important not to get too hung up on wireless; it’s an awesome technology, but it’s easy to forget about wire and we should look at how we can use it.
Use wire if you can. AC wire, CableTV go to almost everyone’s home. So why not power line networking? There are issues there, too.
John
He has problems and he’s interested in help – so he’s sharing the problems
In Ghana, there’s a link that needs to go 30K, in the TCP/IP handshake there can be problems with time out of the SYN before the ACK gets back so it resends. Of course, there’s a requirement that this be doable with inexpensive, non-enterprise equipment.
BGWireless has a 60K link and they changed their acktimeout. (discussion ensued). Right now, John’s system works, it’s just retransmitting packets a lot.
Could be the software. Others who have long links like this working are using MikroTik software.
Michael
Wanted to get native reservations in the county connected again (southern California tribal) 3 tribes divided into 18 reservations. Idea to bring broadband access to these rural communities.
Used 5.8 for long links (greater than 5K). Some links use 2.4 because can’t meet the power demands. The power is solar for all the backbone towers which are placed on the tops of mountains. Some sites have propane generators as a backup; lesson learned from an extremely rainy season in 2004 where they had 2 months of rain.
Had shadow project where took teens to learn everything they did and at end of project they were given the job of building 2 towers.
Sustainability: costs $8K/month for DH3 link—need to find ways to fund, have created some businesses to generate revenue.
Michael showed some photos of extreme weather conditions in which things are operating; rain, ice, snow, rockslides.
Technorati Tag:NS4CWN