Slashdot Mirror

Back in March there was a bit of discussion on the wireless-longhaul list regarding setup in Nepal.

http://openict.net/pipermail/wireless-longhaul/

The list-archive front end seems to eat much of the text, but it's all there in the gzipped archive:

http://openict.net/pipermail/wireless-longhaul/200 4-March.txt.gz

Back in March there was a bit of discussion on the wireless-longhaul list regarding setup in Nepal.

http://openict.net/pipermail/wireless-longhaul/

The list-archive front end seems to eat much of the text, but it's all there in the gzipped archive:

http://openict.net/pipermail/wireless-longhaul/200 4-March.txt.gz

Hardly ivory tower.

If you take some time to look at Longhaul Wireless Networks That Really Work(ed) you'll see that they are all ground up projects.

There's many "impose from above" technology in development, but Wi-Fi doesn't have to be one of them.

simon

> Wireless services based on Wi-Fi cost less to deliver than do services
> offered through other broadband technologies such as DSL and 3G
> (third-generation) wireless, Gelsinger said, making Wi-Fi "the only
> way to build a broadband infrastructure" in developing nations. Wi-Fi
> is an interoperability specification for wireless LAN technology based
> on the IEEE 802.11 standards, but is often used loosely as a synonym
> for wireless LAN technology in general.
>
> However, many of those nations are taking actions that are detrimental
> to Wi-Fi development, he argued.
>
> "We're seeing developing nations be the slowest and the most
> conservative in terms of making unregulated, unlicensed spectrum
> available," he said. "We see this idea of a scarcity mentality, this
> 'We have this spectrum, we're holding onto it and maybe getting a few
> dollars from licensing it.' "
>
> Gelsinger later clarified his remarks, saying that by "unregulated" he
> doesn't mean governments should take an entirely hands-off approach
> toward overseeing spectrum allocation, but rather that governments
> should set aside spectrum bands with no end-user licensing
> requirements for wireless device use, as the Federal Communications
> Commission has done in the U.S.

> Wireless services based on Wi-Fi cost less to deliver than do services
> offered through other broadband technologies such as DSL and 3G
> (third-generation) wireless, Gelsinger said, making Wi-Fi "the only
> way to build a broadband infrastructure" in developing nations. Wi-Fi
> is an interoperability specification for wireless LAN technology based
> on the IEEE 802.11 standards, but is often used loosely as a synonym
> for wireless LAN technology in general.
>
> However, many of those nations are taking actions that are detrimental
> to Wi-Fi development, he argued.
>
> "We're seeing developing nations be the slowest and the most
> conservative in terms of making unregulated, unlicensed spectrum
> available," he said. "We see this idea of a scarcity mentality, this
> 'We have this spectrum, we're holding onto it and maybe getting a few
> dollars from licensing it.' "
>
> Gelsinger later clarified his remarks, saying that by "unregulated" he
> doesn't mean governments should take an entirely hands-off approach
> toward overseeing spectrum allocation, but rather that governments
> should set aside spectrum bands with no end-user licensing
> requirements for wireless device use, as the Federal Communications
> Commission has done in the U.S.

Remember...with that 500 kilometer straight line connection, one box dying brings down the entire network...

> The reason you need a 30 m tower is because of the curvature of the earth.

Actually, according to this link, they would need to be over 50 m high:
http://www.oreillynet.com/cs/weblog/view/wl g/197

> I'm not sure how it follows that a tower of $5000 pushes the price
> into six digits.

$5000 x 20 towers = $100,000. And this is just for the backbone.

>> The fact remains that longhaul wireless can reach beyond the grid,
>> and battery operated, generator charged, or solar power can keep them
>> running.
>
> This I simply cannot fathom. The solar panels and batteries described
> in
> the Bhutan project (http://www.bhutan-notes.com/clif/) cost $4500, and
> that's just for one repeater!

$4500 is a lot but it's not a showstopper.

Senegal is 192 000 km^2. Let's say that there is one WiFi mesh tower
ever 20km in a square grid completely covering senegal.
20^2 = 400 km^2 per tower
192 000 / 400 = 480
480 * $4500 = $2 160 000

Remember...with that 500 kilometer straight line connection, one box dying brings down the entire network...

> The reason you need a 30 m tower is because of the curvature of the earth.

Actually, according to this link, they would need to be over 50 m high:
http://www.oreillynet.com/cs/weblog/view/wl g/197

> I'm not sure how it follows that a tower of $5000 pushes the price
> into six digits.

