Orientador: Nelson Luis Sadanha da Fonseca / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Computação / Made available in DSpace on 2018-08-06T04:44:40Z (GMT). No. of bitstreams: 1 Ostrowsky_LarissaOliveira_M.pdf: 1550430 bytes, checksum: 823c0f45a69f06a55a90d4abbffa7e24 (MD5) Previous issue date: 2005 / Resumo: Uma característica importante do tráfego gerado pelo protocolo Internet Protocol (IP) é a existência de padrões scaling, que impactam significantemente o desempenho dos mecanismos de controle de tráfego e que, por isto vem sendo foco de atenção de diversas pesquisas. A natureza scaling do tráfego IP tem sido alvo de bastante polêmica. Em pequenas escalas de tempo o tráfego IP é altamente variável e a variabilidade difere da natureza fractal encontrada em grandes escalas de tempo, não existindo ainda um consenso em relação a natureza do tráfego IP nestas pequenas escalas de tempo. No presente estudo, foram revisadas as evidências da multifractalidade nas pequenas escalas de tempo, através da análise experimental de diversos traços de tráfego real. Constatou-se que não se pode generalizar a natureza dos fluxos do protocolo IP e do protocolo Transmission Control Protocol (TCP) como monofractal ou como multifractal, enquanto que a natureza do fluxo do protocolo User Datagram Protocol (UDP) é sempre multifractal. O crescente uso do protocolo UDP pelas emergentes aplicações que necessitam requisitos de tempo real altera consideravelmente a natureza scaling do tráfego IP, dado que este tipo de tráfego n¿ao reage a situações de congestionamento. Apesar de existirem diversos modelos para tráfego TCP, pouca atenção tem sido dada a modelagem de tráfego UDP. Outra contribuição desta dissertação é a proposta de um modelo de tráfego para fluxos UDP, desenvolvido a partir da caracterização das distribuições estatísticas de traços reais. Este modelo consegue reproduzir com precisão várias características do tráfego UDP real, inclusive a natureza scaling. O modelo foi validado via simulação e a precisão dos resultados do modelo para avaliar um sistema de filas foi comparada às precisões do Modelo Wavelet Multifractal (MWM) e do Modelo MAP Multifractal. Resultados indicam que o modelo proposto reproduz melhor a ocupação em uma fila para diferentes capacidades de armazenamento e taxas de serviço do que os outros modelos avaliados. A geração de traços sintéticos a partir do modelo adotado pode ser realizada em pequenos intervalos de tempo. Assim, o modelo proposto é adequado para o dimensionamento de recursos da rede e provisionamento de Qualidade de serviço / Abstract: An important characteristic of the traffic generated by IP protocol is the existence of scaling, since it has great impact on the performance of traffic control mechanism. Therefore, it has been the focus of attention in many researches. The scaling nature of IP traffic has generated lots of controversy. At small time scales the traffic is very irregular and the variability is different from that found in fractal nature at larges scales. There is no general agreement in relation to the nature of the traffic at these small time scales. In the present study, the evidences of multifractality at small time scales were revised via experimental analyses of several real traffic traces. It was concluded that is not possible to generalize that the nature of IP and TCP flows is either monofractal or multifractal, while the nature of the UDP flows is always multifractal. The increasing use of the UDP protocol by real time applications is changing substantially the scaling nature of IP traffic, since this type of traffic does not react to congestion situations. Although there are models for TCP traffic, not much attention has been given to the modeling of UDP flows. Another contribution of this dissertation is a proposition of UDP traffic model, developed from the characterization of the statistics distributions of real traces. This model can reproduce with precision several characteristics of UDP real traffic, including the scaling nature. The model was validated via simulation and the precision of results of the model was compared with the precisions of the MWM and MAP models. Results indicate that the proposed model better reproduces the queue occupation for different storage capacities and service rates then the others evaluated models. The generation of synthetic traces using the adopted model can be realized with low execution times. Thus, the proposed model is acceptable to the dimension of network resources measurement and to provide Quality of Service / Mestrado / Redes e Multimidia / Mestre em Ciência da Computação

