Title from PDF of title page (University of Missouri--Columbia, viewed on Feb. 12, 2010). The entire thesis text is included in the research.pdf file; the official abstract appears in the short.pdf file; a non-technical public abstract appears in the public.pdf file. Thesis advisor: Dr. Praveen K. Edara. Includes bibliographical references.

Nowadays, with the increasing deployment of modern packet-switching networks, traffic classification is playing an important role in network administration. To identify what kinds of traffic transmitting across networks can improve network management in various ways, such as traffic shaping, differential services, enhanced security, etc. By applying different policies to different kinds of traffic, Quality of Service (QoS) can be achieved and the granularity can be as fine as flow-level. Since illegal traffic can be identified and filtered, network security can be enhanced by employing advanced traffic classification. There are various traditional techniques for traffic classification. However, some of them cannot handle traffic generated by applications using non-registered ports or forged ports, some of them cannot deal with encrypted traffic and some techniques require too much computational resources. The newly proposed technique by other researchers, which uses statistical methods, gives an alternative approach. It requires less resources, does not rely on ports and can deal with encrypted traffic. Nevertheless, the performance of the classification using statistical methods can be further improved. In this thesis, we are aiming for optimising network traffic classification based on the statistical approach. Because of the popularity of the TCP protocol, and the difficulties for classification introduced by TCP traffic controls, our work is focusing on classifying network traffic based on TCP protocol. An architecture has been proposed for improving the classification performance, in terms of accuracy and response time. Experiments have been taken and results have been evaluated for proving the improved performance of the proposed optimised classifier. In our work, network packets are reassembled into TCP flows. Then, the statistical characteristics of flows are extracted. Finally the classes of input flows can be determined by comparing them with the profiled samples. Instead of using only one algorithm for classifying all traffic flows, our proposed system employs a series of binary classifiers, which use optimised algorithms to detect different traffic classes separately. There is a decision making mechanism for dealing with controversial results from the binary classifiers. Machining learning algorithms including k-nearest neighbour, decision trees and artificial neural networks have been taken into consideration together with a kind of non-parametric statistical algorithm — Kolmogorov-Smirnov test. Besides algorithms, some parameters are also optimised locally, such as detection windows, acceptance thresholds. This hierarchical architecture gives traffic classifier more flexibility, higher accuracy and less response time.

The purpose of this study is to develop a tool that will assist in the development of traffic control plans. Using information from TxDOT Dallas District and several other sources, a GIS-based map has been assembled that combines data from many different formats into one user-friendly environment. The capability provided in the tool includes geometric properties of all TxDOT-controlled freeway segments, as well as daily and hourly demand data and capacity information specific to both general and HOV lanes. By selecting a segment on the map and opening the tool, the user can easily navigate through the collection of data in just seconds, which will significantly reduce the amount of time required by the traffic engineer to gather site-specific information when analyzing temporary traffic control. / text

This dissertation reported on the findings of the characteristics of the pedestrian flows in Chinese New Year Eve Fair in Hong Kong and, the recommendations to its crowd control measures. Since most of the pedestrian flow models were developed for general purposes under normal condition, special models developed specifically for major events such as bazaars, fairs and festivals in Hong Kong were required to understand their pedestrian flow patterns. In this dissertation, pedestrian flows in the Fair were videotaped and data was extracted for calibrating several pedestrian flow models. These included the conventional models developed by Greenshields, Greenberg, Pipes-Munjal, Underwood, Drake, Wong et al. and four modifications of these models for simulating isotropic and bi-directional pedestrian flow scenarios. The free flow speed of the pedestrians in the Chinese New Year Eve Fair was found to be around 0.69-0.84 m/s, slower than those identified in other researches. Besides, the results of these models showed the relationships between walking speed, density and flow of the pedestrians in the Fair. Also, the effects of bi-directional flow to pedestrian flows were assessed and quantified. These findings obtained from models were then used as a basis for formulating crowd control strategy of major events in Hong Kong. / published_or_final_version / Transport Policy and Planning / Master / Master of Arts in Transport Policy and Planning

Walking is an environmentally friendly mode of transportation. A better understanding of pedestrian activities and effective planning of walking facilities are particularly important for densely populated Asian cities like Hong Kong. Empirical studies on pedestrian flows can be classified as controlled experiments and observational surveys. Controlled experiments are flexible and can be designed to fulfill the requirements of specific studies and eliminate influence from unrelated factors. Observational surveys provide data for pedestrian movements in different types of walking facilities. This thesis aims to develop a mathematical model for multidirectional pedestrian flows based on knowledge obtained from both a controlled experiment and observational surveys on three different walking facilities. Bayesian inference is adopted for model calibration, as it can combine the prior information from the controlled experiment and the observed data from the observational surveys. The proposed model is based on Drake’s (1967) model of traffic flow. However, multidirectional pedestrian flows are much more complicated than the unidirectional and bidirectional flows. Therefore, instead of relating the speed of a pedestrian stream solely to pedestrian density, the flow ratio and intersecting angles between streams are introduced as factors that may influence stream speed. The proposed model takes the form of a set of structural equations rather than a single deterministic function. By applying Bayesian inference, the proposed model is calibrated with the three sets of observed data respectively, based on the prior distribution specified by the controlled experiment results. Finally, pedestrian movements in three different walking facilities are analyzed based on the properties of the calibrated model. / published_or_final_version / Civil Engineering / Master / Master of Philosophy

