Maps of chosen areas in Chicago, San José, London and Shibuya, areimported from Open Street Map into matlab in order to run LTE networksimulations for various scenarios. Firstly, two path loss models arecompared, the empirically based WINNER model and a set of sitespecificmodel. Secondly, low load network simulations are runseparately at two different carrier frequencies, 700MHz and 2GHz, forcity specific base station deployments. Simulation results show thatuser performance is quite unique for each city and that deploymentstrategies and city environments are strongly influencing path gain,SINR and throughput. In general, user performance in UL issignificantly worse at 2GHz than at 700MHz, whereas DL performance isnot as affected by the change in carrier frequency.

No description available.

This thesis research is involved with the development of new methodologies for enhancing the experimental use of computer simulations to optimize predicted human performance in a work domain. Using a computer simulation called Computer modeling Of Human Operator System Tasks (CoHOST) to test the concepts in this research, methods are developed that are used to establish confidence limits and significance thresholds by having the computer model self report its limits. These methods, along with experimental designs that are tailored to the use of computer simulation instead of human subject based research, are used in the CoHOST simulation to investigate the U.S. Army battalion level command and control work domain during combat conditions and develop recommendations about that domain based on the experimental use of CoHOST with these methodologies. Further, with the realization that analytical results showing strictly numerical data do not always satisfy the need for understanding by those who could most benefit from the analysis, the results are further interpreted in accordance with a team performance model and the CoHOST analysis results are mapped to it according to macroergonomic and team performance concepts. The CoHOST computer simulation models were developed based on Army needs stemming from the Persian Gulf war. They examined human mental and physical performance capabilities resulting from the introduction of a new command and control vehicle with modernized digital communications systems. Literature searches and background investigations were conducted, and the CoHOST model architecture was developed that was based on a taxonomy of human performance. A computer simulation design was implemented with these taxonomic based descriptors of human performance in the military command and control domain using the commercial programming language MicroSaint™. The original CoHOST development project developed results that suggested that automation alone does not necessarily improve human performance. The CoHOST models were developed to answer questions about whether human operators could operate effectively in a specified work domain. From an analytical point of view this satisfied queries being made from the developers of that work domain. However, with these completed models available, the intriguing possibility now exists to allow an investigation of how to optimize that work domain to maximize predicted human performance. By developing an appropriate experimental design that allows evaluative conditions to be placed on the simulated human operators in the computer model rather than live human test subjects, a series of computer runs are made to establish test points for identified dependent variables against specified independent variables. With these test points a set of polynomial regression equations are developed that describe the performance characteristics according to these dependent variables of the human operator in the work domain simulated in the model. The resulting regression equations are capable of predicting any outcome the model can produce. The optimum values for the independent variables are then determined that produce the maximum predicted human performance according to the dependent variables. The conclusions from the CoHOST example in this thesis complement the results of the original CoHOST study with the prediction that the primary attentional focus of the battalion commander during combat operations is on establishing and maintaining an awareness and understanding of the situational picture of the battlefield he is operating upon. Being able to form and sustain an accurate mental model of this domain is the predicted predominant activity and drives his ability to make effective decisions and communicate those decisions to the other members of his team and to elements outside his team. The potential specific benefit of this research to the Army is twofold. First, the research demonstrates techniques and procedures that can be used without any required modifications to the existing computer simulations that allow significant predictive use to be made of the simulation beyond its original purpose and intent. Second, the use of these techniques with CoHOST is developing conclusions and recommendations from that simulation that Army force developers can use with their continuing efforts to improve and enhance the ability of commanders and other decision makers to perform as new digital communications