No description available.

No description available.

Connected and autonomous vehicles (CAVs) are an emerging trend in the transport sector and their impact on transportation, the economy, society and the environment will be tremendous. Much like the automobile shaped the way humans travelled, lived and worked during the 20th century, CAVs have yet again the potential to affect and reform all of these areas. Besides the imminent technological challenges on the robotic aspect of making CAVs become a market-ready reality, a plethora of ethical, social and legal questions will have to be addressed along the line. Knowledge of and interaction with the surrounding infrastructure and other actors in the system will be essential for CAVs in order to pave the way for progressive solutions to urgent sustainability and mobility issues in transportation. Road networks, i.e. the networks of roads and intersections, are the core infrastructure on which CAVs will operate. Thus, having detailed knowledge about them is key for CAVs in order to take the right decisions on both short-term actions that will affect individual traffic users in immediate situations and long-term actions that will affect entire transportation systems in the long run. Machine learning is nowadays a popular choice to extract and conglomerate knowledge from large amounts of data – and large amounts of data can be obtained about road networks. However, classical machine learning models are incapable of harnessing the graph-structured nature of road networks sufficiently. Graph neural networks (GNNs) are machine learning models of growing popularity that can explicitly leverage the complex topological structure of node dependencies in graphs, such as the ones observed in road networks. Road networks are sparse graphs that reside in a euclidean space, and therefore different to typical graphs studied in the literature. Also, crowd-sourced road network graphs often have incomplete attributes and are generally lacking the fine-grained level of detail in their encoded information that would be required for CAVs. Identifying the best representation of road network graphs and complementing their lacking detail with auxiliary data is therefore an important research direction. This thesis, therefore, addresses data-driven classification in road networks from two directions: A) the general approach of learning on spatial graphs of road network with GNNs, and B) complementing road network graphs with auxiliary data. Specifically, this thesis and the included papers address the exemplary task of road classifications and make the following contributions to the field: Paper A analyses how GNNs can be applied to road networks and how the networks are best represented. Different aggregator functions are compared on final classification performances. A novel aggregator and a neighbourhood sampling method are introduced, and the line graph transformation is identified as a suitable representation of road network graphs for GNNs. Paper B complements the road network graphs with mobility data from millions of GPS trajectories and introduces an equitemporal node spacing to create road segments of equal travel time. It further introduces remote sensing vision data as a potent complement to overcome shortcom-ings of the graph-based representation for road networks. Simple hand-crafted low-level vision features are used in this work. However, both the equitemporal node spacing and the simple vision features clearly exhibit improved classification performances. Finally, Paper C consolidates the complement of remote sensing data to the road network graphs. Through a general visual feature encoding of state-of-the-art pretrained vision back-bones that are carefully fine-tuned to the remote sensing domain, a further performance boost on the road classification task is achieved. / Vernetzte und autonome Fahrzeuge sind ein aktueller Trend in der