Driving under the influence of alcohol (DUIA) is well recognized as one of the key human factors that increase the risks of traffic accidents and associated injuries and fatalities. These risks increase rapidly with the alcohol concentration in human body. Unfortunately, drink-driving not only increases the accident and injury risks of drivers, but also increases the risks to innocent road users. Despite the decrease in the number of alcohol-related accidents over the past decade, few comprehensive studies have investigated the relationship between the drivers’ alcohol consumption, the actual impairment in driving performance due to alcohol, and the resulting accident and injury risks. Therefore, the development of an appropriate road safety policy framework would help to formulate appropriate measures and penalties for combating drink-driving, thus reducing the associated injuries and fatalities in the long term. To better understand the relationship between drink-driving and traffic safety, this study attempts to evaluate the alcohol-impaired effect on traffic safety in a three-pronged approach: risk analysis, physiological performance, and attitudinal evaluation. For the crash risk analysis, impairment effects of alcohol on the severity of injuries and road crashes in Hong Kong were determined using the comprehensive random breath test (RBT) data. For the physiological performance, simulation study was conducted to develop a temporal profile of driver alcohol concentration and to determine the relationship between driver alcohol concentration and the actual alcohol-impaired driving performance for Chinese populations. Finally, for the attitudinal evaluation, a perception survey was used to evaluate the effectiveness of legislation on drivers’ intention to reduce drink-driving. Findings of the three-pronged analysis were then integrated to provide a useful tool for formulating effective legislation and enforcement measures to combat drink-driving, with respect to the driver’s perceptions, observed driving performance, and actual road safety levels in the long term. / published_or_final_version / Civil Engineering / Doctoral / Doctor of Philosophy

Autonomous vehicles (AVs) may soon be publicly available and are expected to increase both network capacity and travel demand. Reduced safety margins from computer precision may increase network capacity and allow for more efficient intersection controls. AVs also offer the option of repositioning trips to avoid parking fees or share the vehicle between household members, which may increase the total number of vehicle trips and decrease the relative utility of transit. Since AVs may be available within one or two decades, which is within the span of long-term planning models, practitioners may soon wish to predict the effects of AVs on traffic networks. This thesis modifies the four-step planning model commonly used by practitioners to include AV behaviors and capacity improvements. Because dynamic traffic assignment (DTA) offers more realistic flow propagation and intersection control options, the four-step model is modified to incorporate DTA with endogenous departure time choices. To facilitate modeling of AV intersections, the tile-based reservation (TBR) control policy is simplified into a conflict region (CR) model compatible with general simulation-based DTA and with greatly improved computational tractability. Results suggest that although the total number of personal-vehicle trips may almost double (due to repositioning trips to the origin to avoid parking costs), increases in network and intersection capacity can mostly offset or even improve network conditions. Use of dynamic flow propagation instead of static travel time functions in the four-step model results in predictions of increased average travel speed although both static and dynamic planning models predict a high reliance on repositioning trips (i.e., empty-vehicle travel). To study AV behaviors in DTA, this thesis first integrates DTA into the four-step model with the addition of departure time choice. This model alone may be useful for practitioners as departure time modeling is a major concern with DTA planning models. Also, the TBR intersection policy has only been studied in micro-simulation with heuristic routing strategies. The CR model opens this new technique to study under UE behavior, which is the first step for the bridge between technology demonstration simulations to models practitioners can use to evaluate implementation. . Therefore, the models developed here for the purposes of predicting AV trip and mode choices may themselves become useful tools for other applications. / text

M. Tech. Policing / This study was undertaken due to indications that the comprehensive, integrated and scientifically accountable data were largely lacking that were needed to counter road traffic crashes in South Africa. The aim of the study was to explore whether there was a need for developing a model that would provide pointers for establishing and maintaining a scientifically accountable National Road Traffic Safety Databank. In short, the specific objectives were: To assess the characteristics of data and data-gathering systems on road traffic crashes in South Africa, needs in this respect, and the contribution the Council for Scientific and Industrial Research could make towards the establishment and maintenance of such a National Road Traffic Safety Databank.

No description available.

In this thesis, we investigate the impact of channels that have a bursty nature in a cognitive radio network scenario. Our goal is to design a general channel usage model that can handle bursty primary user channel usage. The proposed model describes idle periods with a discrete platoon arrival process and describes busy periods with a discrete phase type distribution. The performance of the proposed model is compared with two more traditionally encountered channel usage models in three different secondary user access schemes. First, we design a reactive access scheme to show the poor performance results an in- vestigator can potentially obtain when ignoring bursty data traffic. We have also analyzed two proactive secondary network access schemes. Numerical results show that the achiev- able utilization and interference probability of the network are affected when traditional channel models are used in a bursty PU channel.