Auswahl der wissenschaftlichen Literatur zum Thema „Classification of network devices“

Automatic classification of devices in computer network can be used for detection of anomalies in a network and also it enables application of security policies per device type. The key to creating a device classifier is a quality data set, the public availability of which is low and the creation of a new data set is difficult. The aim of this work is to create a tool, that will enable automated annotation of the data set of network devices and to create a classifier of network devices that uses only basic data from network flows. The result of this work is a modular tool providing automated annotation of network devices using system ADiCT of Cesnet's association, search engines Shodan and Censys, information from PassiveDNS, TOR, WhoIs, geolocation database and information from blacklists. Based on the annotated data set are created several classifiers that classify network devices according to the services they use. The results of the work not only significantly simplify the process of creating new data sets of network devices, but also show a non-invasive approach to the classification of network devices.

Nous vivons actuellement dans une ère avec sans cesse de nouveaux appareils technologiques (smartphone, réseaux de capteurs sans fil, aux caméras haute résolution, etc). En partant des médias sociaux, en passant par des caméras de surveillance très puissantes, et sans oublier la surveillance de la santé en temps réel, on constate qu'une grande quantité de données est stockée dans le cloud et les serveurs. Cela représente un grand défi de stockage et de transmission, en particulier dans les plates-formes aux ressources limitées qui sont caractérisées par : (a) des capacités de calcul limitées, (b) une source d'énergie limitées et (c) des infrastructures ouvertes qui transmettent des données sur des réseaux sans fil peu fiables. Dans cette thèse, nous nous concentrons sur l'amélioration de la sécurité des contenus multimédia transmis sur des plates-formes à capacité de calcul limitée, tout en préservant un niveau de sécurité élevé. Dans la première partie, nous avons étudié les réseaux ad hoc véhiculaire. Nous avons proposé un état de l'art qui permet de résumer la plupart des travaux récents et d'explorer presque tous les aspects de ce domaine en illustrant les différents aspects que possède cette plateforme. Ensuite, afin de proposer une nouvelle solution de sécurité et de valider sa robustesse et le niveau de caractère aléatoire d'une image chiffrée, nous avons proposé un test simple et efficace. Celui-ci est basé sur des outils pour tester statistiquement le caractère aléatoire de nombres pseudo aléatoires, TestU01 et Practrand. Après avoir effectué ces tests sur des algorithmes de chiffrement bien connus, certaines failles ont été exposées et une nouvelle proposition visant à améliorer le système de chiffrement ultra-léger Speck est proposée. La principale contribution de ce travail est d'obtenir une meilleure version par rapport à Speck. Dans cette nouvelle proposition, appelée Speck-R, nous utilisons seulement 7 itérations contrairement à Speck qui en utilise 26 et nous réduisons le temps d'exécution d'au moins 50%. Tout d'abord, nous validons que Speck-R répond aux tests de statistiques pour mesurer l'aléatoire, proposés précédemment. De plus, nous avons rajouté un système de clé dynamique qui procure plus de sécurité contre les attaques liées à la clé. Speck-R a été implémenté sur différentes cartes de type arduino et dans tous les cas, Speck-R était plus rapide que Speck. Ensuite, afin de prouver que ce chiffrement peut être utilisé pour sécuriser les images, en particulier dans les réseaux VANETS/IoV, plusieurs tests ont été effectués et les résultats montrent que Speck-R possède effectivement le haut niveau de sécurité souhaité. Des expérimentations valident notre proposition du point de vue de la sécurité et de la performance et démontrent la robustesse du système proposé face aux types d'attaques les plus connus<br>Living in an era where new devices are astonishing considering their high capabilities, new visions and terms have emerged. Moving to smart phones, Wireless Sensor Networks, high-resolution cameras, pads and much more, has mandated the need to rethink the technological strategy that is used today. Starting from social media, where apparently everything is being exposed, moving to highly powerful surveillance cameras, in addition to real time health monitoring, it can be seen that a high amount of data is being stored in the Cloud and servers. This introduced a great challenge for their storage and transmission especially in the limited resourced platforms that are characterized by: (a) limited computing capabilities, (b) limited energy and source of power and (c) open infrastructures that transmit data over wireless unreliable networks. One of the extensively studied platforms is the Vehicular Ad-hoc Networks which tends to have many limitations concerning the security field. In this dissertation, we focus on improving the security of transmitted multimedia contents in different limited platforms, while preserving a high security level. Limitations of these platforms are taken into consideration while enhancing the execution time of the secure cipher. Additionally, if the proposed cipher is to be used for images, the intrinsic voluminous and complex nature of the managed images is also taken into account. In the first part, we surveyed one of the limited platforms that is interesting for many researchers, which is the Vehicular Ad-hoc Networks. In order to pave the way for researchers to find new efficient security solutions, it is important to have one reference that can sum most of the recent works. It almost investigates every aspect in this field shedding the light over different aspects this platform possesses. Then, in order to propose any new security solution and validate its robustness and the level of randomness of the ciphered image, a simple and efficient test is proposed. This test proposes using the randomness tools, TestU01 and Practrand, in order to assure a high level of randomness. After running these tests on well known ciphers, some flaws were exposed. Proceeding to the next part, a novel proposal for enhancing the well-known ultra lightweight cipher scheme, Speck, is proposed. The main contribution of this work is to obtain a better version compared to Speck. In this proposal, 26 rounds in Speck were reduced to 7 rounds in Speck-R while enhancing the execution time by at least 50%. First, we validate that Speck-R meets the randomness tests that are previously proposed. Additionally, a dynamic substitution layer adds more security against key related attacks and highly fortifies the cipher. Speck-R was implemented on different limited arduino chips and in all cases, Speck-R was ahead of Speck. Then, in order to prove that this cipher can be used for securing images, especially in VANETS/IoV, where images can be extensively re/transmitted, several tests were exerted and results showed that Speck-R indeed possesses the high level of security desired in any trusted cipher. Extensive experiments validate our proposal from both security and performance point of views and demonstrate the robustness of the proposed scheme against the most-known types of attacks

