The rise of micromobility, most notably electric standing scooters (e-scooters), has resulted in new challenges for traffic planning and road safety. One such issue is the fact that in most European countries, e-scooter users are obliged to ride their vehicle on cycling infrastructure and thereby share this infrastructure with bicyclists. This increases the use of and, subsequently, demand for bicycle lanes, which is an obvious challenge for transport planning. However, for adequate planning and construction of cycling infrastructure, information on route choice behavior of bicyclists and e-scooter users and its influencing factors is necessary. While research on bicyclists' route choice is well advanced, research on e-scooter riders is still in its infancy. For bicyclists, the presence of bicycle facilities, traffic volume, and travel time are among others particularly important for route choice. However, the question arises whether this also applies to e-scooter riders as vehicle dynamics are different and riders are, at least for now, less skilled due to lack of training and exposition. In order to fill this research gap, we aimed to analyze the determinants for route choice of e-scooter users in comparison to bicyclists in a field study. [from Introduction]

The purpose of this qualitative study was to examine teachers' perspectives of the key factors contributing to the retention of rural teachers who entered teaching through an alternate route certification program in Mississippi. It was specifically the goal of this study to understand how alternatively certified teachers perceive their own characteristics (e.g., teacher preparation, personal experiences), school conditions (e.g., students, administration), and compensation (e.g., salary, benefits) to be related to their decision to remain in the profession. In this study, 9 rural alternate route teachers were interviewed from 8 schools in Mississippi. The research questions were: (1) How do rural alternate route teachers who stay describe their decision to continue teaching in terms of teacher characteristics?; (2) How do rural alternate route teachers who stay describe their decision to continue teaching in terms of school conditions?; and (3) How do rural alternate route teachers who stay describe their decision to continue teaching in terms of compensation? Sher's (1983) rural retention 3 C's framework provides a model for understanding retention. Sher proposed that attracting and retaining teachers in rural schools is a function of 3 C's: teacher characteristics, school conditions, and compensation. The data revealed that for teacher characteristics teacher preparation that included practice teaching combined with coursework was important, and participants valued experience working/teaching children. Data also revealed school conditions factors as student were a source of satisfaction for teachers, most teachers had little induction and mentoring support, teachers lacked administration and collegial support, and teachers found networks of support outside the school setting. The data revealed that the relationship between compensation and retention is complex, and that compensation was less important than intangible benefits. Although the study failed to find a simple and direct cause of retention, these findings do provide further insight into teacher retention. The findings of the study suggest implications for teacher preparation, school districts, and policy.

This thesis was motivated to simulate the evacuation traffic of Mississippi Stated University (MSU) main campus using the Path-Following logic of TSIS/CORSIM and to evaluate a set of traffic management plans. Three scenarios of traffic management plans were developed and tested. A NCT of 123 minutes was projected if evacuate without any plan. In comparison, under a pre-planned traffic management plan the NCT would decrease to 39 minutes. Further, if implement contra flow the NCT would reduce to 21 minutes. If even further adjust the signal timing plans at the university exits a NCT of 20 minutes would be achieved. The sensitivity analysis found that the NCT was sensitive to the CORSIM parameters of free flow speed, time to react to sudden deceleration of lead vehicle and the configuration of driver type, while the effects of discharge headway and start up lost time were not found to be significant.

No description available.

No description available.

No description available.

