Given the breadth of constraint satisfaction problems (CSPs) and the wide variety of CSP solvers, it is often very difficult to determine a priori which solving method is best suited to a problem. This work explores the use of machine learning to predict which solving method will be most effective for a given problem. We use four different problem sets to determine the CSP attributes that can be used to determine which solving method should be applied. After choosing an appropriate set of attributes, we determine how well j48 decision trees can predict which solving method to apply. Furthermore, we take a cost sensitive approach such that problem instances where there is a great difference in runtime between algorithms are emphasized. We also attempt to use information gained on one class of problems to inform decisions about a second class of problems. Finally, we show that the additional costs of deciding which method to apply are outweighed by the time savings compared to applying the same solving method to all problem instances.

Ultrafast electron diffraction (UED) is a method of directly imaging system dynamics at the atomic scale with picosecond time resolution. In this thesis I present theoretical analyses of the experimental processes, and construct models in order to better understand UED experiments and to guide future refinements. In particular, I derive a model of electron bunch propagation and a model of electron bunch diffraction, where both models take into account all bunch parameters. To analyse the propagation of electron bunches, I present a mean-field analytic Gaussian (AG) model. I derive a system of ordinary differential equations that are solved quickly and easily to give the bunch dynamics. The AG model is compared to N -body numerical simulations of initially Gaussian bunches, and I demonstrate excellent agreement between the two result sets. I also present a comparison of the AG model with numerical simulations of quasi-Gaussian and non-Gaussian distributions, extending the applicability of the AG model to the propagation of ``real-world'' bunches. During propagation, electron bunches can be shaped by electron-optic devices, which are necessary to attain high brightness, sub-100 fs bunches. I investigate two types of electron-optic devices: one is a magnetic lens used for collimating or focusing bunches, the other is a bunch compressor. I derive bunch parameter transformations for each of the electron-optic devices, and present numerical calculations using these transformations along with the AG model showing the effects of the devices on the evolution of the bunch parameters. To analyse electron bunch diffraction in UED experiments, I present a general scattering formalism. Using single-scattering and far-field approximations, I derive an expression for the diffracted signal that depends on the electron bunch properties just before scattering. Using this expression I identify the transverse and longitudinal coherence lengths and discuss the importance of these length scales in diffraction pattern formation. I also discuss the effects of different bunch parameters on the measured diffracted flux, and present sample numerical calculations for scattering by nanosize particles based on this model. This simulation demonstrates the cumulative effects of the bunch parameters, and shows the complex interplay of the bunch and target properties on the diffracted signal.

The importance of understanding non-metallic material behaviour at high strain rates is becoming ever more important as new materials are being developed and used in shock loading applications. Applying conventional methods for high strain rate testing to non-metallic materials proved ineffective due to impedance mismatch between the specimen and apparatus and so a new test method was developed. A polymeric Hopkinson bar was developed enabling non-metallic materials, such as polycarbonate and rubber, to be tested at strain rates from 500 s^-1 to 4000 s^-1. Conventional Hopkinson bar analysis is invalid due to the viscoelastic nature of the polymeric bar material. As waves propagate along the bar length, the inherent material behaviour causes the waves to undergo a degree of attenuation and dispersion. Through the use of spectral analysis, a comparison of the dispersive relationships for 6061 T-6 aluminium, extruded acrylic and low density polyethylene is presented. The application of spectral methods to viscoelastic wave analysis was validated by three separate methods. A comparison of predicted and measured waves along the bar length allowed the dispersive relationship to be substantiated. The use of an enhanced laser velocity system further verified the predicted wave magnitude. A comparison of results for polycarbonate and ballistic gelatin to published results showed good agreement.

This thesis is a project carried out at the “Centre for RF measurements Technology“at the University of Gävle. The first aim was basically the characterization of different industrial indoor environments to get a model that describes dispersive features of each environment.   The results of previous measurements campaign on three industrial environments as steel mill, storage paper and industrial process mill are used. Also new Power Delay Profile (PDP) on corridor and laboratory has been developed.   Measurements for three frequency bands are done (183-683 MHz, 1640-2140MHz and 2200-2700MHz) and for line-of-sight (LOS) and non-line-of-sight (NLOS) industrial and laboratory scenaries cases are presented.   All these models have been compared with other existing models as Saleh-Valenzuela Model, Two Cluster Model and Indoor Power Delay Profile Model (IPDP Model) and fit-line, typical deviation are shown.   Finally we present a study of the different systems used in the industry and the best suited system to the conditions is chosen.

The report treats the development and evaluation of a three dimensional user distribution model for a real network simulator. The simulator is used to create realistic predictions of real networks with the use of high resolution maps including a building data base and network data and also an advanced radio model for LTE. Previously all simulations have been performed with a two dimensional user distribution, i.e. all users situated on the ground level. Since it is considered plausible that many LTE users will be indoors in buildings with multiple floors, several three dimensional user distribution models with users not only on the ground floor but also on the higher floors has been developed and implemented in the simulator. The models all account for the change in path loss and SINR to be expected and have been compared in computational time and credibility. The simulations show that by the use of such a three dimensional model there is a significant improvement at low loads but at high loads the interference becomes dominant and the results show a deterioration and approaches the results of the ordinary two dimensional model. The seventh and last model to be investigated shows a desirable computational speed that still does not compromise too much with the accuracy and detailing of the model and is therefore recommended for normal use.