Mobilitätsindustrie und ihr Einfluss auf Transportsysteme, die Wirtschaft, die Gesellschaft und die Umwelt wird enorm sein. Die Erwartungen an diese Roboterautos sind dabei extrem hoch. Sie sollen helfen, dringliche Mobilitätsprobleme unserer zunehmend globalisierten, urbanen Lebensweise zu überwinden und nachhaltige Lösungen ermöglichen. Das Wissen um und die Interaktion mit der Infrastruktur und anderen Verkehrsteilnehmern sind dabei ausschlaggebend, um mit vernetzten und autonomen Fahrzeuge progressive Mobilitätslösungen entwickeln zu können. Das Straßennetzwerk ist dabei die Kerninfrastruktur, auf der diese Fahrzeuge agieren werden. Maschinelles Lernen und künstliche Intelligenz sind heutzutage populäre Methoden, um Wissen und Entschlüsse aus riesigen Datenmengen zu ziehen - und über das Straßennetzwerk können riesige Datenmengen gesammelt werden. Klassische Modelle des maschinellen Lernens jedoch nutzen die Graphstrutktur, die den Straßennetzwerken zugrunde liegt, nicht hinreichend aus. Doch in den vergangenen Jahren haben graphbasierte neuronale Netze (GNNs) eine immense Beliebtheit erlangt. GNNs sind Modelle des maschinellen Lernens, die explizit komplexe Topologien und Abhängigkeiten von Knoten in Graphen ausnutzen. Diese Abschlussarbeit befasst sich mit der Anwendung von GNNs auf Straßennetzwerke. Es wird erläutert, welche Repräsentationen von Straßennetzwerke für GNNs vonnöten sind. Gleichzeitig werden die Unzulänglichkeiten der Graphstruktur von Straßennetzwerken dargestellt. Die enthaltenen Publikationen befassen sich damit, Zusatzinformation aus GPS-Sensoren und Satellitenbildern zu entnehmen, um diese Unzulänglichkeiten zu überwinden. Diese Abschlussarbeit umfasst drei Publikationen und in allen dreien wird eine Klassifizierung von Straßen mittels GNNs vorgenommen. Die Klassifizierung ordnet den Straßen verschiedene Kategorien zu, die ihre relative Bedeutsamkeit innerhalb des Straßennetzwerks darstellen. Publikation A analysiert die generelle Anwendung von GNNs auf Straßennetzwerke. Verschiedene Aggregatorfunktionen - d.h. Funktionen, die die Informationen von benachbarten Straßen zusammenfassen - werden im Hinblick auf das Klassifizierungsvermögen miteinander verglichen. Des Weiteren werden eine neue Aggregatorfunktion sowie eine neue Methode zur Nachbarschaftsbestimmung von Straßen im Straßennetzwerk vorgestellt. Außerdem, wird in dieser Publikation der Kantengraph des Straßennetzwerkes, der eine Umkehrung der Knoten und Kanten des ursprünglichen Graphen darstellt, als geeignete Repräsentation hervorgehoben. Publikation B komplettiert das Straßennetzwerk mit Mobilitätsdaten von mehreren Millionen GPS-Trajektorien. Eine äquitemporale Segmentierung des Straßennetzwerkes wird vorgeschlagen. Die Segmentierung erzeugt eine Aufteilung des Straßennetzwerkes, in der jedes Straßensegment dieselbe Reisezeit hat. Außerdem, wird aufgezeigt, wie Satellitenbilder in die Graphstruktur des Straßennetzwerks integriert werden können. Von den Bildern werden handgefertigte, niedrigdimensionale Merkmale extrahiert und dem Straßennetzwerksdaten beigefügt. Sowohl die äquitemporale Segmentierung als auch die Beifügung der Satellitenbilder führen zu einer deutlichen Verbesserung des Klassifizierungsvermögens der GNNs. Abschließend bekräftigt Publikation C, dass Satellitenbilder geeignete Hilfsdaten für Straßennetzwerke sind. Es werden vortrainierte mehrschichtige neuronale Netze genutzt, um hochdimensionale Kodierungen der visuellen Merkmale von Satellitenbildern zu erzeugen. Diese Kodierungen führen erneut zu einer deutlichen Verbesserung des Klassifizierungsvermögens gegenüber der Modelle aus Publikation B. Jedoch zeigen die Ergebnisse aus Publikation C auch, dass diese mehrschichtigen neuronalen Netze behutsam auf die visuellen Merkmale von Satellitenbildern angepasst werden müssen. Dies geschieht durch sogenanntes Transferlernen mithilfe von generellen Datenbanken von Satellitenbildern.

