During the last few years Unmanned Air Vehicles have seen a widespread utilization, both in civilian and military scenarios, because of the benefits of replacing the human presence in unsuitable or hostile environments and dangerous or dull tasks. Examples of their use are surveillance, firefighting, rescuing, extreme photography and environmental monitoring. The main interest of this work is autonomous navigation of such air vehicles, specifically quadrotor helicopters (quadrocopters), and the focus is on convergence to a target destination with collision avoidance. In this work, a general model for a quadrocopter UAV is obtained, making use of a first principles modeling approach, and system identification is exploited in order to relate in a suitable manner the control signals to the effective behavior of the vehicle. The main contribution is the design of a controller for convergence and avoidance of static obstacles, based both on considerations on the dynamics of the agent and knowledge of the testbed for the experiments. The controller is composed of a layered structure. The external layer consists in the computation of a collision-free path leading to the target position and is based on a navigation function approach. The inner layer is meant to make the vehicle follow the waypoints imposed by the outer layer and thus consists in a position controller. Experiments have been conducted in different scenarios in order to analyze the behavior of the controlled system. The final part of the work regards the design of a controller for 3D navigation and collision avoidance for an air vehicle with more constrained dynamics in respect to the quadrocopter. This controller exploits both dipolar navigation functions and model predictive control in order to obtain the control inputs that safely lead the vehicle to its destination with the desired orientation.

To reduce the consequences of a major fault in the electric power grid, functioning parts of the grid can be divided into smaller grid islands. The grid islands are operated isolated from the power network, which places new demands on a faster frequency regulation. This thesis investigates a Kaplan turbine hydropower plant operating an island grid. The Kaplan turbine has two control signals, the wicket gate and the turbine blade positions, controlling the mechanical power. The inputs are combined to achieve maximum turbine efficiency at all operating points. In relative terms, the wicket gate has a fast dynamic but small effect on the mechanical power, while the turbine blade has slow dynamic and large effect on the output, seen around an operating point. The proposed method to get a faster frequency control uses a different combination of the turbine inputs, transferring control effect from the turbine blades to the wicket gates at the cost of loss of turbine efficiency. The method is investigated with time domain simulations on a model containing all essential parts of a Kaplan turbine hydropower plant.

The expansion of renewable energies and the deregulations of the energy market are increasing the demand of regulating power. In absence of hydropower, thermal power is often used for this purpose instead. This thesis focuses on the Vattenfall owned power plant Jänschwalde located in Germany. The goal is to optimize the start-up procedure, so that the start-up time is reduced without causing too much thermal stress on the important thickwalled components in the boiler. By reducing the start-up time, the plant can become more protable, 　exible and better suited to regulate the electricity market. A start-up model was built in Dymola and validated against measurement data and a simulation model of Jänschwalde, which is too complex to use for optimization purposes. The JModelica.org platform was used for the optimization part of the project. It was possible to nd optimal solutions for the start-up process of the Jänschwalde power plant, but the convergence of the optimization algorithm was very dependent on the optimization options used and the scaling of the plant model. Further work includes development of the components used in the start-up model, rening the discretization and scaling for the optimization problem.

An extension of the linear consensus protocol for agents moving in the plane is considered. For single integrator agents the use of a vector perpendicular to the standard consensus feedback leads to a large family of trajectories. If the new perpendicular term is applied only sustained oscillations are facilitated. For special congurations the form of the system trajectories is given in form of eigenvalues and {vectors of the system matrix. A proof is given that this additional term does not eect stability. On the other hand it is motivated that robustness against discrete implementation and switching topologies can be decreased. The control strategy is also applied to agents with double integrator dynamics. Stability can be archived with suciently high velocity feedback and the eect of this feedback on the system performance is further discussed. Using the results for single integrators a self{triggered consensus control strategy is proposed based on the assumption of bounded input magnitude of the other agents. Additional communication of the actual input leads to asymptotic convergence. By applying similar reasoning it is shown how local controllers at the agents can avoid circular regions in state{space while moving towards consensus.