A rapid global decline in biodiversity has been observed in the past few decades, especially in large vertebrates and the habitats supporting these animal populations. This widely accepted fact has made it very important to understand how animals respond to modern ecological threats and to understand the ecosystems functions. The motion activated camera (also known as a camera trap) is a common tool for research in this field, being well-suited for non-invasive observation of wildlife. The images captured by camera traps in biological studies need to be classified to extract information, a traditionally manual process that is time intensive. Recent studies have shown that the use of machine learning (ML) can automate this process while maintaining high accuracy. Until recently the use of machine learning has required significant computing power, relying on data being processed after collection or transmitted to the cloud. This need for connectivity introduces potentially unsustainable overheads that can be addressed by placing computational resources on the camera trap and processing data locally, known as edge computing. Including more computational power in edge and IoT devices makes it possible to keep the computation and data storage on the edge, commonly referred to as edge computing. Applying edge computing to the camera traps enables the use of ML in environments with slow or non-existent network accesss since their functionality does not rely on the need for connectivity. This project shows the feasibility of running machine learning algorithms for the purpose of species identification on low-cost hardware with similar power to what is commonly found in edge and IoT devices, achieving real-time performance and maintaining high energy efficiency sufficient for more than 12 hours of runtime on battery power. Accuracy results were mixed, indicating the need for more tailor-made network models for performing this task and the importance of high quality images for classification.

This paper shows that the two most common impedance transformation networks for power amplifiers (PAs) can be designed to achieve optimum transformation at two frequencies. Hence, a larger bandwidth for the required impedance transformation ratio is achieved. A design procedure is proposed, which takes imperfections like losses into account. Furthermore, an analysis method is presented to estimate the maximum uncompressed output power of a PA with respect to frequency. Based on these results, a fully integrated PA with a dual-band impedance transformation network is designed and its functionality is proven by large signal measurement results. The amplifier covers the frequency band from 450 MHz to 1.2 GHz (3 dB bandwidth of the output power and efficiency), corresponding to a relative bandwidth of more than 100%. It delivers 23.7 dBm output power in the 1 dB compression point, having a power-added efficiency of 33%.

