Meshnet

From Makers Local 256
Jump to: navigation, search

Creator:
hfuller
Status:
Planning
Born On:
19:38, 26 September 2016 (CDT)
Last Updated:
17:33, 06 April 2018 (CDT)

This project aims to create an accessible mesh network for Huntsville. This network would operate independently from the Internet and ideally would not rely on the Internet for any of its functions.

Goals

These goals are roughly in order of descending priority.

  1. The mesh should provide connectivity between mesh participants.
  2. The mesh should be as wide-reaching as possible.
  3. Joining the mesh should be accessible and understandable to the general public.
  4. The mesh should integrate with participants' Internet CPE (customer premises equipment) if they wish to do so. If they do not, it should also be capable of operating independently.
  5. The mesh should provide access to communication services that can be used even when the Internet is inaccessible for some or all participants.
  6. The mesh should provide access to existing Internet overlays with the spirit of Maker culture, such as dn42, as well as those that promote anonymity and privacy, such as the I2P or the Tor Project.

Strategy

Here's the current strategy plan for each of the goals

  1. Basic connectivity
    • We will build the links using commodity point-to-point and point-to-multipoint hardware.
      • Popular vendors include Ubiquiti and Mikrotik.
      • Popular spectrums include 900MHz, 2.4GHz, and 5.2-5.7GHz.
    • We will then use a protocol that can automatically select a path between two nodes.
      • We can use the normal Spanning Tree Protocol, but it might require a lot of configuration to be optimal. It is pretty dumb by default.
      • Or we can use batman-adv, but the compatibility will be worse. (We couldn't use Mikrotik RouterOS for this, for instance.)
  2. Coverage
    • At first, we will light some fairly long-range links, in order to get some "islands" of connectivity spread throughout Five Points and the neighboring areas.
    • Then, we can do outreach to the areas surrounding the "islands" in order to get more participants and a wider range of nodes.
    • Finally, the nodes will interconnect further, creating redundant paths for reliability and speed.
  3. Accessibility
    • The creators/maintainers of the mesh should be able to hand off CPE to participants in a state that requires little maintenance or configuration.
  4. Integration
    • A static route can be added to almost all CPE provided by Internet service providers. A static route would be used in order to make the IP space of the mesh accessible from the participants' "normal" wireless networks, if desired.
    • If that's not desired, the mesh CPE would need to broadcast a small (maybe 5GHz-only) network so that they can connect to that to use the mesh.
    • Or maybe even that will not be necessary, as some participants may wish only to further the purpose of the mesh without providing a point of connection.
  5. Services
    • Makers Local can, on-premises, host some communication services for mesh users.
  6. Peerings
    • These peerings could be located anywhere, but will likely be located at Makers Local, at least initially.
    • Redundancy could later be added, in case Makers Local loses its Internet connection.

Links (of the radio variety)

From To Frequency Status
ML HF 900MHz Pending proof-of-concept testing.
ML HF 2.4GHz Proven infeasible with NanoStation panel antennas. Need to try directional antenna.
ML SA 900MHz Weak link came up. No further testing conducted.
ML SA 2.4GHz Pending proof-of-concept testing. Will almost certainly require directional antenna.
ML WO 900MHz Pending proof-of-concept testing.
ML JR 900MHz Pending proof-of-concept testing.

Notes

Of course, we almost always want to set up the link with the highest frequency possible because it will provide more bandwidth and utilize cheaper gear. So here's the theory:

  1. Use expensive 900MHz gear to get a rudimentary initial link set up, usually using the built-in semi-directional panel antenna (in the case of the NanoStation Loco M900).
  2. Aim this link as precisely as possible at both ends.
  3. Bust out the 2.4GHz radios and set them up right next to the 900MHz units.
  4. Point the 2.4GHz units in the exact same direction as the 900MHz ones and see if the link comes up. Fiddle with settings and antenna combinations as necessary.
  5. If / Once the 2.4GHz link is stable, remove the 900MHz gear and reuse it at another site.

This strategy can be reused for moving to 5GHz, if possible.

On 5 Dec 2017, another strategy was discussed - that of point-to-point laser links. They are not susceptible to the Fresnel zone, which requires vertical clearance for radio communications, nor are they susceptible to radio interference, but they require perfect line of sight. Open source projects like RONJA and KORUZA achieve this, but the assembly is hellacious. There are commercial products from companies like LightPointe we might be able to acquire from the used market for this purpose. Finally, a visitor to the make shop brought up the simple use of generic laser emitters and focusing gear, purchasable for general purposes from the Internet. The transmitter could be connected to a RS232 UART on a single-board computer or microcontroller, and the receiving diode could similarly be connected to the UART on the other end. Even if only relatively slow speeds could be attained, a link of at least dial-up quality could potentially be established at a very low cost.

Road Map (actually in no particular order)

  • Testing 900MHz to HF site because I am selfish.
  • Testing 900MHz to WO site because of its potential as a relay location.
  • Re-testing 900MHz to JR site because of its potential as a relay location. (Initial tests were botched such that they produced no results.)
  • Investigate placing a radio at the Five Points intersection to act as an uplink for a camera, as well as potentially a relay station.
    • The light pole next to Holmes
    • Tenders
    • Star Market
    • Gas station?
    • Any other nearby locally-owned businesses...?

Hardware

  • Point-to-point hardware for outdoor
    • Ubiquiti NanoStations
      • Pros: Cheap and readily available. Can often run OpenWRT, which could eliminate the need for an indoor meshnet CPE aside from the user's existing network
    • MikroTik RBLDF-5nD
      • Pros: Cheap. Reuse a satellite dish as a directional antenna.
      • Cons: Only available in 5GHz.
  • Indoor CPE
    • MikroTik RB951Ui-2nD
      • Pros: Has five Ethernet interfaces. Passes through PoE (switchable I believe).
      • Cons: Kind of expensive. Does not come with a PoE injector, but could be used as one with an external 24V brick.
    • MikroTik RB750UPr2
      • Pros: Has five Ethernet interfaces. Takes PoE on one of them and passes through to the other four! Could be used for a site with multiple PtP links.
      • Cons: Even more expensive than the RB951Ui-2nD, and no wireless. OpenWRT doesn't look that mature for this.
    • MikroTik RB941-2nD
      • Pros: Has five Ethernet interfaces. CHEAP.
      • Cons: No PoE support at all. Not that much RAM for OpenWRT and it looks hard to install.
    • MikroTik RB931-2nD
      • Pros: Has three Ethernet interfaces. CHEAP. Tiny.
      • Cons: Not that much RAM. Can't run OpenWRT(?).
    • Ubiquiti AirGateway
      • Pros: Compact. Cheap and readily available. Docks with the standard UBNT power injector. Can run OpenWRT.
      • Cons: Only two Ethernet interfaces. Hardware is kind of wimpy for OpenWRT and the stock OS is limited.
    • MikroTik RBmAP-2nD
      • Pros: Accepts 802.3af/at PoE or DC barrel power, and passes through passive PoE. Compact.
      • Cons: Only two Ethernet interfaces. Kind of expensive. Can't run OpenWRT(?).

Resources

Retrieved from "https://256.makerslocal.org/wiki/index.php?title=Meshnet&oldid=22663"