'Quality of Service Compliance Verification Using Federated Learning Empowered by Blockchain' (26/03/2025)

João de Brito Gonçalves

• 26 March 2025
• 7 p.m.
• ​Online​
• Promotors: prof. dr. Johann Marquez-Barja & prof. dr. Rodolfo da Silva Villaça  
• Faculty of Applied Engineering

Abstract

Edge computing brings a new paradigm in which computing, storage, and bandwidth resources are shared as closely as possible to mobile devices or sensors, generating a large amount of data, but the sensitive nature of the data means that there are risks and responsibilities to storing them in a centralized location. Moreover, to address the data privacy required for some data in these devices, we propose the use of Federated Learning (FL) so that specific data on services performed by clients do not leave the source machines. FL is a new sub-field of Machine Learning (ML) that allows training models without collecting the data itself. Instead of sharing data, users collaboratively train a model by only sending weight updates to a server. However, the naive use of FL in the scenarios aforementioned exposes it to a risk of corruption, whether intentional or not, during the training phase. This is due to the lack of training monitoring and the difficulty in verifying the quality of training datasets. To improve the security of FL systems, we propose a Blockchain-based framework in an edge computing scenario. Blockchain, with its immutability and traceability, can be an effective tool to prevent malicious attacks in FL. More specifically, the immediate updates made by each participant to its local model can be chained together on the distributed ledger offered by a blockchain such that those model updates are audited, and malicious trainers can be removed of the system. We also apply blockchain to create a reward mechanism in FL to enable an incentive strategy for trainers.

'Impact of rational catalyst design on stability for electrochemical ammonia synthesis revealed by electron microscopy' (21/01/2025)

Saskia Hoekx

• 21 January 2025
• 4.30 p.m.
• ​Campus Middelheim - Building A, room m.A.143
• Promotor: prof. dr. Tom Breugelmans & prof. dr. Sara Bals ​
• Faculty of Applied Engineering

Abstract

Ammonia is a high commodity chemical that is predominantly used as a fertilizer in the agricultural industry. Over 180 million metric tons are produced annually, and the demand is expected to increase by 3-5% each year. Currently, this is synthesized using the energy-intensive and polluting Haber-Bosch process, which produces over 1% of the world’s total yearly greenhouse gas emissions, and 15% of the total carbon dioxide emitted by the chemical industry. One promising alternative is the electrochemical nitrate reduction reaction, which can convert polluting nitrates in agricultural wastewater back into useful ammonia under ambient conditions, with no carbon footprint. However, in order to be industrially applicable, this reaction requires an efficient and stable catalyst. In this work, characterization by electron microscopy and electrochemical tests are combined to approach catalyst optimization rationally. This resulted in a more stable catalyst that can effectively catalyze the nitrate reduction reaction for at least 24 hours without a significant loss in electrochemical performance.

'Development of a reversibly immobilised cell reactor for the valorisation of lignocellulosic wastewaters to microbial oil' (14/01/2025)

Waut Broos

• 14 January 2025
• 4 p.m.
• ​Campus Drie Eiken- Building O, room d.O.01
• Promotor: prof. dr. prof. dr. Iris Cornet, prof. dr. Siegfried E. Vlaeminck & prof. dr. Jan Dries ​
• Faculty of Applied Engineering

Abstract

Lignocellulose-based biorefineries produce a significant amount of wastewater containing several phenolic compounds. Today, these wastewaters are mostly considered a burden. However, some microorganisms, e.g., Rhodotorula kratochvilovae and Cutaneotrichosporon oleaginosum can convert phenolics into valuable intracellular components, namely single-cell oil (SCO). This turns the waste stream into a raw material and an economic opportunity. However, valorising wastewater to SCO is challenging due to its low substrate concentration. Therefore, supplementation with a high-value carbon source is used in literature to obtain appreciable SCO titters. To date, the production of SCO from high-value carbon sources has proven to be economically unfeasible. As a result, this study explores the use of a repeated batch fermentation strategy to overcome the low substrate concentration. In a repeated batch, the cells are recycled for use in the next batch. Cell recycling and repeated feeding allow intracellular SCO to accumulate in the cells. This allows high SCO concentrations to be achieved from substrate-poor wastewaters. Three technologies can be used for cell recycling: centrifugation, membrane technology and immobilisation. However, each of these technologies has its drawbacks: centrifugation is energy-consuming and requires a significant investment, membrane technology is prone to fouling, and conventional immobilisation technologies make recovery of the intracellular product problematic.
Our hypothesis is that the design of a new reactor type, namely a reversible immobilised cell reactor (RICR), offers a possible solution. In this reactor, the cells are first immobilised on a suitable support, repeated batch fermentation occurs, and finally remobilised to recover the intracellular components. As a case study, the wastewater obtained from thermochemical pre-treatment of lignocellulose is investigated as a substrate for microbial oil production. This study aimed to design an economically feasible process for valorising this lignocellulosic wastewater.