systems and procedures are producing radical changes to the paradigm that describes the command and control work domain. The general benefits beyond the Army domain of this research fall into the two areas of methodological improvement of simulation based experimental procedures and in the actual application area of the CoHOST simulation. Tailoring the experimental controls and development of interrogation techniques for the self-reporting and analysis of simulation parameters and thresholds are topics that bode for future study. The CoHOST simulation, while used in this thesis as an example of new and tailored techniques for computer simulation based research, has nevertheless produced conclusions that deviate somewhat from prevailing thought in military command and control. Refinement of this simulation and its use in an even more thorough simulation based study could further address whether the military decision making process itself or contributing factors such as development of mental models for understanding of the situation is or should be the primary focus of team decision makers in the military command and control domain. / Master of Science

In the pursuit for ever-increasing compute power, and with Moore's law slowly coming to an end, high-performance computing started to scale-out to larger systems. Alongside the increasing system size, the interconnection network is growing to accommodate and connect tens of thousands of compute nodes. These networks have a large influence on total cost, application performance, energy consumption, and overall system efficiency of the supercomputer. Unfortunately, state-of-the-art routing algorithms, which define the packet paths through the network, do not utilize this important resource efficiently. Topology-aware routing algorithms become increasingly inapplicable, due to irregular topologies, which either are irregular by design, or most often a result of hardware failures. Exchanging faulty network components potentially requires whole system downtime further increasing the cost of the failure. This management approach becomes more and more impractical due to the scale of today's networks and the accompanying steady decrease of the mean time between failures. Alternative methods of operating and maintaining these high-performance interconnects, both in terms of hardware- and software-management, are necessary to mitigate negative effects experienced by scientific applications executed on the supercomputer. However, existing topology-agnostic routing algorithms either suffer from poor load balancing or are not bounded in the number of virtual channels needed to resolve deadlocks in the routing tables. Using the fail-in-place strategy, a well-established method for storage systems to repair only critical component failures, is a feasible solution for current and future HPC interconnects as well as other large-scale installations such as data center networks. Although, an appropriate combination of topology and routing algorithm is required to minimize the throughput degradation for the entire system. This thesis contributes a network simulation toolchain to facilitate the process of finding a suitable combination, either during system design or while it is in operation. On top of this foundation, a key contribution is a novel scheduling-aware routing, which reduces fault-induced throughput degradation while improving overall network utilization. The scheduling-aware routing performs frequent property preserving routing updates to optimize the path balancing for simultaneously running batch jobs. The increased deployment of lossless interconnection networks, in conjunction with fail-in-place modes of operation and topology-agnostic, scheduling-aware routing algorithms, necessitates new solutions to solve the routing-deadlock problem. Therefore, this thesis further advances the state-of-the-art by introducing a novel concept of routing on the channel dependency graph, which allows the design of an universally applicable destination-based routing capable of optimizing the path balancing without exceeding a given number of virtual channels, which are a common hardware limitation. This disruptive innovation enables implicit deadlock-avoidance during path calculation, instead of solving both problems separately as all previous solutions.

[EN] Intelligent Transportation Systems (ITS) can make transportation safer, more efficient, and more sustainable by applying various information and communication technologies. One of these technologies are \acfp{VN}. \acp{VN} combine different communication solutions such as cellular networks, \acfp{VANET}, or IEEE 802.11 technologies to provide connectivity among vehicles, and between vehicles and road infrastructure. This thesis focuses on VNs, and considers that the high speed of the nodes and the presence of obstacles like buildings, produces a highly variable network topology, as well as more frequent partitions in the network. Therefore, classical \ac{MANET} protocols do not adapt well to VANETs. Under these conditions, \ac{DTN} have been proposed as an alternative able to cope with these adverse characteristics. In DTN, when a message cannot be routed to its destination, it is not immediately dropped but it is instead