Painful lesson in OLPC mesh networking for Mongolians

Painful lesson in OLPC mesh networking for Mongolians

Summary: The Mongolians have had a painful lesson on mesh networking according to the OLPC current events webpage.  Broadcast storms in the overly dense mesh environment along with excessive mDNS broadcast traffic seem to have crippled the Gobi desert experiment.

SHARE:
TOPICS: Wi-Fi, Networking
40

The Mongolians have had a painful lesson on mesh networking according to the OLPC current events webpage.  Broadcast storms in the overly dense mesh environment along with excessive mDNS broadcast traffic seem to have crippled the Gobi desert experiment.  Here's an excerpt:

We have painfully discovered the limitations of the mesh and current collaborative software in Mongolia, where the convolution of the number of laptops with bugs #5335 (more mDNS traffic than expected) and #5007 (mesh repeats multicast too much) make the perfect storm, which prevents anybody from using the network. We will continue to improve the mesh performance, but clear guidelines are needed as to what network infrastructure to deploy under what conditions. Once a certain density of students is exceeded, a wired backbone and conventional access points will be required.

The limitations of mesh topology are well known in the wireless engineering community and I've raised the issue and pointed out the limitations last September.  Each mesh hop you add increases the propagation delay as well as multiply the radio traffic and congestion.  Performance on a mesh network is fundamentally many times slower than a non-mesh network and when the density gets high enough, the system simply breaks down.

When on a tight budget, I had always recommended the usage of a cheap $60 router running open source DD-WRT would have sufficed and you get a free router with it which you need for IP sharing anyways.  The addition of a high-powered antenna would allow the access point to hear distant signals from faint clients and it will amplify the broadcast signal.  A simple in-door $26 9 dBi antenna placed up high can easily cover a small school.  A $60 12 dBi outdoor antenna positioned on the roof would easily cover an entire campus.  If you put two centralized Access Points and large antennas on channel 1 and 11 (avoid adjacent channels because of channel bleeding) in the 2.4 GHz spectrum, you can load balance and have redundancy if one set of AP/antenna fails.

My fellow blogger and teacher Chris Dawson feels that the ability to do peer-to-peer collaboration with or without an Access Point has great potential.  But peer-to-peer wireless collaboration could have been done with regular ad hoc networking technology without the expense or problems of a full 802.11s mesh implementation.

The inclusion of full 802.11s stack has been challenging.  The need for a radio system that stays on and continues to forward packets even while the laptop is off added unnecessary expenditure to the OLPC XO and it unnecessarily drains the laptop batteries.  When you multiply this expense and complexity across all the clients and realize that the wireless access point comes free with the router, it becomes clear that this may not have been the best design decision.

Topics: Wi-Fi, Networking

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

Talkback

40 comments
Log in or register to join the discussion
  • You all Jump around too much!!

    The biggest problem I see with the Mesh is that a hop basically jumps around from computer to computer looking for a way back home. That is a lot of jumping. Then you have the fact that one in how ever many is the AP for how many different OLPCs. Not that the under powered little guy couldn't server as a router, but after a few hundred or more decide to route through him, he gets pretty busy. In which case, I would hate to be the rich kid on the block with Internet Access.

    Oh well.
    nucrash
    • Just imagine 2 mesh nodes relaying

      Node 1 sends message to node 4 but it has to go through node 2 and 3. Node 2 repeats to node 3 but can't talk when node 1 and node 3 are talking. You have effectively slashed your throughput by 3 at the very least.
      georgeou
  • One size does not fit all

    Obviously there will be growing pains with the networking topology and yes there are trade-offs when one goes with a mesh network.

    The thing is though, your solution always requires an additional piece of hardware, a power source for that hardware, and someone to maintain it. Since the OLPC is being marketed at areas where all of those items are in limited supply you might be surprised to find that your first-world solutions aren't always going to be practical.

    (BTW - a solution that requires you to go outside of your design restrictions is not a valid solution.)
    Robert Crocker
    • If you have power for the satelite uplink and modem

      If you have a power source (solar, wind, generator, infrastructure) for the satelite uplink and modem, then you have power for your 3-5 watt wireless router.
      georgeou
  • Another point/reason for the mesh

    Another design point was that OLPC laptops would help maintain the mesh even when they're not being used (this was also why they went with a particular chip-set even though it didn't have fully open source drivers). I'm not sure you could get that effect using ad-hoc networking.
    Robert Crocker
    • You missed my point, I don't want those features

      You missed my point, I don't want off-mode mesh relay because it drives up the cost/complexity of the product and it drains the power. You've done this for no good reason when all the peer-to-peer collaboration can be done without a full 802.11s stack. Neither the Nintendo DS nor the Sony PSP utilize 802.11s.
      georgeou
      • You missed THEIR point

        George,

        I'm really sorry that you don't like the OLPC design, but it wasn't designed for you. They came up with their set of design conditions and implemented the mesh WiFi to address those conditions.
        Robert Crocker
        • Tell me what those conditions are that match real-world needs

          You keep saying there are conditions. What are they? Name a condition that forced the need for mesh networks that mean something in the civilian world where practicality, performance, and low cost is paramount.
          georgeou
          • Go check the OLPC wiki

            They claim the 802.11s feature set is "Small enough to run
            autonomously on Marvell's 88W8388 802.11 wireless module":

            http://wiki.laptop.org/go/Mesh_Network_Details

            Implicitely saying it doesn't consume that much power...

