No description available.

State-of-the-art onboard spacecraft avionics use SpaceWire networks to interconnect payload data-handling sub-systems. This includes high data-rate sensors and instruments, processing units, and memory devices. SpaceWire is an interconnection network composed of nodes and routers connected by bi-directional, point-to-point, high-speed, serial-data communication links. SpaceWire is established as one of the main data-handling protocols and is being used on many ESA, NASA and JAXA spacecraft. SpaceWire is very successful for being fast, flexible and simple to use and implement. However it does not implement Quality of Service mechanisms, which aim to provide guarantees in terms of reliability and timely delivery to data generated by network clients. Quality of Service is increasingly being deployed in commercial ground technologies and its availability for space applications, which requires high reliability and performance, is of high interest for the space community. This thesis researches how Quality of Service can be provided to existing SpaceWire networks. Existing solutions for ground-based technologies cannot be directly used because of the constraints imposed by the limitations of space-qualified electronics. Due to these limitations SpaceWire uses wormhole routing which has many benefits but makes it more challenging to obtain timing guarantees and to achieve a deterministic behaviour. These challenges are addressed in this work with a careful analysis of existing Quality of Service techniques and the implementation of a novel set of protocols specifically designed for SpaceWire networks. These new protocols target specific use cases and utilise different mechanisms to achieve the required reliability, timely delivery and determinism. Traditional and novel techniques are deployed for first time in SpaceWire networks. In particular, segmentation, acknowledgements, retry, time-division multiplexing an cross-layer techniques are considered, analysed, implemented and evaluated with extensive prototyping efforts. SpaceWire provides high-rate data transfers but the next generation of payload instruments are going to require multi-gigabit capabilities. SpaceFibre is a new onboard networking technology under development which aims to satisfy these new requirements, keeping compatibility with SpaceWire user-applications. As a new standard, SpaceFibre offers the opportunity to implement Quality of Service techniques without the limitations imposed by the SpaceWire standard. The last part of this thesis focuses on the specification of the SpaceFibre standard in order to provide the Quality of Service required by next generation of space applications. This work includes analytical studies, software simulations, and hardware prototyping of new concepts which are the basis of the Quality of Service mechanisms defined in the new SpaceFibre standard. Therefore, a critical contribution is made to the definition and evaluation of a novel Quality of Service solution which provides high reliability, bandwidth reservation, priority and deterministic delivery to SpaceFibre links.

The main aim of my master thesis was to get acquainted with the implementation of quality-of-service support methods in UMTS 3G mobile network and the evaluation of these methods in a simulation environment with main interest in the characteristic features of the traffic classes. The impact of the QoS methods available on different service types were verified in the OPNET Modeler simulation environment. I examined mainly the potentials of the model to support the required quality parameters for the connections. The models of typical network applications, available in real UMTS networks, together with two examples of mobile networks, operating voice and video services in combination with standard data services, were defined in this simulation environment. My work also contains the detailed description of configuration parameters fundamental for correct behavior of the simulation models. The simulation also shows the case, when the transport service is rejected due to the lack of network resources. The key parameters of the model are also described and their effect on the service quality with and without QoS control are compared in separate scenarios. Finally, I investigated the effect of mobility of the end stations on the communication process, especially on the duration of a file transmission.

Der Trend zu kleinteiligeren und kapitalintensiveren Transportgütern führt in Kombination mit der in Europa angestrebten Reduktion von Treibhausgasemissionen im Transportsektor zu einer Zunahme der Bedeutung von effizienten Konsolidierungsnetzwerken des Schienengüterverkehrs. Zugehörige Produktionsformen mit Bündelung von Warenströmen wie der Einzelwagenverkehr und der intermodale Verkehr sind somit erfolgskritisch für die Zukunftsfähigkeit des Schienengüterverkehrs. Deren Wettbewerbsfähigkeit kann durch die Einführung und Nutzung von Innovationen gestärkt werden. Beispiele hierfür sind eine Digitale Automatische Kupplung (DAK) sowie Sensoren an Güterwagen und Lokomotiven. Diese Innovationen werden oftmals von einem hohen monetären Aufwand sowie Unsicherheiten hinsichtlich ihrer genauen betrieblichen Wirkung in den Netzwerken begleitet. Für strategische Entscheidungen hinsichtlich einer Einführung solcher Innovationen sind die ökonomischen und betrieblichen Effekte für gezielte Nutzen-Kosten-Betrachtungen aufzuzeigen sowie mögliche Pfade für eine Migration der jeweiligen Innovation für die Produktionsformen und deren zugehörigen Netzwerke zu eruieren. Dabei sind insbesondere die