This thesis studies the physical layer (PHY layer) of the IEEE 802.11n wireless local area network (WLAN) standard. The possibility of integrating a receiver designed according to the standard with software-defined radios is investigated. The proposed design was implemented in MATLAB and tested using two softwaredefined radios. One of the radios transmitted IEEE 802.11n signals whilst the other one captured them and sent them to a computer for decoding. In this way, evaluation of the proposed receiver design was done. The tests resulted in successfully decoded WLAN packets, although errors occured regularly due to distortions in the air. The proposed MATLAB design can be developed further, with more features, for future tests and research.

The launch of the 5th Generation (5G) mobile network allows for wireless communication at increased throughput rates and reduced latency compared to its predecessors and has opened for new possibilities in terms of computation offloading where demanding processes can be referred to powerful servers.User equipment (UE), i.e. wireless devices connected to a network, can benefit from offloading computational tasks to servers. Not only does it extend UE's computational resources, but also has the potential to reduce its energy consumption as it effectively redistributes the computational load to the server.This thesis is a study into the energy consumption trade-offs of this procedure for which a case study is done on a computer connected to a 5G network and tasked with the computation of a specific algorithm. Specifically, a comparison is made on the power consumption of computing the algorithm on the UE's central processing unit (CPU) and offloading it via a 5G modem, respectively, and a theoretical framework describing algorithms in terms of their utilization of these components is presented. By experimentally profiling the power consumption of the components and an algorithm's utilization thereof, these trade-offs can be quantified for a variety of signalling conditions. While the empirical study is a test case of a characteristic algorithm on a specific set of hardware components, the developed theoretical framework and methodology allows for the results to be extended to other hardware and algorithms, and general conclusions to be drawn on the energy consumption trade-offs in computation offloading. The results show that computation offloading is overwhelmingly beneficial in terms of power consumption and that the trade-offs only become comparable in certain edge cases. In particular, unless the CPU has an uncommonly low power consumption, the signal quality conditions are very poor or if the algorithm to be offloaded has a combination of low CPU load and high throughput requirements, offloading should always be considered a viable computational procedure in terms of energy consumption.

Internet of Things (IoT) has come in reality to improve our living quality. Automation is embraced in all the possible business verticals that have diverse communication needs ranged from static devices’ sporadic transmission to mobile devices’ every minute transmission. Despite, there are many technologies available today to support IoT services; cellular systems can play a vital role for IoT services, like wearables, vehicular, and industrial IoT, rollout which have either mobility or security concern.  IoT services generated traffic are foreseen as a sporadic-bursty traffic. As the cellular networks are designed to serve continuous data traffic, the existing system’s access control mechanism cannot efficiently conform to the burstiness of traffic. This limits the scope of the network scalability in terms of simultaneous serving devices’ capacity. Also, this bursty pattern can extensively increase the rate of network’s congestion incident. In this thesis, we focus on these underlying challenges to support a large number of heterogeneous IoT services with existing services over the same radio network. An important question for supporting IoT services over cellular networks is how detrimental are the effects of IoT services on other services of cellular networks. This dissertation seeks to answer this with quantitative results to indicate the real constraints of existing networks. An important conclusion is that existing cellular system is incompetent to support the bursty arrival of massive IoT devices in terms of radio networks’ access control plane’s scalability. Therefore, this dissertation presents solutions to overcome the identified limitations of access control planes. To improve the performance of the access control plane, we incorporate a vertical core network controlled group management scheme that can assure the operator’s granular control over capillary gateways. Besides, this introduces a unique handover opportunity between cellular and capillary network vertices. Then, we present a simple but efficient initial access mechanism to overcome the initial access collision at the very early stage. Finally, we show the impact of access collision and retransmission on the initial access resource dimensioning.We present a practical traffic model that is realistic for the traffic scenario for mixed-traffic. Our presented results and analysis depict the trade-offs between access rate, retransmission and resource allocation over time and frequency.Our results reveal that with proposed schemes of the cellular system’s access control plane can be scalable and resilient to accommodate a large number of IoT devices without incurring extra delay or need of resources to the system. / <p>QC 20170928</p>

No description available.

No description available.

