|
|
Movement Prediction of Wireless Nodes in Mobile Ad Hoc Networks (MANETS)
Makhlouf, Nermin ; Šimák, Boris (oponent) ; Slavíček, Karel (oponent) ; Koton, Jaroslav (vedoucí práce)
The rapid evolution in the field of mobile computing has led to a new alternative way for mobile communication, in which mobile nodes form a self-organising wireless network, called a Mobile Ad hoc Network (MANET). The specific characteristics of MANETs impose many challenges to network protocol designs on all layers of the protocol stack because of unpredictable topology changes and mobile nature. Mobility prediction is a tool to deal with the problems emerging from the nodes’ mobility by predicting future changes in the network topology. This is crucial for different tasks such as routing. In this doctoral thesis, two mobility prediction methods for MANET networks are developed. The first method supposes that each node can build its virtual map depending on its location over the time. This method is called mobility prediction using virtual map. In order to evaluate the developed prediction algorithm, it has been implemented in the network simulator NS-2. I have investigated existing mobility models, and how the prediction method can be applied to them. Simulations respectively realize performance improvement in terms of average end to end delay, packet delivery ratio and network throughput under different mobility model. The proposed prediction concept is implemented over AODV (Ad Hoc On-Demand Distance Vector) routing protocol. In the second method, I have developed an artificial neural network for movement prediction in MANETs. The prediction model for mobility has been done by the data collected from location patterns. The Bayesian technique was used for learning or training ANNs. It has been implemented in software for training Bayesian neural networks called Model Manager. The best way to evaluate the final model is done by making predictions and comparing predictions with target data. The predictions are made by using 50 patterns as input variables. The reached and in the thesis discussed results show that improvement in the most significant network parameters, i.e. delay, throughput and packet delivery ratio, are reached even by 30% compared to AODV routing protocol, where the proposed prediction model is not utilized.
|
|
|
Problematika optimální šířky přenosového pásma pro přenos medicinských obrazových dat
Schindler, Vladimír ; Šimák, Boris (oponent) ; Šárek, Milan (oponent) ; Dostál, Otto (vedoucí práce)
Disertační práce je zaměřena na optimalizaci parametrů přenosového pásma pro transport medicínských obrazových dat mezi zdravotnickými zařízeními a vzdálenými datovými úložišti. Jako reálná a plně funkční struktura, jež bude v této práci analyzována, byl zvolen systém MeDiMed (Metropolitan Digital Imaging System in Medicine). Práce nejprve rozebírá provoz menších zdravotnických organizací a jejich modalit, které využívají tento systém pro vzdálenou archivaci dat. Analýza provozu je poté statisticky zpracována. Disertační práce se dále zabývá analýzou zvýšení zabezpečení přístupu pracovních stanic do zdravotnického systému a posuzuje jeho vliv na přenášená data. Je zde porovnáván vliv nastavení přenosových parametrů a nejpoužívanějších typů šifer na rychlost přenosu.
|
|
|
Audio Classification with Deep Learning on Limited Data Sets
Harár, Pavol ; Platoš,, Jan (oponent) ; Šimák, Boris (oponent) ; Mekyska, Jiří (vedoucí práce)
Standard procedures of dysphonia diagnosis by a clinical speech therapist have their downsides, mainly because the process is very subjective. Recently, an automatic objective analysis of a speaker's condition gained in popularity. Researchers successfully based their methods on various machine learning algorithms and handcrafted features. These methods, unfortunately, are not directly scalable to other voice disorders and the process of feature engineering is laborious and thus financially and talent expensive. Based on the previous successes, a deep learning approach might help to ease the problems with scalability and generalization, but an obstacle is a limited amount of training data. This is a common denominator in almost all systems for automated medical data analysis. The main aim of this work is to research new approaches to deep-learning-based predictive modeling using limited audio data sets, focusing especially on voice pathology assessment. This work is the first to experiment with deep learning in this field and on so far the largest combined database of dysphonic voices, which was created in this work. It provides a thorough examination of publicly available data sources and identifies their limitations. It describes the design of novel time-frequency representations based on Gabor transform and it presents a new class of loss functions, that yield target representations beneficial for learning. In numerical experiments, it demonstrates improvements in the performance of convolutional neural networks trained on limited audio data sets using the augmented target loss function and the newly proposed time-frequency representations, namely Gabor and Mel scattering.
|
|
|
Advanced Parameterisation of Online Handwriting in Writers with Graphomotor Disabilities
Mucha, Ján ; Šimák, Boris (oponent) ; Drotár,, Peter (oponent) ; Mekyska, Jiří (vedoucí práce)
Graphomotor disabilities (GD) significantly affect the quality of life beginning from the school-age, when the graphomotor skills are developed, until the elderly age. The timely diagnosis of these difficulties and therapeutic interventions are of great importance. As GD are associated with several symptoms in the field of kinematics, the basic kinematic features such as velocity, acceleration, and jerk were proved to effectively quantify these symptoms. Nevertheless, an objective computerized decision support system for the identification and assessment of GD is still missing. Therefore, the main objective of my dissertation is the research of an advanced online handwriting parametrization utilized in the field of GD analysis, with a special focus on methods based on fractional calculus. This work is the first to experiment with fractional-order derivatives (FD) in the GD analysis by online handwriting of Parkinson’s disease (PD) patients and school-age children. A new online handwriting parametrization technique based on the Grünwald-Letnikov approach of FD has been proposed and evaluated. In the field of PD dysgraphia, a significant improvement in the discrimination power and descriptive abilities was proven. Similarly, the proposed methodology improved current state-of-the-art techniques of GD analysis in school-aged children. The newly designed parametrization has been optimized in the scope of the computational performance (up to 80 %) as well as in FD order fine-tuning. Finally, various FD-approaches were compared, namely Riemann-Liouville, Caputo’s, together with Grünwald-Letnikov approximation to identify the most suitable approach for particular areas of GD analysis.
