Invited Speaker at ICTAC 2023 in Lima, Peru

I was delighted to be invited to lecture at ICTAC 2023, the 20th International Colloquium on Theoretical Aspects of Computing which took place in Lima, Peru between the 4th and 8th of December, 2023.

I not only had the chance to lecture there but also to build and strengthen bonds with great people with whom, in bigger events, I miss the chance to have long and deep conversations.

In addition, I was not only warmly welcomed to Lima as an Argentinian, but also discovered a beautiful city that reveals how formidable metropolis must have been on the times of the Spanish colonies. Unfortunately, I only had a glimpse of the majesty of pre-colonial civilizations. (A note to myself: a second visit to Peru beyond the borders of Lima is imperatively needed)

The data of the lecture is as follows:

Title: Optimal Route Synthesis in Space DTN using Markov Decision Processes.
Abstract: Delay-tolerant networks (DTN) are time-evolving networks that do not provide continuous and instantaneous end-to-end communication. Instead, the topological configuration of DTN changes continuously: connections are available only during some time intervals and thus the network may suffer from frequent partitions and high delay. In this sense, the DTN paradigm is fundamental to understand deep-space and near-Earth communications. A particular characteristic of space networks is that, due to the orbital and periodic behavior of the different agents (e.g. satellites and terrestrial or lunar stations), contact times and durations between nodes can be accurately predicted. This type of DTNs is called scheduled and expected contacts can be imprinted in a contact plan that exhaustively describes the future network connectivity. In addition, the contacts may suffer of some quantifiable failure that can be included in the contact plan. Thus, this behavior can be encoded in a Markov decision process (MDP) where the non-determinism corresponds precisely to the routing decisions. With this model at hand, we have developed and studied several offline techniques for deriving optimal and near-optimal routing solutions that ensure maximum likelihood of end-to-end message delivery. In particular, we have devised an analytical solution that exhaustively explores the MDP very much like probabilistic model checking does, and have also explored simulation-based techniques using lightweight scheduler sampling (LSS). The objective of this presentation is to report this research as well as current ongoing developments for multi-objective routing optimization on space DTN.

The research presented here have been the result of the cooperation with Juan Fraire, Arnd Hartmanns, Fernando Raverta, Ramiro Demasi, Jorge Finochietto, and Pablo Madoery.