In this era of rapid wireless technological advancements, wireless connectivity between humans, humans with machines, and machines with machines is gradually becoming an absolute necessity. The initial motivation for wireless connectivity was to enable voice communication between humans over a geo-graphical area. Thanks to cellular communications advancements in the past decade, cellular wireless connectivity has become a global success, starting from 1G to the present generation 5G. However, the needs of humans often evolve with time, and now the world is witnessing an ever-growing demand for the internet with high data rates besides reliable voice communication. Current cellular networks suffer from non-uniform data rates across a cell, i.e., users at the cell center and the cell edges experience significant variations in signal-to-noise ratio, making the cellular technology less reliable to meet the future data demands. Moreover, cellular networks operating as cells, i.e., an access point (AP, the term we would use instead of base station) serving the users within its geographical location, cannot leverage the network’s total capacity without cooperation among APs of the neighboring cells. One potential solution is moving away from the cell to cell-free networks wherein all the APs will serve all the users within the geographical coverage area. Thus, there is a need for a paradigm shift in how cellular networks operate. Towards the goal mentioned above to fully leverage the network capacity, the Cell-Free Massive multiple-input-multiple-output (MIMO) technology is expected to be the next potential technology beyond 5G combining the benefits of Massive MIMO and cell-free distributed architectures.  Distributed architectures require distributed signal processing algorithms, and also energy consumption of the network is crucial. Keeping in view the practical ease in deployment, we consider a sequentially connected Cell-Free Massive MIMO network called a “radio stripe”. In the first part of the thesis, we focus on developing an optimal sequential algorithm in the sense of mean-square-error (MSE) which has the same performance as that of centralized Cell-Free Massive MIMO implementation with the minimum MSE (MMSE) receiver. We also develop an optimal sequential algorithm that decentralizes the centralized bit LLR computation. Another attractive aspect of these proposed algorithms is that the fronthaul (number of real symbols required by the central processing unit (CPU) to decode the transmitted signal) is independent of the number of APs. On the contrary, centralized implementation fronthaul is dependent on the number of APs, causing scalability problems with the increase in APs.  In the second part of the thesis, we develop an algorithm focused on maximizing the energy efficiency of the RadioWeave network in an underlay spectrum sharing. RadioWeave is a technology envisioned to combine Cell-Free Massive MIMO and possibly large intelligent surfaces. We first present the energy efficiency problem, which is non-convex in its original form. Then, a convex lower bound on the problem is provided with an iterative algorithm to solve the problem efficiently. / I denna tid av snabba trådlösa tekniska framsteg blir trådlös anslutning mellan människor, mellan människor och maskiner och mellan maskiner och maskiner gradvis en absolut nödvändighet. Den ursprungliga motivationen för trådlös anslutning var att möjliggöra röstkommunikation mellan människor över ett stort geografiskt område. Tack vare framsteg inom mobilkommunikation under det senaste decenniet har mobil trådlös anslutning blivit en global framgång, från 1G till den nuvarande generationen 5G. Men människors behov utvecklats med tiden, och nu bevittnar världen, förutom pålitlig röstkommunikation, en ständigt växande efterfrågan på internet med höga datahastigheter. Nuvarande cellulära nätverk lider av olikformiga datahastigheter över cellen, d.v.s. användare i cellcentret och cellkanterna upplever betydande variationer i signal-brusförhållande, vilket gör den cellulära tekniken mindre lämplig för att möta framtida databehov. Dessutom kan cellulära nätverk som fungerar som celler, d.v.s. att varje accesspunkt (AP, termen vi använder istället för basstation) betjänar användarna inom dess geografiska område, inte utnyttja nätverkets totala kapacitet utan samarbete mellan AP:er i de angränsande cellerna. En potentiell lösning är att gå från cellulära till cellfria nätverk där alla basstationer betjänar alla användare inom det geografiska täckningsområdet. Det finns alltså ett behov av ett paradigmskifte i hur cellulära nätverk fungerar. För att fullt ut utnyttja nätverkskapaciteten, förväntas Cell-Free Massive MIMO vara nästa potentiella teknik bortom 5G som kombinerar fördelarna med Massive MIMO och cellfria distribuerade arkitekturer. Distribuerade arkitekturer kräver distribuerade signalbehandlingsalgoritmer, och även energiförbrukningen i nätverket är av stor vikt. Vi studerar sekventiellt anslutna Cell-Free Massive MIMO-nätverk som kallas “radio stripe”, eftersom dessa är enkla att placera ut. I den första delen av avhandlingen fokuserar vi på att utveckla en optimal, ur ett MSE perspektiv, sekventiell algoritm som har samma prestanda som den för centraliserad Cell-Free Massive MIMO-implementering med en MMSE-mottagare. Vi utvecklade också en optimal sekventiell algoritm som decentraliserar den centraliserade bit LLR. En kritisk aspekt av dessa föreslagna algoritmer är att fronthaul (antal reella symboler som krävs av CPU:n för att avkoda den sända signalen) är oberoende av antalet AP:er. Tvärtom är fronthaulen i en centraliserad implementering beroende av antalet AP:er, vilket orsakar skalbarhetsproblem med ökningen av AP:er. I den andra delen av avhandlingen utvecklar vi en algoritm fokuserad på att maximera energieffektiviteten i RadioWeave-nätverk med en underliggande spektrumdelning. RadioWeave är en teknik som är tänkt att kombinera Cell-Free Massive MIMO och möjligen stora intelligenta ytor. Vi presenterar först energieffektivitetsproblemet, som är icke-konvext i sin ursprungliga form. Sedan förses en konvex nedre gräns för problemet med en iterativ algoritm för att lösa problemet effektivt.

