Liberouter: Towards autonomous neighborhood networking (2014) | Hacker News Reader

Liberouter: Towards autonomous neighborhood networking (2014)

ieeexplore.ieee.org

52 pointscunidev12 comments

I'm the main author of this paper. It was a part of my PhD work on opportunistic networking. I'm surprised to see it here, but happy to answer any questions.

I still believe this approach has a lot of unrealized potential, but unfortunately I never had the resources (time, students, funding) to push it as far as I think it could go. It has been maintained on a best effort basis over the years.

Hi, from a first glance at the project, how would it compare with yggdrasil, cjdns, etc (pinecone, batman and other mesh networks) ?

Edit: looks like the main difference is caching, allowing for disconnected networks (also sneakernets) to store and forward messages later, a bit like Secure Scuttlebutt handles it. Interesting.

Yes, the difference is that mesh networks and MANETs attempt to build end-to-end packet paths. This approach (opportunistic networking, or Delay-Tolerant Networking) relies purely on infrequent peer-wise contacts to forward messages.

The protocol stack in the Liberouter system is the DTN stack, ie., the Bundle Protocol (RFC 5050, 9171).

That sounds like an extremely interesting topic, thanks for the answer! I wish it was spelled a bit more clearly on the website too.

That would interoperate well with RFC 1149 (IP over avian carriers!).

This is pretty cool, I've imagined similar concepts.

Are you running it anywhere or do you know anyone who is running it?

http://liberouter.mobi/

https://i.imgur.com/VbRSdZg.png ouch

I'm at such a loss for what they were trying to do with that that I started wondering if that was intentional.

They forgot to update the link to the background image. It is missing this background image reference - http://www.liberouter.mobi/img/censorship.jpg

https://github.com/tk721/liberouter-people-finder

If I'm reading Figure 5 ("Fraction of clients reached by each message.") correctly, in a simulated "neighborhood network" with 50 users, broadcast messages reach less than 50% of users?!

Note that there is already another networking project called Liberouter: https://www.liberouter.org/

Reply on news.ycombinator.com