|
|
AI-based classification of RF signals
Turák, Samuel ; Ulovec, Karel (referee) ; Polák, Ladislav (advisor)
Táto práca sa zameriava na klasifikáciu rádiofrekvenčných (RF) signálov založenú na hlbokom učení. Pre tento účeľ, tri neuronové siete sú vybrané a prezentované: Konvolučná Neurónová Sieť (CNN), Sieť s Bránovými Rekurentnými Jednotkami (GRU), Konvolučná Hlboká Neurónová Sieť s Bránami (CGDNN). Všetky sú trénované a vyhodnotené na viacerých datasetoch, ovplyvnené rôznymi RF rušeniami, pre klasifikáciu rôznych bezdrátových štandardov. Signály v jednotlivých datasetoch boli vytvorené pomocou aplikácie Wireless Waveform Generator v programu MATLAB. Jeden verejne dostupný dataset na klasifikáciu modulácie je takisto testovaný na modeloch. Použité prístupy k predspracovaniu dát, tréningu modelov a vyhodnoteniu modelov sú implementované v programovacom prostredí Python s využitím knižníc ako Scikit-learn a Keras. \mbox{Získané výsledky} sú prehľadne prezentované a diskutované.
|
|
|
Advanced tool for generating modern Slow DoS attacks
Hrůza, Tomáš ; Člupek, Vlastimil (referee) ; Sikora, Marek (advisor)
In today’s world, cyber threats are becoming increasingly sophisticated. Those threats include SDoS (Slow Denial of Service) and SDDoS (Slow Distributed Denial of Service) attacks, which employ advanced methods to disrupt normal service operations. These attacks are particularly difficult to detect and are effective because they simulate the behavior of legitimate users with slow internet connections. The topic of SDoS attacks is relatively new and not thoroughly documented. To respond to potentially newly devel- oped attacks, it is necessary to understand the principles of currently known attacks and have the practical capability to create them in order to develop effective countermeasures in the future. This thesis focuses on the development of an advanced tool for generating modern SDoS attacks. The main contribution of this work is the enhancement of the generator to create distributed attacks, the creation of an intuitive interface, and more options for monitoring the progress of individual attacks. The theoretical part introduces the topic of internet connection establishment and explains the properties of TCP and IP protocols in detail. It then clarifies the theory of secure communication over the internet using the HTTPS protocol and provides a comparison of currently used web servers. The final theoretical section addresses the topic of denial of service, discussing some types of currently known SDoS attacks, the tools that generate these attacks, and their shortcomings. Next chapter details the implementation of functionalities, which includes performance enhancements of the tool through the use of multiple processes. The developed tool features Slow Read, Slow Next, and Slow Drop attacks, as well as the ability to combine these attacks. This is followed by a description of how a local network of virtual machines was created for the purpose of testing the implemented tool. The final chapter presents the results and effectiveness of the tool in conducting SDDoS attacks against Apache and NGINX web servers in a local network.
|
|
|
Time Series Forecasting Using Machine Learning
Elrefaei, Islam ; Galáž, Zoltán (referee) ; Hošek, Jiří (advisor)
The aim of this thesis is to explore the application of various artificial intelligence (AI) techniques for the prediction of time series data, which is prevalent in fields such as finance, economics, and engineering. Accurate time series prediction is essential for effective decision-making and planning. This thesis reviews several traditional and state-of-the-art AI techniques used for time series prediction, including linear regression, ARIMA, support vector regression, random forests, and deep learning. These techniques are applied to different time series datasets, encompassing both univariate and multivariate data. The performance of the predictive models is evaluated using various scalar metrics. The performance of the models was different depending on the type of the dataset. Additionally, this thesis includes the development of a user interface application that allows users to change parameters and forecast new results based on their entries. Furthermore, the thesis discusses the challenges and limitations of using AI techniques for time series prediction and provides suggestions for future research directions.
|
|
|
Development and Deployment of a Platform for Efficient Data Collection and Visualization in Opportunistic Sensor Networks
Chovancová, Emília ; Miklánek, Štěpán (referee) ; Musil, Petr (advisor)
Oportunistické senzory predstavujú zaujímavú alternatívu ku získavaniu cenných informácií a dát, ktoré šetria zdroje ekologicky a ekonomicky. Do tejto kategórie patria aj dané meteorologické meradlá a snímače, ktoré sú opísené v rámci tejto diplomovej práce a ktoré budú v budúcnu využívané pre porovnávanie s inými mikrovlnnými spojmi. Cieľom práce bolo vytvoriť technickú dokumentáciu ku príslušným meradlám, v nej opísať ich technické vlastnosti, tok dát a navrhnúť bezepečnostné vylepšenia infraštruktúry. Ďalším cieľom bolo vytvorenie webovej aplikácie, ktorá má umožňovať optimalizovanú vizualizáciu dát v reálnom čase a tým informovať užívateľov o aktuálnych klimatických podmienkach v rámci FEKT kampusu.
