Add New Page: You are not allowed to add pages Select section/namespace. New page title.
 

Project Implementation Challenges and Questions

Questions

What's the status of Commotion code that's checked in?

  • Not very well tested because of recent complete refactoring
  • script + OpenWRT package feed.

How do you get from 0 -> Mesh?

Downloads OpenWRT build tree and installs Commotion scripts.

Autconfigures:

  • SSID
  • IPv4 addresses for all interfaces
  • Mesh routing parameters
  • DHCP parameters
  • firewall
  • creates extensible hotpluggable script framework for adding additional service autoconfiguration
  • TODO: patching web interface to support custom mesh interfaces, custom web theme using sweet new logo, figuring out bugs with udhcpc, cleaning up some comments, more testing, less naive defaults for wireless interfaces, GUI config wizard, VPN support, secure AP support, automatic (ideally signed) remote update support, dashboard support, IPv6 and batmand/batman-adv autoconfig support. Integration of measurement tools which report back relevant wireless parameters to a collector which can then make proposals for channel assingments etc. (Better to push measurement tasks to a dashboard tool? (I believe measurment is an essential to have a good high performance stable network))

Does commotion provide channel/noise scanning/hopping?

Not yet.

  • Ben West said ROBIN does some of this. At least a provision to tell entire mesh to switch a channel.
  • Aaron: that is not enough to tell all boxes to switch to a new channel. Then they will all simply interfere with each other on a different channel.
  • Need AGC (autmatic gain control (txpower)) also
  • ?: Big challenge/need is to manage power w/ automatic gain control, so nodes only connect to closests nodes (like cell-phone). If everything's running at full-power, it gets really noisy!
  • Ruckus Wireless offers some interesting proprietary SDR stuff: http://www.ruckuswireless.com/technology/beamflex

What do we know about (specific) hardware adoption in communities of interest to prioritize target platforms?

commotion's strategy is to use testbeds in US (detroit/phila/??) as models of different usages/adoption patterns – these models can inform deployments globally.

What about specific devices (mobile device model/manufacturer)? Or is the plan to include delivering hardware more than repurposing hardware already on the ground?

  • 'What's the point of putting a $100 piece of hardware on a roof when you know it will not be used in x years … ' ? (re: repurposing / found)
  • Focus on building from the ground up, open source hardware with “luck”

Is Commotion trying to deal with aggressive filtering (e.g. China) by governments? How to avoid governments using technology adoption as a way to identify targets for repression/spying?

  • In some countries, the use of encryption is illegal, have to work in on-the-ground context.
  • Access is taking this into account by supporting a variety of projects.
  • It helps to make a technology a “commonly used thing for legitimate purposes”
  • Commotion aims to integrate existing devices which may make it harder to identify users of mesh networks
  • Concern that infrastructure vs. ad-hoc mode can be used to identify users of mesh
  • Countries have spectrum regulation groups that scan for network use.
  • Some countries (e.g. Russia) prohibit running long-distance wifi w/o license.
  • But: in flashpoint regions there can be enough momentum that people don;t care so much about being identified and become more concerned about being able to maintain communications and be effective. “I am Sparticus”
  • Important to communicate the risk to people on the ground.
  • Need to consider tech. needs of activists vs. “everybody”
    • one strategy can be to put tech. in the same marketplace as other technologies (strategy taken by Guardian project)
    • another strategy might be to promote mesh as social network
  • E.g. skype is unpopular with some governments because it has such widespread adoption that it becomes hard to surveil.
  • Tor is a good example of using documentation to convey capabilities and risks to users. This is an important goal for Commotion. Don't want people to think that they're secure and end up in the gulag.
  • Need to foster applications on top of Commotion that allows for plausible deniability
  • Voip
  • Media streaming/sharing
  • Social chat
  • Sensor networks
  • Community safety

Mesh Administrativia

  • What provisions for mesh dashboard, i.e. map/list of active nodes, latest checkin status?
  • Dashboard is ideal for reporting operational status and throughput measurements, since dashboard data is not realtime, doesn't burden limited CPU of nodes.
  • What about CoovaAAA-style captive portal that is editable by all (or select) mesh users? Such a portal would be the first thing seen by everyone connecting to mesh. LUCI web configuration interface has a rudimentary splash and also has a new web interface for advertising OLSR services, combining those two (and ideally scraping services from mDNS) would be a great way to show services on the mesh.
  • How do mobile-based micropayment systems interface with the GSM infrastructure? How does $ fluidity over networks play a part in civil disobedience objectives.*

Challenges

How to push documentation to upstream partners as well as code + offer resources for documentation of partner technologies. (Does launchpad.net provide a good platform for this?) ← I see the challenge as more social than platform.