This thesis describes an object oriented programming software that implements a set of automatic controllers for the management-over-the-cloud of Heating, Venting and Air Conditioning (HVAC) systems. More precisely, the toolbox provides general Model Predictive Controls (MPCs), Proportional, Integrative and Derivatives (PIDs), and Linear Quadratic Regulators (LQRs) objects for automatically and remotely actuate any HVAC system over the Internet. The aim is to endow users with some libraries that enable the fast and easy prototyping of control strategies, and thus to boost research on the management of variables such as temperatures, humidity and CO2 levels in buildings. The library has been then exploited to manage a laboratory testbed located at the KTH in Stockholm, and has been shown to provide the research community an instrument for sharing research results and eorts.

This report presents a layer-optimized streaming technique for delivering video content over the Internet using quality-scalable motion-compensated orthogonal video. We use Motion-Compensated Orthogonal Transforms (MCOT) to remove temporal and spatial redundancy. The resulting subbands are quantized and entropy coded by Embedded Block Coding with Optimized Truncations (EBCOT). Therefore, we are able to encode the input video into multiple quality layers with sequential decoding dependency. A layer-distortion model is constructed to measure the trade-off between expected streaming layer and expected distortion. Due to the sequential dependency among streaming layers, we build a cost function with concave properties. With that, we develop a fast algorithm to find the optimal trans-mission policy at low computational complexity. The experiments demonstrate the advantages of expected distortion and computational complexity for challenging streaming scenarios.

Electrical arcing in the railway environment radiates wideband noise that can disturb nearbysensitive electronics e.g. signaling systems. High frequency disturbances follow conducting wiresand rails. For Electromagnetic Compatibility reasons, radiation from wire structures is studied atspecific frequencies, to assess the risk to sensitive electrical devices further along the track.The specific problem is to construct a fast and accurate solver for computing the currents inducedby a dipole source on an arbitrary number of parallel wires inside a layered medium. In this project, atwo-layered medium is considered, each medium defined by arbitrary permittivity and permeability.Once the wire currents are known, the fields may be calculated everywhere.The approach is to transform the circular wires into equivalent strips, creating a planarly layeredstructure. The fields from a source in a layered structure may be expanded as a sum of planewaves propagating in the direction of stratification. The strip currents are expanded into Chebyshevpolynomials and together with the fields and boundary conditions, the currents are solved with themode matching technique in the spectral domain.Moreover, a simplified model for strips much narrower than the wavelength is derived, onlyconsidering axial currents, further reducing the complexity of the problem and still exhibiting highaccuracy.Unlike most full wave methods, the spectral domain approach does not rely on spatial discretisationof wires. Since infinitely long wires are considered, they are electrically large and thusconventional full wave methods yield to massive computations as many grid points are required, thusmotivating the spectral domain approach.The result is a linear system of integral equations solving the currents on an arbitrary numberof narrow strips. Results for different configurations of strips and dipole locations are shown andverified by comparing them with a commercial Method of Moments based solver, 4NEC2, using finitelengths of wire.In conclusion, a simple, fast and accurate method has been developed in Matlab for solving thecurrents along parallel strip conductors. / Blixtar längs järnvägen strålar bredbandigt brus som kan störa närligande känslig elektronik t.ex.signaleringssystem. Högfrekventa störningar följer ledningar och spår. För elektromagnetisk förenlighet,studeras strålning från ledningstrukterer vid specifika frekvenser, för att bedömma risken förkänsliga system längre bort längs ledningen.Problemet är att konstruera en snabb och nogrann numerisk lösare för att beräkna strömmarnainducerade av en dipolkälla hos en godtycklig antal parallella ledare innuti en lagrad struktur. Idetta projekt används en tvålagrig struktur, varje medium definierad av godtycklig permittivitet ochpermeabilitet. Då strömmarna i ledningarna är kända, kan fälten bestämmas överallt.Metoden är att transformera de circulära ledarna till ekvivalenta strips, vilket skapar ett plantlagrat struktur. Fält som härör en källa i ett lagrat medium kan expanderas som en summa planavågor som utökar sig i avlagringsrikningen. Strömmar på stripen expanderas som Chebyshev polynomoch tillsammans med fält och gränsvärden, kan strömmarna tas fram med modanpassning ispektrala rummet.Utöver detta, har en simplifierad model för strips mycket smalare än våglängen härleds fram,där enbart strömmar i stripens parallella riktning utses vilket vidare förenklar problemet utan attförlora mycket nogrannhet.Olikt de alra flest helvågsmetoderna, använder sig inte spektrala metoden av rumdiskretiseringav ledare. Eftersom ledarna i frågan är oändligt långa är de elektriskt stora som gör att rumdiskretiseringkommer att orsaka för stora beräkningar då det kräver för många diskretiserings punkter,vilket motiverar valet av spektrala metoden.Resultatet är ett linjärt ekvationssystem bestående av integralekvationer för att lösa strömmarnalängs en godtycklig antal smala ledare. Resultat för olika konfigurationer av strips och dipolpositionär redovisade och överensstämmer väl med den kommersiella Method of Moments baserade lösaren4NEC2 med ändligt långa ledare.Sammanfattningsvis, har det tagits fram en enkel, snabb och nogrann metod i Matlab för attlösa strömmarna längs parallella ledande strips.