|
|
|
Controllable Fractional-Order Analogue Electronic Circuits
Dvořák, Jan ; Vávra, Jiří (oponent) ; Šimák, Boris (oponent) ; Jeřábek, Jan (vedoucí práce)
The doctoral thesis focuses on the synthesis and analysis of novel non-integer-order (fractional-order) circuit structures with electronically adjustable parameters. The main goal of the thesis is the design of new solutions of fractional-order current-mode filtering structures, fractional-order passive elements and also oscillators. The thesis contains the designs of three emulators of fractional-order elements, three filtering structures and two oscillators based on the usage of a fractional-order passive element in their circuit structure, and two general conceptions of fractional-order filters based on an approximation of the fractional-order transfer function. Based on general conceptions of the filtering structures, the fractional-order low-pass and high-pass filters are designed. The adjustability of the order, the pole frequency and in several cases also the quality factor of the proposed circuits is provided by used active elements with adjustable parameters. The features of the proposed circuits are verified by simulations using behavioural simulation models of the active elements. Several of these circuits were implemented on PCB and verified by laboratory measurement.
|
|
|
Pokročilé možnosti zabezpečení medicínských obrazových dat
Roček, Aleš ; Šimák, Boris (oponent) ; Molnár, Karol (oponent) ; Dostál, Otto (vedoucí práce)
Zdravotnická zařízení začala využívat výhody digitálního ukládání medicínských obrazových dat jako jsou dostupnost, jednoduché sdílení, vysoké rozlišení atd. Digitální podoba zdravotních záznamu však přináší kromě výše zmíněných výhod i nevýhodu v oblasti nutnosti zajistit bezpečnost těchto dat. Ta jsou jednodušeji napadnutelná, zcizitelná a použitelná bez autorizace. Tato práce se zabývá bezpečností medicínských obrazových dat, popisuje potřeby a přístupy k zabezpečení, vysvětluje důvody nasazení zabezpečení vodoznačením. Vyjmenovává hlavní typy vodoznačení a porovnává jejich klady a zápory. V práci je navržena nová metoda kombinující výhody a potlačující nevýhody tří základních principů vodoznačení v oblasti medicíny: nulového, vratného vodoznačení a vodoznačení v oblasti, která nenese důležitou medicínskou informaci (Region Of Non Interest, RONI). Pro praktické testy vlastností navržené metody byla použita rozsáhlá databáze medicínských snímků. Tyto testy přinesly velice slibné výsledky. Jejich rozbor a porovnání s ostatními metodami vodoznačení medicínských obrazových dat jsou uvedeny v závěru této práce.
|
|
| |
|
| |
|
|
Laboratoř pro výuku a testování síťových prvků
Poláček, Marcel ; Šimák, Boris (vedoucí práce) ; Holý, Radek (oponent)
Název: Laboratoř pro výuku a testování síťových prvků Autor: Marcel Poláček Katedra: Katedra informačních technologií a technické výchovy Vedoucí práce: prof. Ing. Boris Šimák, CSc. Abstrakt: Práce se zaměřuje na problematiku týkající se telekomunikačních a počítačových sítí. Vypisuje základní témata tohoto oboru. Cílem práce je najít řešení, vytvoření laboratoře pro testování síťových prvků, které lze libovolně nastavovat, přičemž se zachová maximální autenticita pro odbornou výuku. Je také nutné najít řešení ekonomicky výhodné, aby bylo možné tuto práci aplikovat v co největší možné míře v institucích vyučující danou problematiku. Součástí práce je i vymyšlení základních laboratorních úloh s různou obtížností. Klíčová slova: počítačové sítě, operační systémy, telekomunikační technologie, přenos dat, laboratoř pro testovaní sítí, virtualizace a emulace síťových zařízení
|
|
|
Advanced Parameterisation of Online Handwriting in Writers with Graphomotor Disabilities
Mucha, Ján ; Šimák, Boris (oponent) ; Drotár,, Peter (oponent) ; Mekyska, Jiří (vedoucí práce)
Graphomotor disabilities (GD) significantly affect the quality of life beginning from the school-age, when the graphomotor skills are developed, until the elderly age. The timely diagnosis of these difficulties and therapeutic interventions are of great importance. As GD are associated with several symptoms in the field of kinematics, the basic kinematic features such as velocity, acceleration, and jerk were proved to effectively quantify these symptoms. Nevertheless, an objective computerized decision support system for the identification and assessment of GD is still missing. Therefore, the main objective of my dissertation is the research of an advanced online handwriting parametrization utilized in the field of GD analysis, with a special focus on methods based on fractional calculus. This work is the first to experiment with fractional-order derivatives (FD) in the GD analysis by online handwriting of Parkinson’s disease (PD) patients and school-age children. A new online handwriting parametrization technique based on the Grünwald-Letnikov approach of FD has been proposed and evaluated. In the field of PD dysgraphia, a significant improvement in the discrimination power and descriptive abilities was proven. Similarly, the proposed methodology improved current state-of-the-art techniques of GD analysis in school-aged children. The newly designed parametrization has been optimized in the scope of the computational performance (up to 80 %) as well as in FD order fine-tuning. Finally, various FD-approaches were compared, namely Riemann-Liouville, Caputo’s, together with Grünwald-Letnikov approximation to identify the most suitable approach for particular areas of GD analysis.
|
host ::
RSS.