On 8/3/07, <b class="gmail_sendername">Hilton Garcia Fernandes</b> <<a href="mailto:hgfernan@lsi.usp.br">hgfernan@lsi.usp.br</a>> wrote:<div><span class="gmail_quote"></span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
thanks for the enligthening, informative and thought provoking answer. for<br>instance, i failed to notice the Wireless Repeater pages.<br><br>now think of a school with not all class rooms filled with XO laptops.</blockquote>
<div><br>I think the potential is there and my comments are below, but lets clarify the deployment strategy. Given some value (say, all 6 year olds, or all primary school students. I don't know the precise plans), every member of the group in a school gets one. So it is reasonable to plan on clusters of students as it is not likely that schools have the same grade class on opposite ends of a building, etc.
<br><br>The other piece to consider is that the school server, which will do routing to the gateway, along as bridging the 3 unique wifi channel (1,6,11) will have semi-directional antennas, which should geographically split the school. That is, the north end of the school might be on the first school server antenna, the east and south on the second, and the west on the third. In deployment, strategic placement of the antennas and the school server is planned.
<br><br>Nevertheless, we must plan for poor wireless signals and connectivity gateways of some kind...<br></div><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
in the general case, it will be necessary to have a more powerful device<br>to connect the XOs to the school network, and then to the Internet. that<br>kind of device will be in the class rooms or school corridors and will
<br>provide the neighbor XOs with the desired connectivity.</blockquote><div><br>Another option if buying cabling is not out of the question (though in many of our cases, it is) a Cat cable from a teacher's XO back to the school server, etc. could be quite useful.
<br></div><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">that device usually will be a conventional 802.11g AP, since 802.11s cards<br>
are not widely available now.</blockquote><div><br>In fact there are *no* wireless 802.11s cards available publicly that I have heard of. I strongly suspect none exist. So it is for the time being, it will only be 802.11b
/g access points. <br><br>Eventually, the peripherals mentioned previously will speak native XO-friendly 802.11s to share the mesh.<br></div><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
due to 802.11 infrastructured mode, the AP<br>will create a NAT for its clients, and due to the hypothesis above, the<br>client of an AP will have no way to see directly the clients of other APs.</blockquote><div><br>The AP would not have to NAT connected devices. They would more likely act as a smart repeater/bridge.
<br></div><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">in the general case, the school will be an archipelago of AP-centered<br>islands. and that is not good for many of modern teaching techniques the
<br>XO will offer.</blockquote><div><br>Agreed. But I think the premise is flawed. Indeed, NAT in any use creates an unfortunate set of limitations (that is, next to none available) for end-to-end connectivity that these XOs really ought to have.
<br></div><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">so, the idea would be to use an AP-like device that joins the mesh and<br>lets all XOs talk with each other, without isolating them in islands.
</blockquote><div><br>This is indeed the purpose of the Wireless Repeaters, when they are available, as they will talk native-802.11s. In the meantime, I believe standard APs and a bit of designing will get a similar (though less efficient and adorable looking) result.
<br></div><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">if we use XOs for APs, the additional advantage will be that no wireline<br>structure will be necessary to connect each AP to the school network, as
<br>they will connect to each other, due to mesh.</blockquote><div><br>With WDS [1], we could merge multiple APs together and have them connect over their own wireless connection doing longer-haul connectivity to get back to the school server.
<br></div><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">so, this kind of architecture requires only power outlets, that will be<br>already available, due to the illumination. and not network outlets, that
<br>in general will be not available.</blockquote><div><br>Sure. I see precisely the solution you are trying to solve and I think it is a good one to tackle. In the short-term, proper placement of the School Server and classroom layout should let more than sufficient connectivity to exist. In the medium-term, things like the Wireless Repeater will be able to extend the mesh network natively, and do so without need for a power outlet as they will have solar panels and sub-watt power requirements.
<br><br></div></div>[1] <a href="http://en.wikipedia.org/wiki/Wireless_Distribution_System">http://en.wikipedia.org/wiki/Wireless_Distribution_System</a><br><br>-- <br>Michael Burns * Intern<br> One Laptop Per Child