stored and carried until a new route becomes available. The combination of VN and DTN is called \acp{VDTN}. In this thesis, we propose a new VDTN protocol designed to collect information from vehicular sensors. Our proposal, called \ac{MSDP}, combines information about the localization obtained from a GNSS system with the actual street/road layout obtained from a Navigation System (NS) to define a new routing metric. Both analytical and simulation results prove that MSDP outperforms previous proposals. Concerning the deployment of VNs and VANET technologies, technology already left behind the innovation and the standardization phases, and it is about time it reach the first early adopters in the market. However, most car manufacturers have decided to implement VN devices in the form of On Board Units (OBUs), which are expensive, heavily manufacturer dependent, and difficult to upgrade. These facts are delaying the deployment of VN. To boost this process, we have developed the GRCBox architecture. This architecture is based on low-cost devices and enables the establishment of V2X, \emph{i.e.} V2I and V2V, communications while integrating users by easing the use of general purpose devices like smartphones, tablets or laptops. To demonstrate the viability of the GRCBox architecture, we combined it with a DTN platform called Scampi to obtain actual results over a real VDTN scenario. We also present several GRCBox-aware applications that illustrate how developers can create applications that bring the potential of VN to user devices. / [ES] Los sistemas de transporte inteligente (ITS) son el soporte para el establecimiento de un transporte más seguro, más eficiente y más sostenible mediante el uso de tecnologías de la información y las comunicaciones. Una de estas tecnologías son las redes vehiculares (VNs). Las VNs combinan diferentes tecnologías de comunicación como las redes celulares, las redes ad-hoc vehiculares (VANETs) o las redes 802.11p para proporcionar conectividad entre vehículos, y entre vehículos y la infraestructura de carreteras. Esta tesis se centra en las VNs, en las cuales la alta velocidad de los nodos y la presencia de obstáculos como edificios producen una topología de red altamente variable, así como frecuentes particiones en la red. Debido a estas características, los protocolos para redes móviles ad-hoc (MANETs) no se adaptan bien a las VANETs. En estas condiciones, las redes tolerantes a retardos (DTNs) se han propuesto como una alternativa capaz de hacer frente a estos problemas. En DTN, cuando un mensaje no puede ser encaminado hacia su destino, no es inmediatamente descartado sino es almacenado hasta que una nueva ruta esta disponible. Cuando las VNs y las DTNs se combinan surgen las redes vehiculares tolerantes a retardos (VDTN). En esta tesis proponemos un nuevo protocolo para VDTNs diseñado para recolectar la información generada por sensores vehiculares. Nuestra propuesta, llamada MSDP, combina la información obtenida del servicio de información geográfica (GIS) con el mapa real de las calles obtenido del sistema de navegación (NS) para definir una nueva métrica de encaminamiento. Resultados analíticos y mediante simulaciones prueban que MSDP mejora el rendimiento de propuestas anteriores. En relación con el despliegue de las VNs y las tecnologías VANET, la tecnología ha dejado atrás las fases de innovación y estandarización, ahora es el momento de alcanzar a los primeros usuarios del mercado. Sin embargo, la mayoría de fabricantes han decidido implementar los dispositivos para VN como unidades de a bordo (OBU), las cuales son caras y difíciles de actualizar. Además, las OBUs son muy dependientes del fabricante original. Todo esto esta retrasando el despliegue de las VNs. Para acelerar la adopción de las VNs, hemos desarrollado la arquitectura GRCBox. La arquitectura GRCBox esta basada en un dispositivo de bajo coste que permite a los usuarios usar comunicaciones V2X (V2V y V2I) mientras utilizan dispositivos de propósito general como teléfonos inteligentes, tabletas o portátiles. Las pruebas incluidas en esta tesis demuestran la viabilidad de la arquitectura GRCBox. Mediante la combinación de nuestra GRCBox y una plataforma de DTN llamada Scampi hemos diseñado y probado un escenario VDTN real. También presentamos como los desarrolladores pueden crear nuevas aplicaciones GRCBox para llevar el potencial de las VN a los dispositivos de usuario. / [CA] Els sistemes de transport intel·ligent (ITS) poden crear un transport més segur, més eficient i més sostenible mitjançant l'ús de tecnologies de la informació i les comunicacions aplicades al transport. Una d'aquestes tecnologies són les xarxes vehiculars (VN). Les VN combinen diferents tecnologies de comunicació, com ara les xarxes cel·lulars, les xarxes ad-hoc vehiculars (VANET) o les xarxes 802.11p, per a proporcionar comunicació entre vehicles, i entre vehicles i la infraestructura de carreteres. Aquesta tesi se centra en les VANET, en les quals l'alta velocitat dels nodes i la presència d'obstacles, com els edificis, produeixen una topologia de xarxa altament variable, i també freqüents particions en la xarxa. Per aquest motiu, els protocols per a xarxes mòbils ad-hoc (MANET) no s'adapten bé. En aquestes condicions, les xarxes tolerants