Within Ericsson there is a continuous activity of traffic modeling. Traffic modeling is a practice to analyze traffic patterns and determine necessary resources to handle it optimally. This activity focuses on gathering and analyzing live network measurements, implementing and presenting traffic models. One example of concept in packet traffic modeling is transmission object log which is an aggregation of packet data traces from a measured network over a transmission period. These trace logs that are simple list of all transmission objects contain a vast number of data. When the amount of data increases in these logs several problems can occur such as expensive analysis time, costly data storage and even statistical analysis and data processing in software environments run out of memory. On the other hand, sophisticated and costly computing systems are required for analysis and storage of the data. Therefore, monitoring and analyzing these large traces motivate data reduction. The goal of this thesis is to reduce the number of traffic objects in large object trace logs while preserving the statistical characteristics of the original transmission objects.    Sampling techniques are wildly used to cope with the issues of large amount of data in network monitoring. First, this thesis aims to assess the impact of two sampling techniques as a reduction method. Second, to analyze traffic characteristics and showing the effects of sampling, some statistical properties of both original and sampled datasets as well as their distribution plots will be discussed. The distortion introduced by sampling as the distance between the distribution of properties for sampled and unsampled traffic is also presented by a statistical metric. One of the issues in sampling technique is the sampling size. In order to estimate the sampling size and reduce the logs to a certain level, the concept of offline marginal utility as a complementary method to sampling is proposed in this report. The thesis also makes some suggestions as further works to reduce the logs by having less impact on the object characteristics.

The main goals of this thesis are towards an understanding of the delay process in best-effort Internet for both non-congested and congested networks. A novel measurement system is reported for delay measurements in IP routers, which follows specifications of the IETF RFC 2679. The system employs both passive measurements and active probing and offers the possibility to measure and analyze different delay components of a router, e.g., packet processing delay, packet transmission time and queueing delay at the output link. Dedicated application-layer software is used to generate UDP traffic with TCP-like characteristics. Pareto traffic models are used to generate self-similar traffic in the link. The reported results are in form of several important statistics regarding processing and queueing delays of a router, router delay for a single data flow, router delay for multiple data flows as well as end-to-end delay for a chain of routers. They confirm results reported earlier about the fact that the delay in IP routers is generally influenced by traffic characteristics, link conditions and, to some extent, details in hardware implementation and different IOS releases. The delay in IP routers may also occasionally show extreme values, which are due to improper functioning of the routers. Furthermore, new results have been obtained that indicate that the delay in IP routers shows heavy-tailed characteristics, which can be well modeled with the help of several distributions, either in the form of a single distribution or as a mixture of distributions. There are several components contributing to the OWTT in routers, i.e., processing delay, queueing delay and service time. The obtained results have shown that, e.g., the processing delay in a router can be well modeled with the Normal distribution, and the queueing delay is well modeled with a mixture of Normal distribution for the body probability mass and Weibull distribution for the tail probability mass. Furthermore, OWTT has several component delays and it has been observed that the component delay distribution that is most dominant and heavy-tailed has a decisive influence on OWTT. / Mätningar och modeller för en-vägskorsningstid presenteras.

Theses deals with network traffic modeling focused on elaboration by time series analysis. The nature of network traffic is discussed above all http traffic. First three chapters are theoretical, which describes time series and basic models, linear AR, MA, ARMA, ARIMA and nonlinear ARCH. Other chapters define terms like self-similarity and long range dependence. It is demonstrated a failure of conventional models which cannot capture these specific properties of network data traffic. On the basis of study in chapter 6. is closely described the combined ARIMA/GARCH model and its parameter estimation procedure. Applied part of this theses deals with procedure of estimation and fitting the estimation model to observed network traffic. After an estimation a few future values are predicted on the basis of estimated model. These predicted values are consequently compared with real data.

The driver-specific data from a naturalistic driving study provides car-following events in real-world driving situations, while additionally providing a wealth of information about the participating drivers. Reducing a naturalistic database into finite car-following events requires significant data reduction, validation, and calibration, often using manual procedures. The data collection performed herein included: the identification of commuting routes used by multiple drivers, the extraction of data along those routes, the identification of potential car-following events from the dataset, the visual validation of each car-following event, and the extraction of pertinent information from the database during each event identified. This thesis applies the developed process to generate car-following events from the 100-Car Study database, and applies the dataset to analyze four car-following models. The Gipps model was found to perform best for drivers with greater amounts of data in congested driving conditions, while the Rakha-Pasumarthy-Adjerid (RPA) model was best for drivers in uncongested conditions. The Gipps model was found to generate the lowest error value in aggregate, with the RPA model error 21 percent greater, and the Gaxis-Herman-Rothery model (GHR) and the Intelligent Driver Model (IDM) errors 143 percent and 86 percent greater, respectively. Additionally, the RPA model provides the flexibility for a driver to change vehicles without the need to recalibrate parameter values for that driver, and can also capture changes in roadway surface type and condition. With the error values close between the RPA and Gipps models, the additional advantages of the RPA model make it the recommended choice for simulation. / Master of Science