            You also have to understand that 802.11s is capable of
            supporting many more nodes than regular ad-hoc technologies
            (which you suggest they should have adopted, in the
            next-to-last paragraph of your blog post). Even if their
            implementation of 802.11s stops working once a certain density
            of nodes is reached (as they report in Mongolia), it already
            supports many more nodes than what is possible with regular
            ad-hoc technologies: 802.11s already outperforms them. So of
            course it was the right technology to adopt for a peer-to-peer
            wireless network.
            zpdixon 42
          • Nope, there's a way to set up one virtual Access Point on a laptop

            Nope, there's a way to set up one virtual Access Point on a laptop and allow all the other machines to connect to it with far less overhead. That or a regular access point far out performs any 802.11s topology.
            georgeou
          • Virtual APs: not the solution for the use cases envisioned by the OLPC team

            If a real AP is within proximity -> no need for a virtual AP or
            802.11s.

            If not -> 802.11s is better because a mesh network covers an
            area larger that what a single virtual AP could cover. One of
            the main design desicion of the OLPC team was to optimize cases
            where only a handful of laptops are present in a medium/large
            area (think of a village). What hapenned in Mongolia was a
            rather exceptional scenario ("500 laptops under the same roof":
            http://wiki.laptop.org/go/News)
            zpdixon 42
          • From the OLPC news

            10. Mesh: Cramming 500 laptops under the same roof is a difficult (but tractable) engineering problem. We haven't done any testing of such deployment scenarios and Mongolia is not really the most convenient place for that testing. Despite that, common sense can still carry us a long way. We have set the limit of XO laptops to school servers to 180 (60 per channel in mesh mode)?after optimizing the laptop for ?dense? deployment (which hasn't been a priority in our software development schedules). However, deploying more school servers under the same roof doesn't immediately translate to increased capacity, since school servers don't add spectrum. While a school server still costs several hundred dollars, it is more economical to install standard low-cost access points instead of multiple servers. [b](The OLPC mesh implementation was to maximize the ?connected? time for sparse deployments (children in villages in Cambodia, rural schools in Rwanda) and to simplify and extend connectivity away from an access point.[/b]

            http://wiki.laptop.org/go/News

            So, the "to simplify and extend connectivity away from an access point" would certainly seem to be a design requirement then.
            Robert Crocker
          • A $30 antenna would do a much better job extending

            "So, the "to simplify and extend connectivity away from an access point" would certainly seem to be a design requirement then."

            A $30 antenna would do a much better job extending the network with 360 degrees coverage. As I explain here http://blogs.zdnet.com/Ou/?p=777, you need 10s or 100s of XOs positioned optimally to replace a single well placed antenna.
            georgeou
          • And don't be so arrogant

            I'm wondering what your "real world" experience is with deploying laptops in Cambodia and Ruwanda are. We're not talking about a solution for use here in the US.
            Robert Crocker
          • Wireless is the same all over

            You have dense deployments, you have sparse deployments. In all cases, AP architecture with some fixed and narrowly defined mesh bridges makes the most sense. OLPC is the only organization that insists on using 802.11s client-side mesh. I've spoken to other professors who deploy in poorer nations that agree with me.
            georgeou
          • Infrastructure is NOT the same all over

            You continue to think that everywhere this stuff will go there will be readily available power and support for setting up infrastructure and that's just not the case.

            I'd be interested to see what design restrictions these professors you talk to are operating under and what usage patterns they're expecting.
            Robert Crocker
          • Power issues can be solved very cheaply

            http://www.amazon.com/Northern-Industrial-Tools-Wattage-Panels/dp/B000K2HNZS/ref=pd_sim_auto_shvl_title_5/002-9646450-8008013
            $80 for a 15 watt solar panel which can operate quite a few Access Points and Cell Phone tower equipment.

            http://www.amazon.com/exec/obidos/ASIN/B000C1Z2VE/
            400 Watt wind tower for $500 which works day and night. I'm sure a smaller/cheaper version can be made since 400W isn't really necessary.
            georgeou
          • Censoring

            "No further replies to this post will be accepted." What's the point of a "blog" then, hmm?
            zpdixon 42
          • I see you found the workaround. We do not censor.

            It's a limitation on all of us. I don't/can't sensor anyone. You just reply to the parent thread which allows you to keep posting.
            georgeou
          • You're adding complexity without reliability

            Solar power is obviously dependent on sun light. Wind power of course required wind. So now your $60 access point requires an $80 solar power panel and some battery array to deal with when we don't have sun light. You've driven your complexity up greatly in the pursuit of your AP's shall solve ALL problems approach.
            Robert Crocker