How to create perception and actual buyin that technologies are for “everyone”

JB: How to differentiate the mesh network over other forms of internet

  • how do you gain the community investment
  • business model / how to tie into local cash economy (re: West tactic)

How to accelerate widespread hardware access/adoption

Training/support: How to create a distribituted, collaborative, community support infrastructure to fix problems with mesh/personal area network technologies when they stop working on end user devices… for that matter how to train admins/deployers.

Network Development: making sure the tech matches the needs of the communities in which it will be deployed

  • how can these strategies be used across diverse locations (detroit / phil / abroad), similar / different
  • deployment that reflects the inherent organizing patterns and relationships within a location / community
  • theory / ground level applications, what is the priority

Making nuts and bolts hardware cheap (e.g. masts for deploying GSM hardware)

Peer-to-peer networking in GSM band

  • GSM uses multiple frequencies
  • Phones would need 2 transmitters/receivers. Increases cost of device, power consumption

Naming and different network namng authorities (note if a network is cut off there is a good chance the DNS will be gone too, plus some of the necessary naming may not be ICANN/DNS possible) http://tools.ietf.org/html/draft-irtf-dtnrg-dtn-uri-scheme-00 http://tools.ietf.org/html/draft-davies-dtnrg-uri-find

If your network is cut off from the rest of the world it would be good to have distributed and addressable information storage in the nodes that are still connected.. the addressing stuff above is partly about this - people are calling some aspects of this Information Centric Networking (ICN). In the 'crisis' situation this raises a whole raft of novel security issues. Like it would be good for the information to be valueless if the machine is 'captured' - the equivalent of spraying the banknotes with purple dye when somebody breaches the ATM box. OASIS did some work on this. There is an extension which is really fun.. making bandwidth valueless if somebody breaks in. But then you get back to the problem of knowing when there has been a breakin.

Routing issues. In a DTN network, much of the information that mainstream routing needs can be difficult. DTN research is looking into methods. We (N4C and Bytewalla) worked on developing and implementing a protocol to deal with probablistic movenment in a loose unpredictable mesh. http://tools.ietf.org/html/draft-irtf-dtnrg-prophet-09 “Probabilistic Routing Protocol for Intermittently Connected Networks” Support can be found in both the Android code and the DTN2 code. though at the moment they don't support the same version and there is some more work to be done to get to compatability between the implementations. Espcially since the latest update of the protocol (the one we hope to take to RFC real soon now - for some value of soon) upreved the protocol because of some backwardly incompatitble changes we made and becasue the DTN bundle requirements had changed since the time we wrote the first version of the protocol.

Opportunities

Youth adoption of mesh as social network E.g.:

  • messaging via Serval
  • streaming Pandora over Orbot/Tor from Syria
  • WasabiNET is using Chumby to create music service on the mesh. “Innovative ways to make mesh wifi relevent”
  • Stream local LPFM radio stations over mesh. Barnraise LPFM station and deploy mesh as a package deal.

Support tech. for non-political network blackout situations

Ben West discusses WasabiNet mesh network in St. Louis. http://gowasabi.net 20foot / 6meter Flagpole for mounting mesh hardware: $60. [Vendor link: http://www.harborfreight.com/20-ft-telescoping-flag-pole-kit-95598.html Install photo: https://picasaweb.google.com/wasabinetsupport/WasabiNetOnYourRooftop#5603763990251235426 ] Use Ubiquiti and Ingenious hardware. Shows ROBIN / Cloudtrax dashboard tool. http://cloudtrax.com 50 access points sharing 3 broadband connections.

Uplink management hardware - QoS, shaping for bandwidth availability

  • Need CPU power to manage uplink, traffic filtering, tiered bandiwdth. Commodity linksys hardware doesn't have enough power.
  • Ubiquiti devices trending toward 400MHz CPU /w 32MB+ RAM, making this limitation less painful

WasabiNet uses 2-layer mesh because of different range capabilities (2.4GHz for short-range, 5GHz devices for 2000+ feet)

  1. 2.4GHz devices run ROBIN firmware, since proven to work
  2. 5/8GHz devices run custom-patched OpenWRT backfire / OLSR. Looking for stable distro/platform to migrate to.

West: “Recipe rather than algorithm.” Not a disciplined approach.

west: wasabinet community-supported network model uses subscriptions paid through local merchants with barcode keychains as well as grant funding to distribute subscriptions and computers to people

west: wasabinet uses out-of-the box dashboards/portals hosted on Internet servers. U.S. networks don't have lack of 'net connectivity as an immediate concern, but having the option for a locally-hosted or dis distributed administrative system is a goal

Captive Portal: user controlled content for collaborative content sharing, hyper local

  1. edited right on the spot, incentive to community, stimulant to new and exisiting local economic infrastructure
  2. sale of ads on cap captive portal within the comunity to ensure a sustainable network