Ce mémoire a pour but d'explorer l'utilisation de données de contexte, notamment spatial, pour prédire de la durée que va prendre un garage pour effectuer les réparations à la suite d'un accident automobile. Le contexte réfère à l'environnement dans lequel évolue le garage. Il s'agit donc de développer une approche permettant de prédire une caractéristique précise en utilisant notamment de l'information historique. L'information historique comprend des composantes spatiales, comme des adresses, qui vont être exploitées afin de générer de nouvelles informations relatives à la localisation des garages automobiles. L'utilisation des données accumulées sur les réclamations automobiles va permettre d'établir un niveau initial de prédiction qu'il est possible d'atteindre avec de l'apprentissage supervisé. En ajoutant ensuite petit à petit de l'information de contexte spatial dans lequel évolue le garage responsable des réparations, de nouveaux niveaux de prédiction seront atteints. Il sera alors possible d'évaluer la pertinence de considérer le contexte spatial dans un problème de prédiction comme celui des temps de réparations des véhicules accidentés en comparant ces niveaux de prédiction précédemment cités. L'utilisation de données historiques pour prédire une nouvelle donnée se fait depuis plusieurs années à l'aide d'une branche de l'intelligence artificielle, à savoir : l'apprentissage machine. Couplées à cette méthode d'analyse et de production de données, des analyses spatiales vont être présentées et introduites pour essayer de modéliser le contexte spatial. Pour quantifier l'apport d'analyses spatiales et de données localisées dans un problème d'apprentissage machine, il sera question de comparer l'approche n'utilisant pas d'analyse spatiale pour produire de nouvelles données, avec une approche similaire considérant cette fois-ci les données de contexte spatial dans lequel évolue le garage. L'objectif est de voir l'impact que peut avoir une contextualisation spatiale sur la prédiction d'une variable quantitative. / The purpose of this paper is to explore the use of context data, particularly spatial context, to predict how long it will take a garage to complete repairs following an automobile accident. The context refers to the environment in which the garage evolves. It is therefore a question of developing an approach that makes it possible to predict a precise characteristic by using historical information in particular. The historical information includes spatial components, such as addresses, which will be exploited to generate new information about the location of car garages. The use of the accumulated data on car claims will allow to establish an initial level of prediction that can be reached with supervised learning. By then gradually adding information about the spatial context in which the garage responsible for the repairs evolves, new levels of prediction will be reached. It will then be possible to evaluate the relevance of considering the spatial context in a prediction problem such as that of the repair times of accidented vehicles by comparing these prediction levels previously mentioned. The use of historical data to predict new data has been done for several years with the help of a branch of artificial intelligence, namely: machine learning. Coupled with this method of data analysis and production, spatial analyses will be presented and introduced to try to model the spatial context. To quantify the contribution of spatial analysis and localized data in a machine learning problem, we will compare the approach that does not use spatial analysis to produce new data with a similar approach that considers the spatial context data in which the garage evolves. The objective is to see the impact that spatial contextualization can have on the prediction of a quantitative variable.

Driver heterogeneity in travel behavior has repeatedly been cited in the literature as a limitation that needs to be addressed. In this work, driver heterogeneity is addressed from four different perspectives. First, driver heterogeneity is addressed by models of driver perceptions of travel conditions: travel distance, time, and speed. Second, it is addressed from the perspective of driver learning trends and models of driver-types. Driver type is not commonly used in the vernacular of transportation engineering. It is a term that was developed in this work to reflect driver aggressiveness in route switching behavior. It may be interpreted as analogous to the commonly known personality-types, but applied to driver behavior. Third, driver heterogeneity is addressed via latent class choice models. Last, personality traits were found significant in all estimated models. The first three adopted perspectives were modeled as functions of variables of driver demographics, personality traits, and choice situation characteristics. The work is based on three datasets: a driving simulator experiment, an in situ driving experiment in real-world conditions, and a naturalistic real-life driving experiment. In total, the results are based on three experiments, 109 drivers, 74 route choice situations, and 8,644 route choices. It is assuring that results from all three experiments were found to be highly consistent. Discrepancies between predictions of network-oriented traffic assignment models and observed route choice percentages were identified and incorporating variables of driver heterogeneity were found to improve route choice model performance. Variables from all three groups: driver demographics, personality traits, and choice situation characteristics, were found significant in all considered models for driver heterogeneity. However, it is extremely interesting that all five variables of driver personality traits were found to be, in general, as significant as, and frequently more significant than, variables of trip characteristics — such as travel time. Neuroticism, extraversion and conscientiousness were found to increase route switching behavior, and openness to experience and agreeable were found to decrease route switching behavior. In addition, as expected, travel time was found to be highly significant in the models that were developed. However, unexpectedly, travel speed was also found to be highly significant, and travel distance was not as significant as expected. Results of this work are highly promising for the future of understanding and modeling of heterogeneity of human travel behavior, as well as for identifying target markets and the future of intelligent transportation systems. / Ph. D.

