LERIS research group got three papers accepted (two as full and one as short) in the IEEE NOMS 2024 conference. The paper “TimeGAN as a Simulator for Reinforcement Learning Training in Programmable Data Planes.“, lead by the Post-doc Thiago Caproni, shows that GANs are quite useful for enhancing Reinforcement Learning by means of generating synthetic data from the programmable data plane. The paper “Unlocking Security to the Board: An Evaluation of SmartNIC-driven TLS Acceleration with kTLS“, lead by Msc student Felipe A. S. Novais, reveals several takeaways regarding crypto functions offloading in a smartNIC. The paper shows that, in certain setups, offloading of crypto functions to the smartNIC may not bring gains.
Finally, the paper “Detecting Network-Wide Heavy Hitters in Multi-Pipe Programmable Devices”, accepted as short, lead by the Msc student Thiago Henrique Silva Rodrigues, proposes a mechanism for collecting network-wide heavy-hitter flows in multi-pipes switches.
Congrats to all the team!