Veränderungen im sogenannten Service Network Design (SND) von großer Bedeutung. Das SND ist Teil der taktischen Netzwerkplanung und definiert das Zuggerüst sowie die Wagenroutenplanung im Netzwerk. Dabei werden die Kosten für den Betrieb von Netzwerken unter Einhaltung definierter Qualitätsstandards minimiert. Das Ergebnis des SND stellt den Rahmen für konkrete Wagenrouten in der betrieblichen Durchführung dar und definiert das zu behandelnde Zuggerüst in den Bündelungsknoten der Netzwerke. Trotz der wichtigen Funktion des SND, ist dieser taktische Planungsprozess in der Praxis noch stark manuell geprägt und daher zeitaufwändig. Außerdem liefert er oft zu ungenaue Aussagen. Das trifft insbesondere auf Aussagen zu den Netzeffekten durch Innovationen zu. Aufgrund der hohen Komplexität von Konsolidierungsnetzwerken des Schienengüterverkehrs und fehlender EDV-Unterstützung basieren Betrachtungen zu den Effekten von Innovationen in Netzwerken im Status Quo auf Expertenbefragungen und Abschätzungen. Insbesondere für Innovationen, deren Migration mit weitreichenden Prozessänderungen im Netzwerk und neuen Betriebsstrategien verbunden ist, bedarf es jedoch objektiver modellbasierter Verfahren zur Entscheidungsunterstützung. Durch deren Einsatz könnten die Auswirkungen in Bezug auf Netzwerkstrukturen und Kosten ermittelt und für Entscheidungsträger:innen transparent dargestellt werden. Die vorliegende Dissertation leistet einen wissenschaftlichen Beitrag, um dieses Potenzial zu erschließen. Hierfür wird im Rahmen der Dissertation eine neue Methode als Beitrag zur Entscheidungsunterstützung für die Einführung und Migration von Innovationen in Konsolidierungsnetzwerken des Schienengüterverkehrs entwickelt: TRENO (TRansparent Evaluation of InNOvation Effects in Rail Freight Service Networks). Die Methode kombiniert dabei eisenbahnbetriebswissenschaftliche Grundlagen mit Ansätzen aus dem Operations Research und dem maschinellen Lernen. Prozessveränderungen durch Innovationen werden analysiert und in einem neuartigen mathematischen Optimierungsmodell für das SND abgebildet. Das Modell ermöglicht die Definition von verschiedenen Betriebsstrategien im Netzwerk und bildet erstmals zeitbezogene Infrastrukturnutzungen von Zügen in den Knoten des Netzwerks ab. Da die Einführung von Innovationen mit hoher Unsicherheit hinsichtlich der Annahmen und Eingangsparameter verbunden ist, sind zahlreiche Szenarien für eine fundierte Entscheidungsunterstützung zu definieren und zu bewerten. Aufgrund der hohen Komplexität von SND Modellen sind Berechnungen mittels mathematischer Optimierung sehr zeitintensiv. Daher nutzt TRENO Klassifizierungs- und Regressionsmodelle aus dem Bereich des maschinellen Lernens, welche die mathematische Optimierung ergänzen. Auf Basis eines Pools von Szenarien, für welche optimale Netzwerkstrukturen mithilfe mathematischer Optimierung berechnet wurden, lernen die Klassifizierungs- und Regressionsmodelle den Zusammenhang zwischen Eingangsdaten und den resultierenden Kennzahlen des Netzwerks. Nach diesem Training können die Modelle dazu eingesetzt werden, die Kennzahlen von Netzwerken (insbesondere Kosten, Zuggerüste sowie Auslastungen von Zügen und Bündelungsknoten) für zahlreiche neue Szenarien innerhalb von Sekunden vorherzusagen. Dies stellt eine maßgebliche Beschleunigung gegenüber der mathematischen Optimierung dar. Die Klassifizierungsmodelle werden genutzt, um die grundsätzliche (Un-)Lösbarkeit eines Szenarios vorherzusagen, die beispielsweise durch unzureichende Kapazitäten in den Zugbildungsanlagen resultieren kann. Die Regressionsmodelle prognostizieren spezifische metrische Kennzahlen des Netzwerks wie Kosten, Zuggerüste und Kapazitätsauslastungen. Neben dieser Kernfunktionalität von TRENO ermöglicht die Integration der SHAP Methode (shapley additive explanations) eine Analyse bezüglich des Einflusses der Eingangsparameter auf die Kennzahlen eines Konsolidierungsnetzwerks. Dies erlaubt den Aufbau eines tiefgründigen Verständnisses der Wirkungszusammenhänge in Konsolidierungsnetzwerken des Schienengüterverkehrs (z. B. durch die Identifikation von maßgeblichen Kostentreibern) und wirkt einem grundsätzlichen Problem aus dem Bereich des maschinellen Lernens, der mangelnden Interpretierbarkeit der Modelle, entgegen. TRENO wird anhand eines praxisnahen Anwendungsfalls validiert, der Planung einer Migration einer DAK in einem exemplarischen europäischen Einzelwagenverkehrsnetz. Hierbei wird mit TRENO untersucht, welchen Einfluss verschiedene Betriebsstrategien während einer Migration einer DAK auf die Kosten und Strukturen im Netzwerk haben. Das Anwendungsbeispiel zeigt auf, dass sich mit TRENO verschiedene komplexe Betriebsstrategien mathematisch modellieren lassen. Durch die Anwendung der Methode werden die konkreten Effekte der Betriebsstrategien auf die definierten Kennzahlen des Netzwerks transparent gemacht. Dies ermöglicht neue Schlussfolgerungen aus eisenbahnbetriebswissenschaftlicher Sicht hinsichtlich der Wahl von Betriebsstrategien während einer Migration. Ferner zeigen die Ergebnisse die hohe Qualität der Prognosen durch die Klassifizierungs- und Regressionsmodelle auf. Beim Test von vier Klassifizierungs- und fünf Regressionsmodellen erzielen Modelle auf Basis des Gradient Boosting Verfahrens die besten Ergebnisse. Für die Klassifizierung erzielt das Modell in 94% der Fälle richtige Vorhersagen. Das Regressionsmodell erzielt im Durchschnitt über alle Kennzahlen ein Bestimmtheitsmaß von 93% und kann damit einen Großteil der Varianz in den Datensätzen erklären. TRENO stellt somit einen anwendbaren Beitrag dar, um den manuellen Prozess zur Planung von