$5000 x 20 towers = $100,000. And this is just for the backbone.

>> The fact remains that longhaul wireless can reach beyond the grid,
>> and battery operated, generator charged, or solar power can keep them
>> running.
>
> This I simply cannot fathom. The solar panels and batteries described
> in
> the Bhutan project (http://www.bhutan-notes.com/clif/) cost $4500, and
> that's just for one repeater!

$4500 is a lot but it's not a showstopper.

Senegal is 192 000 km^2. Let's say that there is one WiFi mesh tower
ever 20km in a square grid completely covering senegal.
20^2 = 400 km^2 per tower
192 000 / 400 = 480
480 * $4500 = $2 160 000

Remember...with that 500 kilometer straight line connection, one box dying brings down the entire network...

> The reason you need a 30 m tower is because of the curvature of the earth.

Actually, according to this link, they would need to be over 50 m high:
http://www.oreillynet.com/cs/weblog/view/wl g/197

> I'm not sure how it follows that a tower of $5000 pushes the price
> into six digits.

$5000 x 20 towers = $100,000. And this is just for the backbone.

>> The fact remains that longhaul wireless can reach beyond the grid,
>> and battery operated, generator charged, or solar power can keep them
>> running.
>
> This I simply cannot fathom. The solar panels and batteries described
> in
> the Bhutan project (http://www.bhutan-notes.com/clif/) cost $4500, and
> that's just for one repeater!

$4500 is a lot but it's not a showstopper.

Senegal is 192 000 km^2. Let's say that there is one WiFi mesh tower
ever 20km in a square grid completely covering senegal.
20^2 = 400 km^2 per tower
192 000 / 400 = 480
480 * $4500 = $2 160 000

Remember...with that 500 kilometer straight line connection, one box dying brings down the entire network...

> The reason you need a 30 m tower is because of the curvature of the earth.

Actually, according to this link, they would need to be over 50 m high:
http://www.oreillynet.com/cs/weblog/view/wl g/197

> I'm not sure how it follows that a tower of $5000 pushes the price
> into six digits.

$5000 x 20 towers = $100,000. And this is just for the backbone.

>> The fact remains that longhaul wireless can reach beyond the grid,
>> and battery operated, generator charged, or solar power can keep them
>> running.
>
> This I simply cannot fathom. The solar panels and batteries described
> in
> the Bhutan project (http://www.bhutan-notes.com/clif/) cost $4500, and
> that's just for one repeater!

$4500 is a lot but it's not a showstopper.

Senegal is 192 000 km^2. Let's say that there is one WiFi mesh tower
ever 20km in a square grid completely covering senegal.
20^2 = 400 km^2 per tower
192 000 / 400 = 480
480 * $4500 = $2 160 000

If you have any kind of hills at all wireless should be an option. Get a nice tall tower up on top of your local hill and put an omni antenna up there. Then, at each house, point a nice high gain antenna at the tower. That's the usual star design. If that doesn't cover it, you can bridge the network into multiple stars, create for example a backbone that jumps from the T1 to a tower then to other towers or well-located houses. It will be a LOT cheaper than running any kind of new cable given the distances you quoted. Wifi you can get the kit for each house as low as $200-$300 and put up a repeater with a couple of radios and antennas for $1000 or so.

The only mailing list that I know of dedicated to long-distance WiFi (802.11) links is wireless-longhaul. You can subscribe here. There's also a Wiki with plenty of links to projects that have successfully deployed long-distance wireless networks in all kinds of different places.

Don't go into proprietary wireless unless you absolutely have no other option. There's some interesting new technology that's already available e.g. from Alvarion, using OFDM you can make non-line of sight connections at microwave frequencies. Eventually there will be 802.16 standards for them but right now it's not ready yet. The proprietary solutions are many thousands of dollars for each box.

simon

If you have any kind of hills at all wireless should be an option. Get a nice tall tower up on top of your local hill and put an omni antenna up there. Then, at each house, point a nice high gain antenna at the tower. That's the usual star design. If that doesn't cover it, you can bridge the network into multiple stars, create for example a backbone that jumps from the T1 to a tower then to other towers or well-located houses. It will be a LOT cheaper than running any kind of new cable given the distances you quoted. Wifi you can get the kit for each house as low as $200-$300 and put up a repeater with a couple of radios and antennas for $1000 or so.