|
|
|
Scripts for automated editing of fonts in PDF files
Gmitter, Jakub ; Zeman, Kryštof (referee) ; Hanák, Pavel (advisor)
Master's thesis deals with the issue of font encoding in PDF documents. Proper font encoding is necessary for searching and copying text from such documents. Thesis includes a description of the internal structure of PDF documents, page representation, font types and their encoding, and important objects needed for proper font representation. Understanding of these areas was necessary for development of scripts that are able to repair incorrect font encoding. Two Python scripts were developed as part of the thesis. The first one verifies the integrity of repaired PDF files using SHA-256 hashes computed from their contents. The second script repairs corrupted font encodings in the documents. The necessary information for the functionality of both scripts has been stored in the corresponding JSON structures. The repair script targets PostScipt fonts of type 1. Core function of the repair script is the generation of a ToUnicode object that correctly maps glyphs to Unicode codes within the font. The script has been tested on approximately 200 electronic issues of a Czech magazine that have been provided as sample data. From these sample files, only those that had completely corrupted font encodings were chosen for further work. Other sample magazines only had corrupt encoding of characters with diacritical marks. These magazines were analyzed, but the script is unable to repair them. Commented Python source code, compiled Windows executables and a user guide are available in the electronic attachment and in the author's GitHub repository.
|
|
|
Software-defined antenna array
Sedláček, David ; Prokopec, Jan (referee) ; Maršálek, Roman (advisor)
Hlavným cieľom tejto práce bolo vyvinúť demonštračnú platformu pre implementáciu algoritmov na odhad smeru príchodu signálu. Celkovo bolo opísaných šesť techník, ktoré boli následne implementované v Pythone spolu so simuláciami príchodu signálov. Bola vykonaná štúdia dostupného hardvéru, v ktorej boli porovnané rozdiely medzi na mieru prispôsobenými a voľne dostupnými riešeniami. Navrhli sa konkrétne hardvérové a softvérové konfigurácie pre praktickú implementáciu metód odhadu príchodu signálu. Na záver boli navrhnuté experimentálne merania na overenie účinnosti implementovaných metód. Vykonali sa testovacie merania a ich výsledky boli kriticky zhodnotené.
|
|
|
Optimization of the positioning system of semiconductor chip testing
Kotian, Tomáš ; Jankovský, Jaroslav (referee) ; Otáhal, Alexandr (advisor)
The goal of this diploma thesis was to gather theoretical information on the possibilities of chip testing and subsequently design and apply necessary changes for the semi automatic positioning system, which would ensure its ability to test chips with the use of probe cards. Semi automatic positioning system is a result of a bachelor thesis, which ensured the ability of the machine to provide testing with the use of a resistance system created on a thick layer. The practical part of this thesis is focused on providing an accurate testing of the machine’s parametres and designing and applying necessary changes for the mechanical part of the machine. Furthermore, the thesis also describes the design and creation of an application with automatic detection of a chip, which is used for controlling the machine. The last part of the thesis describes the testing of the machine, which was realized with the use of two different sized chips.
|
|
|
Generative Neural Network for Creating Synthetic Photorealistic Images
Hora, Adam ; Přinosil, Jiří (referee) ; Říha, Kamil (advisor)
The main objective of this work is to select and design a neural network model that will be able to generate realistic images thematically fitting the selected dataset. The architecture used for the solution is Deep convolutional generative adversarial network. This network is than implemented in the Python programming language using the Tensorflow application programming interface and its included interface Keras. Finally, the model is trained on the selected dataset and the resulting generated images are presented. The final model and individual images are then evaluated using various quality assessment methods.
|
|
|
Tools for Wi-Fi and IPv4 penetration testing
Jančík, David ; Lieskovan, Tomáš (referee) ; Martinásek, Zdeněk (advisor)
The master thesis deals with the design and implementation of support tools and methodologies for security penetration testing of Wi-Fi networks and IPv4 network infrastructure. The theoretical part covers penetration testing itself, approaches, phases, and types. It also describes the development of Wi-Fi networks and their security protocols. Various penetration tools for Wi-Fi networks and types of attacks are introduced. In the last theoretical part, a basic overview of IPv4 and tools for IPv4 scanning is provided. Initially, in the practical part, a proprietary methodology for Wi-Fi networks and IPv4 and tools for penetration testing are proposed. The Python programming language is defined, along with the output of various tools for the Penterep platform. A review of tools from the theoretical part is conducted to select suitable tools for new support tools. The implementation of penetration tools is based on the design diagram created. The conclusion summarizes the results achieved and suggestions for further expansion of tools for Wi-Fi and IPv4. The result of this thesis is the implementation of support tools and the design diagram for Wi-Fi networks and IPv4.
|
|
|
Data Mining Based Web Analyzer of Job Advertisements
Wittner, Alex ; Dzurenda, Petr (referee) ; Sikora, Marek (advisor)
Cílem této bakalářské práce bylo vytvoření automatizovaného zadávání nových pracovních inzerátů pomocí vložení URL v rámci již existující webové aplikace https://rewire.informacni-bezpecnost.cz, jejíž cílem je shromažďování pracovních inzerátů v oblasti cybersecurity s podrobnou analýzou pracovních kompetencí. Pracovní inzeráty jsou analyzovány pomocí více vzorového vyhledávacího algoritmu Aho-Corasick, psaného v jazyce Java. K získávání informací ze zadaných pracovních inzerátů slouží Python skript využívající knihovnu Selenium. Výsledná implementace a webová stránka je vytvořena pomocí jazyka PHP a knihovny ReactJS využívající JavaScript.
|
guest ::