a retards (DTN) s'han proposat com una alternativa capaç de fer front a aquests problemes. En DTN, quan un missatge no pot ser encaminat cap a la seua destinació, no és immediatament descartat sinó que és emmagatzemat fins que apareix una ruta nova. Quan les VN i les DTN es combinen sorgeixen les xarxes vehicular tolerants a retards (VDTN). En aquesta tesi proposem un nou protocol per a VDTN dissenyat per a recol·lectar la informació generada per sensors vehiculars. La nostra proposta, anomenada MSDP, combina la informació obtinguda del servei d'informació geogràfica (GIS) amb el mapa real dels carrers obtingut del sistema de navegació (NS) per a definir una nova mètrica d'encaminament. Resultats analítics i mitjançant simulacions proven que MSDP millora el rendiment de propostes prèvies. En relació amb el desplegament de les VN i les tecnologies VANET, la tecnologia ha deixat arrere les fases d'innovació i estandardització, ara és temps d'aconseguir als primers usuaris del mercat. No obstant això, la majoria de fabricants han decidit implementar els dispositius per a VN com a unitats de bord (OBU), les quals són cares i difícils d'actualitzar. A més, les OBU són molt dependents del fabricant original. Tot això està retardant el desplegament de les VN. Per a accelerar l'adopció de les VN, hem desenvolupat l'arquitectura GRCBox. L'arquitectura GRCBox està basada en un dispositiu de baix cost que permet als usuaris usar comunicacions V2V mentre usen dispositius de propòsit general, com ara telèfons intel·ligents, tauletes o portàtils. Les proves incloses en aquesta tesi demostren la viabilitat de l'arquitectura GRCBox. Mitjançant la combinació de la nostra GRCBox i la plataforma de DTN Scampi, hem dissenyat i provat un escenari VDTN pràctic. També presentem com els desenvolupadors poden crear noves aplicacions GRCBox per a portar el potencial de les VN als dispositius d'usuari. / Martínez Tornell, S. (2016). Delay Tolerant Networks for Efficient Information Harvesting and Distribution in Intelligent Transportation Systems [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/68486

In the pursuit for ever-increasing compute power, and with Moore's law slowly coming to an end, high-performance computing started to scale-out to larger systems. Alongside the increasing system size, the interconnection network is growing to accommodate and connect tens of thousands of compute nodes. These networks have a large influence on total cost, application performance, energy consumption, and overall system efficiency of the supercomputer. Unfortunately, state-of-the-art routing algorithms, which define the packet paths through the network, do not utilize this important resource efficiently. Topology-aware routing algorithms become increasingly inapplicable, due to irregular topologies, which either are irregular by design, or most often a result of hardware failures. Exchanging faulty network components potentially requires whole system downtime further increasing the cost of the failure. This management approach becomes more and more impractical due to the scale of today's networks and the accompanying steady decrease of the mean time between failures. Alternative methods of operating and maintaining these high-performance interconnects, both in terms of hardware- and software-management, are necessary to mitigate negative effects experienced by scientific applications executed on the supercomputer. However, existing topology-agnostic routing algorithms either suffer from poor load balancing or are not bounded in the number of virtual channels needed to resolve deadlocks in the routing tables. Using the fail-in-place strategy, a well-established method for storage systems to repair only critical component failures, is a feasible solution for current and future HPC interconnects as well as other large-scale installations such as data center networks. Although, an appropriate combination of topology and routing algorithm is required to minimize the throughput degradation for the entire system. This thesis contributes a network simulation toolchain to facilitate the process of finding a suitable combination, either during system design or while it is in operation. On top of this foundation, a key contribution is a novel scheduling-aware routing, which reduces fault-induced throughput degradation while improving overall network utilization. The scheduling-aware routing performs frequent property preserving routing updates to optimize the path balancing for simultaneously running batch jobs. The increased deployment of lossless interconnection networks, in conjunction with fail-in-place modes of operation and topology-agnostic, scheduling-aware routing algorithms, necessitates new solutions to solve the routing-deadlock problem. Therefore, this thesis further advances the state-of-the-art by introducing a novel concept of routing on the channel dependency graph, which allows the design of an universally applicable destination-based routing capable of optimizing the path balancing without exceeding a given number of virtual channels, which are a common hardware limitation. This disruptive innovation enables implicit deadlock-avoidance during path calculation, instead of solving both problems separately as all previous solutions.