Recently, many new types of wireless networks have emerged, such as mobile ad hoc networks (MANETs), cognitive radio networks (CRNs) and large scale wireless sensor networks. To get better performance in these wireless networks, various schemes, e.g., metrics, policies, algorithms, protocols, etc., have been proposed. Among them, optimal schemes that can achieve optimal performance are of great importance. On the theoretical side, they provide important design guidelines and performance benchmarks. On the practical side, they guarantee best communication performance with limited network resources. In this dissertation, we focus on the modeling and optimization of routing in wireless networks, including both broadcast routing, unicast routing, and convergecast routing. We study two aspects of routing: algorithm analysis and Qos analysis. In the algorithmic work, we focus on how to build optimal broadcast trees. We investigate the optimality compatibility between three tree-based broadcast routing algorithms and routing metrics. The Qos work includes three parts. First, we focus on how to optimally repair broken paths to minimize impact of path break in MANETs. We propose a provably optimal cached-based route repair policy for real-time traffic in MANETs. Second, we focus on the impact of secondary user (SU) node placement on SU traffic delay in CRNs. We design SU node placement schemes that can minimize the multi-hop delay in CRNs. Third, we analyze the convergecast delay of a large scale sensor network which coexists with WiFi nodes. We derive a closed form delay formula, which can be used to estimate sensor packet convergecast delay given the distance between a sensor node and the sink node together with other networking setting parameters. The main contributions of this dissertation are summarized as follows: Optimality compatibility study between tree-based broadcast routing algorithms and routing metrics: Broadcast routing is a critical component in the routing design. While there are plenty of routing metrics and broadcast routing schemes in current literature, arbitrary combination of broadcast routing metrics with broadcast tree construction (BTC) algorithms may not result in optimal broadcast trees. In this work, we study the requirement on the combination of routing metrics and BTC algorithms to ensure optimal broadcast tree construction. When a BTC algorithm fails to find the optimal broadcast tree, we define that the BTC algorithm and the metric are not optimality compatible. We show that different BTC algorithms have different requirements on the properties of broadcast routing metrics. The metric properties for BTC algorithms in both undirected network topologies and directed network topologies are developed and proved. They are successfully used to verify the optimality compatibility between broadcast routing metrics and BTC algorithms. Optimal cache-based route repair policy for real-time traffic in mobile ad hoc networks: Real-time applications in ad hoc networks require fast route repair mechanisms to minimize the interruptions to their communications. Cache-based route repair schemes are popular choices since they can quickly resume communications using cached backup paths after a route break. In this work, through thorough theoretical modeling of the cache-based route repair process, we derive a provably optimal cache-based route repair policy. This optimal policy considers both the overhead of the route repair schemes and the promptness of the repair action. The correctness and advantages of our optimal policy are validated by extensive simulations. Optimal secondary user node placement study in cognitive radio networks: Information propagation speed (IPS) in a multi-hop CRN is an important factor that affects the network's delay performance and needs to be considered in network planning. The impact of primary user (PU) activities on IPS makes the problem of analyzing IPS in multi-hop CRNs very challenging and hence unsolved in existing literature. In this work, we fill this technical void. We establish models of IPS in multi-hop CRNs and compute how to maximize IPS in two cases. The first case, named the maximum network IPS, maximizes IPS across a network topology over an infinite plane. The second case, named the maximum flow IPS, maximizes the IPS between a given pair of source and destination nodes separated by a fixed distance. We reveal that both maximum IPSs are determined by the PU activity level and the placement of SU relay nodes. We design optimal relay placement strategies in CRNs to maximize these two IPS under different PU activity levels. The correctness of our analytical results is validated by simulations and numerical experiments. Convergecast delay analysis of large scale sensor networks coexisting with WiFi networks: Due to the increasing popularity of wireless devices, such as WiFi (IEEE 802.11) and ZigBee (IEEE 802.15.4), the ISM bands have become more and more crowded. Since ZigBee is the de facto radio technology of sensor networks, coexistence of WiFi networks and sensor (ZigBee) networks is challenging because of the great heterogeneity between WiFi and ZigBee technologies. In the presence of interference from WiFi and other sensor nodes, the performance of sensor networks is not clearly understood. In this work, we study delay performance of a large scale sensor network which coexists with WiFi networks. Given the distance from the sensor node to the sink node, we are interested in the expected delay of sensor packets to reach the sink node in the presence of both WiFi and sensor interference. We formulate the delay analysis problem as a two priority M/G/1 preemptive repeat identical queueing system, and analyze the delay using queueing theory and probability theory. First, we use a path probabilistic approach to derive the expected delay. Second, we develop a simplified linear approximation model for delay analysis. The correctness of both models is validated by NS2 simulations.Recently, many