In this information society, wireless network is an indispensable technology supporting the daily information communication and various interaction services. With the motivation to improve the user experience of Wi-Fi service, this thesis presents the front-end development process of a visualization application for home Wi-Fi testing and optimization. To provide the desired service to the users, it is important to understand what they need. With prompt feedback, the development process can have much more customized schedule and specific aim. In this thesis, different methods are adopted to get the valuable feedbacks from potential users. Brainstorming and originality interview support the basic designing of application, which is presented by the wireframe prototype. The prototype were used for a test-run and get feedbacks in order to develop a web application front-end assimilating the ideas and suggestions from real users. As more detailed functionality are designed, the feasibility and practicability should be investigated. Questionnaires are used to do larger scale ranges of user investigation. The front-end is designed using HTML, CSS and JavaScript with Bootstrap framework. The elements within all the pages could be able to interact with each other to give the customers a visualized experienced of service. The web application provides the users with the functions to test the real-time Wi-Fi performance, check the performance trend of data history from a specific time period and get useful optimization solutions. The designing is aiming to synthesize enough accessible information and provide the users with self-helpful testing and optimization of home Wi-Fi, which can increase the efficiency of technical support and reduce the workload of customer service. / I detta informationssamhälle är trådlöst nätverk en oumbärlig teknik som stöder den dagliga informationskommunikationen och olika interaktionstjänster. Med motivationen att förbättra användarupplevelsen av Wi-Fi tjänsten presenterar denna avhandling utvecklingsprocessen för fronten av en visualiseringsapplikation för Wi-Fi-test och optimering av hemmet. För att tillhandahålla önskad service till användarna är det viktigt att förstå vad de behöver. Med snabb feedback kan utvecklingsprocessen ha mycket mer anpassat schema och specifikt syfte. I denna avhandling antas olika metoder för att få de värdefulla feedback från potentiella användare. Brainstorming och originalitet intervju stöder den grundläggande utformningen av applikationen, som presenteras av wireframe prototypen. Prototypen användes för en testkörning och fick feedback för att utveckla ett webbapplikationsfrontend assimilera id´eer och förslag från riktiga användare. Eftersom mer detaljerad funktionalitet är utformad bör genomförbarheten och genomförbarheten undersökas. Frågeformulär används för att göra större omfattning av användarutredningen. Fronten är utformad med hjälp av HTML, CSS och JavaScript med Bootstrap-ramverket. Elementen inom alla sidor skulle kunna interagera med varandra för att ge kunderna en visualiserad erfarenhet av service. Webapplikationen ger användarna funktionerna för att testa Wi-Fi-prestanda i realtid, kontrollera prestandatrenden i dataloggen från en viss tidsperiod och få användbara optimeringslösningar. Designen syftar till att syntetisera tillräckligt tillgång till information och ge användarna självhjälpt testning och optimering av Wi-Fi i hemmet, vilket kan öka effektiviteten av tekniskt stöd och minska arbetsbelastningen hos kundtjänst.