The demand for wireless communication systems has continued to grow steadily and a number oftechnological advances have been made during this growth. To increase system capacity for wirelesssystems by means of uncorrelated propagation paths, a MIMO (Multiple Input Multiple Output) systemhas been considered a better technology for improving the capacity. For a MIMO system to functionefficiently – uncorrelated or at least essentially uncorrelated - transmitted signals are required whichmeans in this context that the polarizations are essentially orthogonal to each other.In this experiment we designed a triple orthogonal polarized antenna with the radiation patterns equalto a dipole antenna for testing purposes. With three orthogonal antennas it should be easy to switchpolarization depending on the signals you want to be measured without having to move the structurephysically.Various possibilities have been examined such as patch antennas and wire antennas required toradiate 360 degrees in one plane. The main challenge was to find a structure that would allow thecabling to be connected to the antenna without interfering with any of the three orthogonal dipolepatterns.Another requirement was to have the same phase center for all three antennas so that measurementsfor all three polarizations can be made without moving the antenna or phase corrections.

In scheduled cooperative localization, devices in a network transmit pulses in a predetermined order. Every node measures inter-arrival time between received pulses to localize other nodes in the network. This thesis has looked into a few aspects of schedule based cooperative localization.First part of the thesis discusses constraints on schedule construction and algorithm to construct a schedule. An algorithm is proposed to design all possible schedules for any number of nodes.The second part of this thesis discusses designing a filter to process inter-arrival times to estimate position of all other nodes in the network. Every node runs a Kalman fi lter. Simulation results are presented for a fi lter model. Complexity of implementing the designed fi lter is discussed. A comparison is made between various ways to invert the matrix by QR decomposition. And finally, increase in fi lter complexity at every node with increase in number of nodes in network is discussed.

Two hop relay based networks consist of three network nodes: source,relay station, and destination in which relay station assists the sourceto communicate reliably and efficiently with the destination. Moreover,these networks provide cost efficient solution for achieving highdata rate via cooperative communication between relays with singleantennas. In two hop relay based networks, communication from a source todestination takes place over two phases, i.e , in first phase from sourceto relay station and in second phase from relay station to the destination.Therefore, it is essential to formulate transmission strategies,i.e, TDMA, SDMA, Hybrid TDMA-SDMA and multicast in terms ofresource allocation, beamforming over two phases so that interferenceis taken into account and high data rates are achieved. In this thesis,some relay selection methods have been proposed to optimize thenetwork performance. Different proposed transmission strategies arecompared in different scenario settings in order to analyse and decidethe best strategy in each setting. Based upon simulation results it is recommended to use adaptivetime split ratio between the two phases. Brute force relay selection givesthe optimal relay assignment but Hungarian assignment algorithm alsoperforms pretty close to brute force performance. SDMA with cooperativerelays connection with multiple antennas at the relays performsmuch better than the other transmission strategies. However, multicaststrategy performs much better if second phase channel knowledge is notavailable at the base station.