new types of wireless networks have emerged, such as mobile ad hoc networks (MANETs), cognitive radio networks (CRNs) and large scale wireless sensor networks. To get better performance in these wireless networks, various schemes, e.g., metrics, policies, algorithms, protocols, etc., have been proposed. Among them, optimal schemes that can achieve optimal performance are of great importance. On the theoretical side, they provide important design guidelines and performance benchmarks. On the practical side, they guarantee best communication performance with limited network resources. In this dissertation, we focus on the modeling and optimization of routing in wireless networks, including both broadcast routing, unicast routing, and convergecast routing. We study two aspects of routing: algorithm analysis and Qos analysis. In the algorithmic work, we focus on how to build optimal broadcast trees. We investigate the optimality compatibility between three tree-based broadcast routing algorithms and routing metrics. The Qos work includes three parts. First, we focus on how to optimally repair broken paths to minimize impact of path break in MANETs. We propose a provably optimal cached-based route repair policy for real-time traffic in MANETs. Second, we focus on the impact of secondary user (SU) node placement on SU traffic delay in CRNs. We design SU node placement schemes that can minimize the multi-hop delay in CRNs. Third, we analyze the convergecast delay of a large scale sensor network which coexists with WiFi nodes. We derive a closed form delay formula, which can be used to estimate sensor packet convergecast delay given the distance between a sensor node and the sink node together with other networking setting parameters. The main contributions of this dissertation are summarized as follows: Optimality compatibility study between tree-based broadcast routing algorithms and routing metrics: Broadcast routing is a critical component in the routing design. While there are plenty of routing metrics and broadcast routing schemes in current literature, arbitrary combination of broadcast routing metrics with broadcast tree construction (BTC) algorithms may not result in optimal broadcast trees. In this work, we study the requirement on the combination of routing metrics and BTC algorithms to ensure optimal broadcast tree construction. When a BTC algorithm fails to find the optimal broadcast tree, we define that the BTC algorithm and the metric are not optimality compatible. We show that different BTC algorithms have different requirements on the properties of broadcast routing metrics. The metric properties for BTC algorithms in both undirected network topologies and directed network topologies are developed and proved. They are successfully used to verify the optimality compatibility between broadcast routing metrics and BTC algorithms. Optimal cache-based route repair policy for real-time traffic in mobile ad hoc networks: Real-time applications in ad hoc networks require fast route repair mechanisms to minimize the interruptions to their communications. Cache-based route repair schemes are popular choices since they can quickly resume communications using cached backup paths after a route break. In this work, through thorough theoretical modeling of the cache-based route repair process, we derive a provably optimal cache-based route repair policy. This optimal policy considers both the overhead of the route repair schemes and the promptness of the repair action. The correctness and advantages of our optimal policy are validated by extensive simulations. Optimal secondary user node placement study in cognitive radio networks: Information propagation speed (IPS) in a multi-hop CRN is an important factor that affects the network's delay performance and needs to be considered in network planning. The impact of primary user (PU) activities on IPS makes the problem of analyzing IPS in multi-hop CRNs very challenging and hence unsolved in existing literature. In this work, we fill this technical void. We establish models of IPS in multi-hop CRNs and compute how to maximize IPS in two cases. The first case, named the maximum network IPS, maximizes IPS across a network topology over an infinite plane. The second case, named the maximum flow IPS, maximizes the IPS between a given pair of source and destination nodes separated by a fixed distance. We reveal that both maximum IPSs are determined by the PU activity level and the placement of SU relay nodes. We design optimal relay placement strategies in CRNs to maximize these two IPS under different PU activity levels. The correctness of our analytical results is validated by simulations and numerical experiments. Convergecast delay analysis of large scale sensor networks coexisting with WiFi networks: Due to the increasing popularity of wireless devices, such as WiFi (IEEE 802.11) and ZigBee (IEEE 802.15.4), the ISM bands have become more and more crowded. Since ZigBee is the de facto radio technology of sensor networks, coexistence of WiFi networks and sensor (ZigBee) networks is challenging because of the great heterogeneity between WiFi and ZigBee technologies. In the presence of interference from WiFi and other sensor nodes, the performance of sensor networks is not clearly understood. In this work, we study delay performance of a large scale sensor network which coexists with WiFi networks. Given the distance from the sensor node to the sink node, we are interested in the expected delay of sensor packets to reach the sink node in the presence of both WiFi and sensor interference. We formulate the delay analysis problem as a two priority M/G/1 preemptive repeat identical queueing system, and analyze the delay using queueing theory and probability theory. First, we use a path probabilistic approach to derive the expected delay. Second, we develop a simplified linear approximation model for delay analysis. The correctness of both models is validated by NS2 simulations. / Ph. D.