Wireless networks are an iconic technology of today’s modern era, theyare present in our daily activities as can be exemplified by cellular communications,wi-fi, bluetooth, and others. Vector measurements play an importantrole in the design, simulation, and testing of wireless networks and are usedto characterize key devices operating in the radio interface, such as amplifiers,filters, and mixers.Accurate characterization is the key for improving the capacity and efficiencyof wireless networks. As the demand for network capacity continuouslyincreases, the accuracy of vector measurements must also improve. Further,it is anticipated that such trends will continue in the years to come. Consequently,the wireless industry needs to include nonlinear behavior in theircharacterization and analysis, to assess and guaranty the operation of the devices,and to comply to the specifications from governmental regulations. Incontrast to linear behavior, nonlinear behavior presents an additional bandwidthrequirement because the signal bandwidth grows when it passes throughnonlinear devices. In this thesis, vector measurements for devices operatingin wireless networks are studied, emphasizing a synthetic approach for theinstrumentation. This approach enables the use of digital post-processing algorithms,which enhances the measurement accuracy and/or speed and canovercome hardware impairments. This thesis presents the design of a vectorialmeasurement system for wireless devices considering the aforementionedtrends and requirements. It also explores the advantages of the proposedapproach, describes its limitations, and discusses the digital signal processingalgorithms used to reach its final functionality. Finally, measurement resultsof the proposed setup are presented, analyzed and compared to those of modernindustrial instruments.<br><p>QC 20130204</p>

With the development of the Internet, new network services with strict network latency requirements have been made possible. These services are implemented as distributed systems deployed across multiple geographical locations. To provide low response time, these services require knowledge about the current network latency. Unfortunately, network latency among geo-distributed sites often change, thus distributed services rely on continuous network latency measurements. One goal of such measurements is to differentiate between momentary latency spikes from relatively long-term latency changes. The differentiation is achieved through statistical processing of the collected samples. This approach of high-frequency network latency measurements has high overhead, slow to identify network latency changes and lacks accuracy. We propose a novel approach for network latency estimation by correlating network paths to network latency. We demonstrate that network latency can be accurately estimated by first measuring and identifying the network path used and then fetching the expected latency for that network path based on previous set of measurements. Based on these principles, we introduce Sudan traceroute, a network latency estimation tool. Sudan traceroute can be used to both reduce the latency estimation time as well as to reduce the overhead of network path measurements. Sudan traceroute uses an improved path detection mechanism that sends only a few carefully selected probes in order to identify the current network path. We have developed and evaluated Sudan traceroute in a test environment and evaluated the feasibility of Sudan traceroute on real-world networks using Amazon EC2. Using Sudan traceroute we have shortened the time it takes for hosts to identify network latency level changes compared to existing approaches.<br>Med utvecklingen av Internet har nya nätverkstjänster med strikta fördröjningskrav möjliggjorts. Dessa tjänster är implementerade som distribuerade system spridda över flera geografiska platser. För att tillgodose låg svarstid kräver dessa tjänster kunskap om svarstiden i det nuvarande nätverket. Tyvärr ändras ofta nätverksfördröjningen bland geodistribuerade webbplatser, således är distribuerade tjänster beroende av kontinuerliga mätvärden för nätverksfördröjning. Ett mål med sådana mätningar är att skilja mellan momenta ökade svarstider från relativt långsiktiga förändringar av svarstiden. Differentieringen uppnås genom statistisk bearbetning av de samlade mätningarna. Denna högfrekventa insamling av mätningar av nätverksfördröjningen har höga overheadkostnader, identifierar ändringar långsamt och saknar noggrannhet. Vi föreslår ett nytt tillvägagångssätt för beräkningen av nätverksfördröjning genom att korrelera nätverksvägar till nätverksfördröjning. Vi visar att nätverksfördröjningen kan vara exakt uppskattad genom att man först mäter och identifierar den nätverksväg som används och sedan hämtar den förväntade fördröjningen för den nätverksvägen baserad på en tidigare uppsättning av mätningar. Baserat på dessa principer introducerar vi Sudan traceroute, ett Verktyg för att uppskatta nätverksfördröjning. Sudan traceroute kan användas för att både minska tiden att uppskatta fördröjningen samt att minska overhead för mätningarna i nätverket. Sudan traceroute använder en förbättrad vägdetekteringsmekanism som bara skickar några försiktigt valda prober för att identifiera den aktuella vägen i nätverket. Vi har utvecklat och utvärderat Sudan traceroute i en testmiljö och utvärderade genomförbarheten av Sudan traceroute i verkliga nätverk med hjälp av Amazon EC2. Med hjälp av Sudan traceroute har vi förkortat den tid det tar för värdar att identifiera nätverksfördröjnings förändringar jämfört med befintliga tillvägagångssätt.