The only mailing list that I know of dedicated to long-distance WiFi (802.11) links is wireless-longhaul. You can subscribe here. There's also a Wiki with plenty of links to projects that have successfully deployed long-distance wireless networks in all kinds of different places.

Don't go into proprietary wireless unless you absolutely have no other option. There's some interesting new technology that's already available e.g. from Alvarion, using OFDM you can make non-line of sight connections at microwave frequencies. Eventually there will be 802.16 standards for them but right now it's not ready yet. The proprietary solutions are many thousands of dollars for each box.

simon

If you have any kind of hills at all wireless should be an option. Get a nice tall tower up on top of your local hill and put an omni antenna up there. Then, at each house, point a nice high gain antenna at the tower. That's the usual star design. If that doesn't cover it, you can bridge the network into multiple stars, create for example a backbone that jumps from the T1 to a tower then to other towers or well-located houses. It will be a LOT cheaper than running any kind of new cable given the distances you quoted. Wifi you can get the kit for each house as low as $200-$300 and put up a repeater with a couple of radios and antennas for $1000 or so.

The only mailing list that I know of dedicated to long-distance WiFi (802.11) links is wireless-longhaul. You can subscribe here. There's also a Wiki with plenty of links to projects that have successfully deployed long-distance wireless networks in all kinds of different places.

Don't go into proprietary wireless unless you absolutely have no other option. There's some interesting new technology that's already available e.g. from Alvarion, using OFDM you can make non-line of sight connections at microwave frequencies. Eventually there will be 802.16 standards for them but right now it's not ready yet. The proprietary solutions are many thousands of dollars for each box.

simon

You probably haven't heard about it, but there's a fat pipe running down the coast of Africa with 20Gbps (yes, that's GIGA) of capacity. I've been following this story, and it's being wasted.

Home Page

Map

The max capacity of the cable is 120Gbps. It cost 0.65 billion to build and was a monumental sign of pan-african development 6 years ago when they bought it. Now it's finally in place.

technical

It's being wasted! It's a fat pipe, it's got something like 20x the bandwidth previously available in Africa (seriously...) but despite the obvious -- to me -- benefits to start using it Right Now, instead nothing seems to be happening.

Analysis: "...the benefits of this new capacity will not be unleashed on the national business environment"

The state telcoms in all these countries that control the access ... they are just going to sit on their hands because they have no vested interest. It's not on their radar screen to do anything with this cable, or to start selling access to regular people, like Guido. Instead, they are all slowly or quickly going out of business and dragging the market down with them. Installing the cable was only half of it. The other half is freeing the bandwidth.

simon

::: Check out rural wireless 802.11 on the wireless-longhaul@openict.net mailing list. subscribe or check out the project page :::

You probably haven't heard about it, but there's a fat pipe running down the coast of Africa with 20Gbps (yes, that's GIGA) of capacity. I've been following this story, and it's being wasted.

Home Page

Map

The max capacity of the cable is 120Gbps. It cost 0.65 billion to build and was a monumental sign of pan-african development 6 years ago when they bought it. Now it's finally in place.

technical

It's being wasted! It's a fat pipe, it's got something like 20x the bandwidth previously available in Africa (seriously...) but despite the obvious -- to me -- benefits to start using it Right Now, instead nothing seems to be happening.

Analysis: "...the benefits of this new capacity will not be unleashed on the national business environment"

The state telcoms in all these countries that control the access ... they are just going to sit on their hands because they have no vested interest. It's not on their radar screen to do anything with this cable, or to start selling access to regular people, like Guido. Instead, they are all slowly or quickly going out of business and dragging the market down with them. Installing the cable was only half of it. The other half is freeing the bandwidth.

simon

::: Check out rural wireless 802.11 on the wireless-longhaul@openict.net mailing list. subscribe or check out the project page :::

And if you're referring to Wi-Fi, then that's also not feasible, since 802.11b only goes about 10 miles max, even with a line-of-sight high-gain directional antenna. You'd have to put repeaters all over the countryside.

Err.. no. Wi-Fi works perfectly well out to 25km and longer distances have been done. For references have a look here and look at the archives for the wireless-longhaul mailing list.

Also Senegal isn't that big a country ... with 25km hops you can go border to border (about 500km) with 20 hops. Each hop is a tower, a couple of antennas and a couple of APs ... costing maybe $1200 total.