Even though first order LWR models have many limitations, they are still widely used in many engineering applications. Second-order models, on the other hand, address many of those limitations. Among second-order models, the inhomogeneous Aw-Rascle-Zhang (ARZ) model is well-received as its structure generates characteristic waves that make physical sense. The ARZ model --- and other $2\times 2$ hyperbolic systems with a relaxation term --- possess a critical phase transition: whenever the sub-characteristic condition (SCC) is violated, uniform traffic flow is unstable, and small perturbations grow into nonlinear traveling waves, called jamitons. The case where the SCC is satisfied has been studied extensively. However, what is essentially unstudied is the question: which jamiton solutions are dynamically stable? To understand which stop-and-go traffic waves can arise through the dynamics of the model, this question is critical. This dissertation first outlines the mathematical foundations of the ARZ model and its solutions, then presents a computational study demonstrating which types of jamitons are dynamically stable, and which are not. After that, a procedure is presented that characterizes the stability of jamitons. The study reveals that a critical component of this analysis is the proper treatment of the perturbations to the shocks, and of the neighborhood of the sonic points. The insight gained from answering the question regarding the dynamical stability of jamitons has many applications. One particular application presented here is deriving an averaged model for the ARZ model. Such a model is as simple to solve (analytically and numerically) as the LWR model, but nevertheless captures the cumulative effects of jamitons regarding fuel consumption, total flow, and braking events. / Mathematics

This dissertation addresses Internet background traffic generation and network intrusion detection. It is organized in two parts. Part one introduces a method to model realistic Internet background traffic and demonstrates how the models are used both in a simulation environment and in a lab environment. Part two introduces two different NID (Network Intrusion Detection) techniques and evaluates them using the modeled background traffic. To demonstrate the approach we modeled five major application layer protocols: HTTP, FTP, SSH, SMTP and POP3. The model of each protocol includes an empirical probability distribution plus estimates of application-specific parameters. Due to the complexity of the traffic, hybrid distributions (called mixture distributions) were sometimes required. The traffic models are demonstrated in two environments: NS-2 (a simulator) and HONEST (a lab environment). The simulation results are compared against the original captured data sets. Users of HONEST have the option of adding network attacks to the background. The dissertation also introduces two new template-based techniques for network intrusion detection. One is based on a template of autocorrelations of the investigated traffic, while the other uses a template of correlation integrals. Detection experiments have been performed on real traffic and attacks; the results show that the two techniques can achieve high detection probability and low false alarm in certain instances.

The dissertation develops analytical models that estimate the capacity of freeway weaving sections. The analytical models are developed using simulated data that were compiled using the INTEGRATION software. Consequently, the first step of the research effort is to validate the INTEGRATION lane-changing modeling procedures and the capacity estimates that are derived from the model against field observations. The INTEGRATION software is validated against field data gathered by the University of California at Berkeley by comparing the lateral and longitudinal distribution of simulated and field observed traffic volumes categorized by O-D pair on nine weaving sections in the Los Angeles area. The results demonstrate a high degree of consistency between simulated and field observed traffic volumes within the various weaving sections. Subsequently, the second validation effort compares the capacity estimates of the INTEGRATION software to field observations from four weaving sections operating at capacity on the Queen Elizabeth Way (QEW) in Toronto, Canada. Again, the results demonstrate that the capacity estimates of the INTEGRATION software are consistent with the field observations both in terms of absolute values and temporal variability across different days. The error was found to be in the range of 10% between simulated and field observed capacities. Prior to developing the analytical models, the dissertation presents a systematic analysis of the factors that impact the capacity of freeway weaving sections, which were found to include the length of the weaving section, the weaving ratio (a new parameter that is developed as part of this research effort), the percentage of heavy vehicles, and the speed limit differential between freeway and on- and off-ramps. The study demonstrates that the weaving ratio, which is currently defined as the ratio of the lowest weaving volume to the total weaving volume in the 2000 Highway Capacity Manual, has a significant impact on the capacity of weaving sections. The study also demonstrates that the weaving ratio is an asymmetric function and thus should reflect the source of the weaving volume. Consequently, a new definition for the weaving ratio is introduced that explicitly identifies the source of the weaving volume. In addition, the study demonstrates that the length of the weaving section has a larger impact on the capacity of weaving sections for short lengths and high traffic demands. Furthermore, the study demonstrates that there does not exist enough evidence to conclude that the speed limit differential between mainline freeway and on- and off-ramps has a significant impact on weaving section capacities. Finally, the study demonstrates that the HCM procedures model the heavy duty vehicle impacts reasonably well. This dissertation presents the development of new capacity models for freeway weaving sections. In these models, a new definition of the weaving ratio that explicitly accounts for the source of weaving volume is introduced. The proposed analytical models estimate the capacity of weaving sections to within 12% of the simulated data, while the HCM procedures exhibit errors in the range of 114%. Among the newly developed models, the Artificial Neural Network (ANN) models performs slightly better that the statistical models in terms of model prediction errors. However, the sensitivity analysis results demonstrate unrealistic behavior of the ANN models under certain conditions. Consequently, the use of a statistical model is recommended because it provides a high level of accuracy while providing accurate model responses to changes in model input parameters (good response to the gradient of the input parameters). / Ph. D.

