Conferences & Papers

As is usual these days, I don’t write here half as much as I would like to. In recent weeks, this is because I have been very busy preparing for a number of conferences and writing papers. Since they all relate in one way or another to the protocol work I have been writing about, I figure people might find them interesting.

I have mentioned before that I am working part time at AnyWi Technologies B.V.. The two jobs are fairly complementary; at Interpeer, protocol design is for peer-to-peer networks, while in the latter, it is for autonomous vehicles, in particular for drones. Really, the Venn diagram between the two, if you would draw it, would probably have about 90% overlap or so. So the appearances and publications relate to both.


Figure: “Conference” by dkalo is licensed under CC BY-SA 2.0

My AnyWi colleague Morten Larsen and I co-wrote a paper related to the COMP4DRONES project, in which we outline the beginnings of EU regulations for safely communicating with commercial drones.

Drones in the EU are subdivided into three broad classes; the first is the toy category, the last is the military category. In the middle there is space for commercial “pizza delivery” type drones.

In order to fly them cost-effectively, it’s best to make use of commercially available communication links - such as LTE/5G or WiFi. The downside is that those may not always provide the reliability you’d need to fly in higher risk areas, e.g. over people’s heads.

AnyWi is working on technology effectively bundling these links into a virtual link with higher availability, failover and even bandwidth.

There is a preprint published at researchgate, and the paper will be published in the ACM proceedings of the conference.

Designing a Human Centric Next Generation Internet

This talk, which I have confirmed for FOSDEM and FOSSASIA, but not yet for LibrePlanet, I want to talk about the vision and progress on the Interpeer Project.

On the one hand, it’s an attempt to shake people up a bit and think about how the web stack isn’t really suitable for the future. I wrote in this blog before about safety in communications, and there will be a bit more on that topic in different words.

The other part of the talk is going to cover some of the protocol design I have been writing about here, as well as in the other paper that may or may not (probably will) be published in MICPRO.

Where the first paper written with Morten is covering the general problem of reliability in commercial drones, this second paper is about the multi-link tunnelling protocol we’re working on; this work is done in the context of ADACORSA.

It’s a bit more abstract than the protocol design things I’ve written about here, in that it focuses on messages and state machines, and stays far away from more concrete information such as byte-level packet encoding. That’s deliberate, because on the one hand, we’re still working on it - and on the other, wire-encoding is the simpler part.

Plus, some version of this protocol will make it into the Interpeer stack somehow. After all, handover between ground stations for a drone is the same kind of scenario - at the networking layer - as keeping up a video conference stream on your mobile phone as you leave your home WiFi and fall back to the LTE cell.

Note 2022/12/22: This paper was not published, in the end.

Published on January 2, 2021