The importance of understanding non-metallic material behaviour at high strain rates is becoming ever more important as new materials are being developed and used in shock loading applications. Applying conventional methods for high strain rate testing to non-metallic materials proved ineffective due to impedance mismatch between the specimen and apparatus and so a new test method was developed. A polymeric Hopkinson bar was developed enabling non-metallic materials, such as polycarbonate and rubber, to be tested at strain rates from 500 s^-1 to 4000 s^-1. Conventional Hopkinson bar analysis is invalid due to the viscoelastic nature of the polymeric bar material. As waves propagate along the bar length, the inherent material behaviour causes the waves to undergo a degree of attenuation and dispersion. Through the use of spectral analysis, a comparison of the dispersive relationships for 6061 T-6 aluminium, extruded acrylic and low density polyethylene is presented. The application of spectral methods to viscoelastic wave analysis was validated by three separate methods. A comparison of predicted and measured waves along the bar length allowed the dispersive relationship to be substantiated. The use of an enhanced laser velocity system further verified the predicted wave magnitude. A comparison of results for polycarbonate and ballistic gelatin to published results showed good agreement.

The enhanced Polar Outflow Probe (ePOP) satellite is to be launched in 2007. One of the 8 instruments it will carry is a Radio Receiver Instrument (RRI) which is a passive radio receiver. The RRI will detect HF (High Frequency band 3 to 30 MHz) radio waves from ground transmitters, one of which is the Saskatoon SuperDARN radar. The modification of an HF radar wave as it propagates through the ionosphere to the satellite is the dominant scientific interest of this thesis. The modification of a radar wave as it propagates through the ionosphere can be used to characterize the ionosphere and reveal a better understanding of magnetoionic radar wave propagation. A ray tracing program has been written to determine characteristics of the wave, including the wave path and the full polarization state, at the satellite receiver. </p> As a confirmation of the ray tracing program abilities, data from a similar transionospheric experiment in 1978, the ISIS II satellite mission, has been analyzed and compared with simulated results. The ISIS II transionospheric experiment received radar signals from a transmitter (9.303 MHz) located in Ottawa, Canada. These signals were analyzed and it was noted that the signal periodically faded in and out both due to differential Faraday rotation effects (due to propagation through an ionized medium and reception on a single dipole antenna) and due to satellite spin rotation at rates up to 13 Hz. Also observed was a splitting of the received signal into Ordinary (O-mode) and Extraordinary (X-mode) components causing a delay between the arrival of the modes at ISIS II of up to 0.8 ms. Simulations have been carried out to model the radar wave propagation from the ground transmitter through the ionized medium of the ionosphere to the spacecraft. The modelled signal shows very similar trends to the observed signal. A linear regression analysis comparing observed to simulated fade rates gave values of slope equal to 1.07 and regression coefficient equal to 0.934. The regression analysis of mode delay gave values of slope equal to 1.14 and regression coefficient equal to 0.930. </p>Ray path modelling has also been simulated for the RRI experiment on ePOP. These simulations have been carried out for various ionospheric and satellite pass configurations. The expected fade rates, mode delays, and latitudinal range of received signal was determined from the simulated data in each case. The dependence of these characteristics on the various pass configurations are presented and discussed.

Given the breadth of constraint satisfaction problems (CSPs) and the wide variety of CSP solvers, it is often very difficult to determine a priori which solving method is best suited to a problem. This work explores the use of machine learning to predict which solving method will be most effective for a given problem. We use four different problem sets to determine the CSP attributes that can be used to determine which solving method should be applied. After choosing an appropriate set of attributes, we determine how well j48 decision trees can predict which solving method to apply. Furthermore, we take a cost sensitive approach such that problem instances where there is a great difference in runtime between algorithms are emphasized. We also attempt to use information gained on one class of problems to inform decisions about a second class of problems. Finally, we show that the additional costs of deciding which method to apply are outweighed by the time savings compared to applying the same solving method to all problem instances.

Careful network planning has become increasingly critical with the rising deployment, coverage, and congestion of wireless local area networks (WLANs). This thesis outlines the achieved prediction accuracy of a direct-ray, single path loss exponent, adapted Seidel-Rappaport propagation model as determined through measurements and analysis of the established 2.4 GHz, 802.11g outdoor WiFi network deployed on the campus of the Georgia Institute of Technology. Additionally, the viability of using the obtained model parameters as a means for planning future network deployment is discussed. Analysis of measured data shows that accurate predictive planning for network coverage is possible without the need for overly complicated modeling techniques such as ray tracing. The proposed model performs with accuracy comparable to other commonly accepted, more complicated models and is offered as a simple, yet strong predictive model for network planning having both speed and accuracy. Results show, that for the area under study, the standard deviation of the prediction error for the proposed model is below 6.8dB in all analyzed environments, and is approximately 5.5dB on average. Further, the accuracy of model predictions in new environments is shown to be satisfactory for network planning.

Taiwan is a small island state surrounded by the oceans but with large population. With limited land space, it would be worthwhile considering how to stabilize the existing coast or to create stable artificial beaches. Under the onslaught of storm surge and large wave from typhoons, beach erosion would occur accompanying by formation of a submerged bar beyond the surf zone with the sand removed from the beach. After the storm, the bar material maybe transport back by the swell and predominant waves which helps recover the original beach, thus producing a beach profile in dynamic equilibrium. The main purpose of this research is to use the back-propagation neural network¡]BPNN¡^, which trains a sample model and creates a system for the estimation, prediction, decision making and verification of an anticipated event. By the BPNN, we can simulate the key characteristic parameters for the storm beach profile resulting from typhoon action. Source data for training and verification are taken from the experimental results of beach profile change observed in large-scale wave tank¡]LWT¡^conducted by Coastal Engineering Research Center¡]CERC¡^in the USA in the 1960s and that from the Central Research Institute of Electric Power Industry in Japan in the 1980s. Some of the data are used as training pairs and others for verification and prediction of the key parameters of berm erosion and bar formation. Through literature review and simulation on the related parameters for storm beach profile, methodology for the prediction of the beach profile and bar/berm characteristics can be established.