Orientador: Varese Salvador Timóteo / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Tecnologia / Made available in DSpace on 2018-08-24T04:02:34Z (GMT). No. of bitstreams: 1 Diorio_RafaelFernando_M.pdf: 6050961 bytes, checksum: bb3814ee43754ae271ad5e4eb3d704b7 (MD5) Previous issue date: 2013 / Resumo: A presente dissertação contempla a implementação e validação funcional de um gateway multimídia baseado em IP a partir da implementação de recursos para classificação, rotulação, identificação e encaminhamento de fluxos de tráfego distintos (elástico e stream), ambos com modelagem de tráfego pré-definida, a partir de informações constantes no cabeçalho dos pacotes IP (campos ToS do IPv4 e TC do IPv6, tidos como DSCP [RFC 2474] e ECN [RFC 3168]) para os serviços multimídia de áudio, voz, vídeo e dados fornecidos por um emulador de tráfego multimídia baseado em IP. De modo complementar a tais implementações, o gateway multimídia também objetiva o fornecimento de recursos de qualidade de serviço (QoS) aos serviços por ele intermediados e que necessitem de tais recursos, tais como serviços multimídia de voz e vídeo, por exemplo. Os testes funcionais e de desempenho com o gateway multimídia, realizados em ambiente cabeado e sem fio, demonstram que sua implementação é válida e que não influência em modificações quanto ao perfil (características) dos tráfegos por ele intermediados, bem como à questões envolvendo erros, atrasos e variações de atrasos (jitter), dentre outras, demonstrando que sua implementação é transparente aos sistemas finais envolvidos no processo de comunicação intermediado por tal gateway multimídia. Seu ambiente de desenvolvimento e testes é baseado em soluções de software livre, fornecido, especialmente, pelo sistema operacional Linux e pela solução Netfilter/Iptables, disponibilizando uma plataforma aberta para o estudo e aprimoramento futuro quanto ao gateway multimídia. Por fim, espera-se que o gateway multimídia contribua como uma nova ferramenta para testes e análises de soluções voltadas às redes multimídia de modo geral / Abstract: This dissertation describes the implementation and functional validation of an IP-based multimedia gateway to the identification, classification and forwarding of different traffic flows (elastic and stream), both with pre-defined traffic profiles, from the ToS field of the IPv4 header and the TC field of the IPv6 header, taken as DSCP [RFC 2474] and ECN [RFC 3168]. The multimedia traffic corresponding to audio, voice, video and data services are provided by a multimedia traffic generator, which generates traffic according to given distributions. The multimedia gateway also aims at providing resources for quality of service (QoS) for services that require these resources, such as multimedia services for voice and video. The functional and performance testing with multimedia gateway, performed in a wired and wireless environment, demonstrates that its implementation is valid and not influence in changes in the traffic profile (characteristics) of the traffic intermediated by him, as well as issues involving errors, delays and variations in delay (jitter), among others, showing that its implementation is transparent to the end systems involved in the communication process intermediated by such multimedia gateway. It's development and testing environment is based on open source solutions, based on the Linux operating system and the Netfilter/Iptables solution, providing an open platform for future research and improvement of the multimedia gateway. Finally, it is expected that the multimedia gateway contributes as a new tool for testing and analysis of multimedia networks in general / Mestrado / Tecnologia e Inovação / Mestre em Tecnologia

