Sabin BELU
Data și ora: 2022-12-05 14:00
Locația: ETTI, Sala consiliu și Microsoft Teams
Rezumat teză de doctorat: Accesează
The thesis focuses on lossless data compression techniques and algorithms with applications in the automotive and IoT industries. We discuss in particular, variations of algorithms based on dictionary compression algorithm, LZ77 and on the entropic coding side, Huffman coding and a new variant called canonical Huffman and Arithmetic encoding. It begins with the presentation of the classic Huffman algorithm, followed by the canonical rules that make up the formation of canonical Huffman prefix codes, hence the name of the variant, canonical. Among the original contributions on canonical Huffman entropy coding we present a new decoding method with the ability to process a maximum of 12 symbols in a single decoding cycle. We also propose a new architecture for an arithmetic coder called quasi-static. Unlike classical implementations, the quasi-static encoder divides the input stream into buffers of certain sizes and uses a quasi-static model for encoding the input data. The advantage of such an approach is the high coding speed obtained, but at the price of a slightly lower compression rate. We further present the differentiation of binary data, especially referential compression (RC), we show how to build a referential dictionary and possible attack scenarios from hackers, and we also propose an original algorithm based on NCD. We continue with the presentation of binary data differentiation algorithms and, in particular, modified dictionary-based compression algorithms, with the aim of using these algorithms for the software update case. We present software update scenarios, followed by the description of the server-side software that creates the delta binary and also the client software that performs the software update installation. The proprietary algorithms were designed to be suitable for a resource-constrained environment and could very well be used in industrial areas such as the automotive industry. They are supported by software implemented and tested on real files from the domains mentioned above, and the articles that support them have been presented and supported in conferences and accepted for publication in specialized magazines and journals.

Conducător de doctorat

Prof. dr. ing. Daniela COLȚUC, Universitatea Politehnica din București, România.

Comisie de doctorat

Prof. dr. ing. Gheorghe BREZEANU, Universitatea Politehnica din București, România
Prof. dr. ing. Ioan TABUȘ, Tampere University, Finlanda
Prof. dr. ing. Daniela TĂRNICERIU, Universitatea Tehnică “Gheorghe Asachi” din Iași, România
Conf. dr. ing. Daniela FAUR, Universitatea Politehnica din București, România.

Comisie de îndrumare

Prof. dr. ing. Inge GAVĂT, Universitatea Politehnica din București, România
Conf. dr. ing. Daniela FAUR, Universitatea Politehnica din București, România
Șl.dr. ing. Lucian PERIȘOARĂ, Universitatea Politehnica din București, România.