Hackaday Prize Entry: A Community Mesh Network

While the Internet of Things is here to stay, and will kill us all, there are a few places left on the planet that will remain unscathed during the robot uprising. These underserved communities still have a need for communications and networking, leading [hlew] to create a Community Engagement Mesh Network as an entry for The Hackaday Prize.

While there are many, many options available for DIY networking solutions out there today, [hlew] is leaning on some work done by some of [Bruce Land]’s students at Cornell. This project used simple and cheap nRF24 radio modules for a true mesh network with multi-node communication, dynamic route discovery, and dynamic route reconfiguration.

The CEMN will rely on this network to provide communications to underserved communities. The primary goal of this network is to broadcast information like crop reports and health advisories, but it can also be used for peer to peer communications between individuals.

26 thoughts on “Hackaday Prize Entry: A Community Mesh Network

          1. 3592 JD Tape @ max of 15TB: 0.045 GB / cu mm

            SD card at max of 512GB: 0.317 GB / cu mm

            microSD card at max of 256GB: 1.552 GB / cu mm

            If you were willing to rely on microSD for your archival purposes (stupid but whatever), you could fit more than 517TB in the same volume as the 3592 tape. But, it’ll cost you around a quarter million dollars.

          2. The thread is about networking, this sub thread was about sneakernet options. However, taking your figures, you could settle for a lot less data and a lot more redundancy, and for read-only operations it may actually be a valid archive method, even if that wasn’t what we were discussing.

    1. Did something similar on my hi-desert property. Needed lo-pwr, lo-bw without wires for about 27 various sensor nodes scattered over 15 acres. Can handle multiple device failure and still find path to talk to sensor server (teensy 3.1). Latency varies, but max has been under 225mSec and does not scale linearly with distance.

      Root cause of most common failure modes for nodes is bird shit.

    2. There are two responses to that:

      A single high-speed connection is very good for getting a bunch of people on email, wikipedia, and other utilitarian resources. Not so hot for torrenting, Youtube, and Netflix, but if you just need to read an article you’re OK. This can spread the connection around without relying on infrastructure that needs much effort to be installed or removed. It’s also good for spreading a connection over a wider area or connecting a bunch of devices to a local area, plus connections like this are used in some clusterbot experiments.

      Second answer is a bit more round-about: Tensor Calculus was invented long before there was a use for it beyond novelty purposes, but Einstein needed something more powerful than what had been used for physics already and found Tensor Calculus fit the bill. So never underestimate the utility of a rubber chicken.

    3. Put an Outernet radio ( http://outernet.is ) on it to provide news weather and education materials received on a low speed satellite feed, Their new dreamcatcher SBC is a combo SBC and RTL/SDR with a L-band patch antenna that looks pretty cool though I wounder why they did not include a wired ethernet port on it.

  1. The problem I see in rural areas is that weak transmitters don’t cut it well. If all the houses are near each other then sure a few meters to a few hundred might make sense but the area’s I have had to work in there is typically terrain and foliage issues, with 1 house per square mile. So rather than dozens of small systems all costing say $25-$50 why not go for a few larger strategic points with more robust setups, then use other products to fill in a denser coverage. I personally have used 900mhz gear at speeds of 10’s of megabits over 5+miles this gear is inexpensive, meshes and has a decent amount of support. (Mikrotik makes a line that does well). Use this for the backhaul/middle mile. Last mile products seem to be abundant, (2.4ghz, 5ghz etc) In the one place that had high density all of the structures where metal and I had to deploy 5+ mesh nodes in a campground full of RV’s and Trees that made 2.4ghz usable to about 50 feet. I also put in larger more centralized node with an amp and high gain antenna to link to the other mesh nodes to keep hop counts reasonable.

    Or you can do something with “whitespace” radios like Carlson Wireless ruralconnect stuff, it’s spendy at a about $2000 per base station and access points around $500, but this would easily let you link up most stuff around a limited access point and then share it out using smaller meshes from there (Ruralconnect has a 10-25mile range). These products are all supported and cost about what you would pay to roll your own.

    As another commenter on the CEMN page suggested look into GSM.

    I am not sure what the project is trying to do that others haven’t done and aren’t already doing? I take this statement from the project “This is a key important element in an environment where ownership of the infrastructure by the locals is not guaranteed” – as way of saying locals don’t maintain the things they want/need. How is installing 50 devices that will all be neglected and unusable any different then 3 or 4 strategic points? This isn’t going to make people fix broken stuff, this just provides more stuff to break that needs fixing.

    Just my 2 cents and while I admire the objective I disagree with the intended implementation.

    1. “I take this statement from the project “This is a key important element in an environment where ownership of the infrastructure by the locals is not guaranteed” – as way of saying locals don’t maintain the things they want/need.”

      I would take that statement to imply the locals might be something like a resistance movement or a persecuted minority.

  2. For rural networks have been using ubiquity hardware to do this kind of work lately. Mostly using the nanobeam devices. Sadly its point to point. However it deals well with foliage based LOS issues. Up to 21km distance at around 50mbps max at that distance I have found. Have some 5mile runs with clear los still pushing 200-500mbps. I haven’t tried their new mesh APs yet. But it looks like anything of a significant distance you will want a point to point link to supplement the mesh nodes.

    Costs are not bad, double what this project is aiming for. At 100-150 for the nano beam devices. But they have been behaving well at this point. With limited problems.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.