Universitatea Politehnica Bucuresti

Founded in 1818, UPB is Romania’s largest engineering school, with 15 faculty departments and 25000 students (of which Electronics-Telecommunications-IT ca. 3000 students). UPB collaborates with over 150 universities, from 27 countries.

UPB ranks 212thin the world according to its publications’ impact, being is dedicated to excellence training – in particular in academia-industry partnerships.  The Electronics-Telecommunications-IT department focuses on the tall-thin engineer paradigm, combining hardware design with modern CAD/CAE software tools. The department is active in student certifications in relation with industry (such as TIE / TIE+, having over 20 years of experience).

Logo of this consortium partner

Universitatea Politehnica Bucuresti
Splaiul Independentei nr. 313, sector 6
Bucharest, 060042
Romania
www.upb.ro

EDU-4.0 for Industry-4.0

Role

UPB will serve as Task-1.1.2 leader, fielding an AES based encryption solution to hopping port-number communications. Additional effort will be dedicated to securing the solution against side-channel attacks. The solutions will be tested with encryption specific software. For Task-1.2.1 – UPB will upgrade its TIE/TIE+ industry-driven student certification challenges to Virtual Prototyping, delivering a solution to Industry-4.0-specific personnel certifications. In Task-1.3.1 UPB will develop a template for a neural network applications. Show-case will be the (SIGINT) problem of autonomous RF-modulation type recognition. Tests for purity and speed will be performed.

Key Contribution

Task-1.1.2 UPB’s solution will make computer communications (via AES/hopping-ports) similar to military grade hopping-frequency communications. In Task-1.2.1 UPB’s contribution to Virtual Prototyping and student certification will provide a symmetric EDU-4.0 training programme for Industry-4.0 engineer training. For Task-1.3.1 UPB’s effort will field a full-suite template-solution, with: data conditioning, pre-definition of higher-level parameters with improved discriminating power, design of master-slave neural networks (neuro-slaves providing “binary” flags and master neural network assembling said flags), neural network training particularities (pruning, division of training data into witness and training, BFGS vs stochastic training advantages/disadvantages, etc).

Share this page on Social Media