Ce mémoire a pour but d'explorer l'utilisation de données de contexte, notamment spatial, pour prédire de la durée que va prendre un garage pour effectuer les réparations à la suite d'un accident automobile. Le contexte réfère à l'environnement dans lequel évolue le garage. Il s'agit donc de développer une approche permettant de prédire une caractéristique précise en utilisant notamment de l'information historique. L'information historique comprend des composantes spatiales, comme des adresses, qui vont être exploitées afin de générer de nouvelles informations relatives à la localisation des garages automobiles. L'utilisation des données accumulées sur les réclamations automobiles va permettre d'établir un niveau initial de prédiction qu'il est possible d'atteindre avec de l'apprentissage supervisé. En ajoutant ensuite petit à petit de l'information de contexte spatial dans lequel évolue le garage responsable des réparations, de nouveaux niveaux de prédiction seront atteints. Il sera alors possible d'évaluer la pertinence de considérer le contexte spatial dans un problème de prédiction comme celui des temps de réparations des véhicules accidentés en comparant ces niveaux de prédiction précédemment cités. L'utilisation de données historiques pour prédire une nouvelle donnée se fait depuis plusieurs années à l'aide d'une branche de l'intelligence artificielle, à savoir : l'apprentissage machine. Couplées à cette méthode d'analyse et de production de données, des analyses spatiales vont être présentées et introduites pour essayer de modéliser le contexte spatial. Pour quantifier l'apport d'analyses spatiales et de données localisées dans un problème d'apprentissage machine, il sera question de comparer l'approche n'utilisant pas d'analyse spatiale pour produire de nouvelles données, avec une approche similaire considérant cette fois-ci les données de contexte spatial dans lequel évolue le garage. L'objectif est de voir l'impact que peut avoir une contextualisation spatiale sur la prédiction d'une variable quantitative. / The purpose of this paper is to explore the use of context data, particularly spatial context, to predict how long it will take a garage to complete repairs following an automobile accident. The context refers to the environment in which the garage evolves. It is therefore a question of developing an approach that makes it possible to predict a precise characteristic by using historical information in particular. The historical information includes spatial components, such as addresses, which will be exploited to generate new information about the location of car garages. The use of the accumulated data on car claims will allow to establish an initial level of prediction that can be reached with supervised learning. By then gradually adding information about the spatial context in which the garage responsible for the repairs evolves, new levels of prediction will be reached. It will then be possible to evaluate the relevance of considering the spatial context in a prediction problem such as that of the repair times of accidented vehicles by comparing these prediction levels previously mentioned. The use of historical data to predict new data has been done for several years with the help of a branch of artificial intelligence, namely: machine learning. Coupled with this method of data analysis and production, spatial analyses will be presented and introduced to try to model the spatial context. To quantify the contribution of spatial analysis and localized data in a machine learning problem, we will compare the approach that does not use spatial analysis to produce new data with a similar approach that considers the spatial context data in which the garage evolves. The objective is to see the impact that spatial contextualization can have on the prediction of a quantitative variable.