This study presents a model that explains the traffic fatality by exploring the Poisson regression model using two types of explanatory variables – referred to as internal and external factors. Internal factors contain variables closely linked to traffic safety, such as speed limits and belt usage (Strandroth et al., 2012), whereas external factors comprise a set of variables that the Swedish Transport Administration cannot control, such as the economy and demographic change (Wiklund et al., 2012). The purpose of the study is to evaluate the impact that internal and external factors have on the traffic fatality. This is done by modeling the traffic fatality using internal factors and then assessing the contribution of adding external factors in the regression model with a forward variable selection strategy. This study uses Swedish traffic fatality data as monthly statistics. The main characteristics of the data are that fatalities have generally decreased with time. Also, the data is characterized by a long term cyclical pattern as well as a yearly cyclical pattern. For the purpose of modeling the impact of internal factors, a model inspired by Brüde (1995) has been adopted, using the variable time as the only explanatory variable. It is concluded that internal factors can be used to significantly explain the general trend of the development of traffic fatalities. The variables chosen to represent external factors were economic development, traffic exposure, demographic development and seasonal trend. The study concludes that the variables economic development, traffic exposure and demographic development significantly contribute to explain the long term cyclical trends, indicating that traffic fatality is a complex multivariate system where no single variable can solely explain its dynamics. The external factor seasonal trend has the most impact of the examined external factors and explains the yearly cyclical pattern by itself. The model presented in this study shows high explanatory power and overall good fit to fatality data, making it a promising tool for statistical analysis of factors contributing to fatality. Especially for the Swedish Transport Administration, the impact of external factors can be evaluated statistically. This study leaves room for further research to assess the impact of additional external factors as well as evaluating the model’s predictive power, both of interest to the Swedish Transport Administration. / Denna studie presenterar en modell som förklarar dödsfall i trafiken genom tillämpning av Poissonregression där två typer av förklaringsvariabler använts – interna och externa faktorer. Interna faktorer innefattar variabler som är direkt knutna till trafiksäkerhet, såsom hastighetsbegränsningar och användande av säkerhetsbälte (Strandroth et al., 2012). Externa faktorer är variabler som Trafikverket inte kan kontrollera, såsom landets ekonomi och demografiska förändringar (Wiklund et al., 2012). Syftet med denna studie är att evaluera påverkan av interna och externa faktorer på dödsfall i vägtrafik. Detta görs genom att analysera hur väl interna faktorer förklarar dödsfall i vägtrafik och sedan undersöka förbättringen av att införa externa faktorer som förklaringsvariabler genom användande av en forward variable selection-strategi. Denna studie använder månatlig data över dödsfall i svensk trafik. Dessa data karaktäriseras av en nedgående trend. Dynamiken av dödsfall visar på ett långt cykliskt mönster samt ett kortare, årligt mönster. I syfte att modellera påverkan av interna faktorer har en modell inspirerad av Brüde (1995) tillämpats. Denna modell använder enbart variabeln tid som förklaringsvariabel. Studien konstaterar att interna faktorer kan användas för att signifikant beskriva en generell trend för utvecklingen av dödsfall i vägtrafik. Variablerna som har valts att representera externa faktorer är ekonomisk utveckling, trafikarbete, demografi samt en säsongstrend. Studien konstaterar att variablerna ekonomisk utveckling, trafikarbete och demografi beskriver det långa cykliska mönstret, vilket tyder på att dödsfall i vägtrafik är av komplex natur och kan inte beskrivas av en ensam variabel. Den externa faktorn säsongstrend förbättrar modellen mest av de externa faktorerna och kan ensam förklara det kortsiktiga cykliska mönstret. Den modell som presenteras i denna studie har hög förklaringsgrad och en överlag bra modellanpassning, vilket gör den till ett lovande verktyg för statistisk analys av faktorer bidragande till dödsfall i trafiken. Modellen är av särskilt intresse för Trafikverket då den tillåter statistisk utvärdering av externa faktorers påverkan. Denna studie lämnar utrymme för framtida forskning att utvärdera påverkan av ytterligare externa faktorer samt att evaluera modellens förmåga att prognostisera framtida antal dödsfall i vägtrafik, vilka båda är intresseområden för Trafikverket.