The solution, I think, is not technical but rather economical. Instead of building some fancy wireless mesh network, developing countries should work to deregulate the phone industry and open it up to competition, thus lowering the cost of Internet access everywhere.

That would be "political" actually, I agree but you're missing a couple of other important policy needs. The country must normalize their spectrum with the international open spectrum bands to make sure WiFi is legal. The other policy need is to make sure that VoIP is fully legal and unrestricted, since it makes much more efficient use of existing resources, thus dropping call costs and allowing more people to make calls.

simon

And if you're referring to Wi-Fi, then that's also not feasible, since 802.11b only goes about 10 miles max, even with a line-of-sight high-gain directional antenna. You'd have to put repeaters all over the countryside.

Err.. no. Wi-Fi works perfectly well out to 25km and longer distances have been done. For references have a look here and look at the archives for the wireless-longhaul mailing list.

Also Senegal isn't that big a country ... with 25km hops you can go border to border (about 500km) with 20 hops. Each hop is a tower, a couple of antennas and a couple of APs ... costing maybe $1200 total.

The solution, I think, is not technical but rather economical. Instead of building some fancy wireless mesh network, developing countries should work to deregulate the phone industry and open it up to competition, thus lowering the cost of Internet access everywhere.

That would be "political" actually, I agree but you're missing a couple of other important policy needs. The country must normalize their spectrum with the international open spectrum bands to make sure WiFi is legal. The other policy need is to make sure that VoIP is fully legal and unrestricted, since it makes much more efficient use of existing resources, thus dropping call costs and allowing more people to make calls.

simon

Senegal is a great place to mount an initiative like this because they have access to some serious bandwidth. The SAT-3/WASC/SAFE undersea fibre optic cable landed in Senegal last year, delivering multiple gigabits of internet bandwidth. This is in contast to the previous situation, where basically all of Africa had less bandwidth than the headquarters of my former employer.

But the idea of wiring Senegal is all wrong. What's needed is wireless. Wireless internet (e.g. 802.11b Wi-Fi) is a far more appropriate solution in a country like Senegal where traditional wireline infrastructure is going to be subject to harsh environmental conditions and being destroyed by political unrest. Wi-Fi long-distance links can span 30 km in a single hop, and the towers like cell towers can be powered with generators. Wi-Fi delivers true broadband, 802.11b is 10Mbps, and 802.11a and 802.11g can deliver more like 30Mbps.

Broadband is essential. With broadband you can deliver the killer app (yeah, I said killer app) of the rural internet which is Voice over IP. People in Senegal, well, the literacy rate isn't so high, and VoIP is what's really useful to people as it allows them to call members of their family who may be off making money in other parts of the world, to call into town to check crop prices, to call their relatives in the city. Of course this requires policy to make sure that VoIP is legal and that the national telco doesn't try to block it to protect their own profits.

If you're interested in wireless long-distance links, you might be interested in a mailing list on the subject, wireless-longhaul@openict.net. You can subscribe here, and the mailing list home page is here.

Senegal is a great place to mount an initiative like this because they have access to some serious bandwidth. The SAT-3/WASC/SAFE undersea fibre optic cable landed in Senegal last year, delivering multiple gigabits of internet bandwidth. This is in contast to the previous situation, where basically all of Africa had less bandwidth than the headquarters of my former employer.

But the idea of wiring Senegal is all wrong. What's needed is wireless. Wireless internet (e.g. 802.11b Wi-Fi) is a far more appropriate solution in a country like Senegal where traditional wireline infrastructure is going to be subject to harsh environmental conditions and being destroyed by political unrest. Wi-Fi long-distance links can span 30 km in a single hop, and the towers like cell towers can be powered with generators. Wi-Fi delivers true broadband, 802.11b is 10Mbps, and 802.11a and 802.11g can deliver more like 30Mbps.

Broadband is essential. With broadband you can deliver the killer app (yeah, I said killer app) of the rural internet which is Voice over IP. People in Senegal, well, the literacy rate isn't so high, and VoIP is what's really useful to people as it allows them to call members of their family who may be off making money in other parts of the world, to call into town to check crop prices, to call their relatives in the city. Of course this requires policy to make sure that VoIP is legal and that the national telco doesn't try to block it to protect their own profits.

If you're interested in wireless long-distance links, you might be interested in a mailing list on the subject, wireless-longhaul@openict.net. You can subscribe here, and the mailing list home page is here.