Konsolidierungsnetzwerken im Schienengüterverkehr insbesondere für den Fall der Einführung von Innovationen maßgeblich zu beschleunigen und zu automatisieren. Der modellbasierte Ansatz objektiviert die Entscheidungsunterstützung gegenüber dem Status Quo und ermöglicht zudem eine weitreichende Exploration des Einflusses von Innovationen auf die Strukturen des Netzwerks über zahlreiche Szenarien. Hierdurch erweitert die Dissertation das Methodenspektrum der Eisenbahnbetriebswissenschaften durch die Verzahnung mit Verfahren aus dem Operations Research sowie des maschinellen Lernens. / The trend towards small-scale and more capital-intensive transport goods means that the importance of rail freight consolidation networks is increasing. Production forms such as single wagonload transport and intermodal transport are therefore critical for the future potential of rail freight. The competitiveness of rail freight networks can be strengthened through the introduction of innovations such as a Digital Automatic Coupling (DAC) and sensors on freight wagons and locomotives. These innovations are often accompanied by high investment costs and uncertainties regarding their specific operational impact in the networks. For strategic decisions regarding the introduction of such innovations, the economic and operational effects must be identified and quantified, e.g., for benefit-cost considerations. In this context, the changes in the so-called Service Network Design (SND) are of particular importance. The SND determines train service structures and the freight distribution in networks at the tactical planning level. The number and schedule of operated train services in the network significantly determines the costs and quality of consolidation networks in rail freight. It provides the framework for specific railcar routings at the operational level and defines the number of trains to be handled in the bundling nodes of the networks. Despite this outstanding importance of the SND, the tactical planning of consolidation networks is still a manual process in practice lacking computer-based decision support. This is particularly true for statements on network effects of innovations. Due to the high complexity of consolidation networks in rail freight transport and the lack of computer-based decision support, analyses regarding the effects of innovations in service networks are mainly based on expert interviews in the status quo. Especially for innovations whose migration is associated with extensive process transformations and new operating strategies in the network, there is a need for objective model-based methods to support decision-making. Thereby, the operational effects of innovations on service network structures and costs could be determined and made transparent to decision-makers. This dissertation contributes to close this gap and to enable an efficient planning of consolidation networks. For this purpose, the dissertation develops a new method as a contribution to decision support for the introduction and migration of innovations in consolidation networks of rail freight transport: TRENO (TRansparent Evaluation of InNOvation Effects in Rail Freight Service Networks). The method combines rail transport planning with approaches from operations research and machine learning. Process changes due to innovations are analyzed and depicted in a novel mathematical optimization model for the SND. The model enables the definition of different operating strategies in the network and incorporates dynamic infrastructure usages of trains in the nodes of the network. Since the introduction of innovations is associated with a high degree of uncertainty regarding the underlying assumptions and input parameters, numerous scenarios have to be defined and evaluated for a sound decision support. Due to the high complexity of SND models, mathematical optimization is computationally expensive and time-consuming. Therefore, TRENO applies classification and regression models from the field of machine learning complementing mathematical optimization. Based on a set of scenarios for which optimal network structures have been computed using mathematical optimization, the classification and regression models learn the relationship between input data and the resulting key figures of the network. After training, the models can be used to predict the key figures of networks for various new scenarios within seconds (in particular cost structures, the number of operated train services and utilization of trains and yards). This represents a significant acceleration compared to mathematical optimization. The classification models are used to predict the feasibility of a scenario, which can, for example, result from insufficient capacities in the nodes of the network. The regression models predict specific metrics of the network such as costs, train service structures and capacity utilization. In addition to this core functionality of TRENO, the integration of the SHAP method (shapley additive explanations) allows an analysis of the influence of input parameters on the key figures of a consolidation network. This contributes to the understanding of the interdependencies in consolidation networks of rail freight transport (e.g., by identifying major cost drivers) and counteracts a fundamental problem from the field of machine learning, the lack of interpretability of the models. TRENO is validated on the basis of a relevant use case, the planning of a migration of a DAC in an exemplary European single wagonload network. Here, TRENO is used to investigate the influence of different operating strategies during a migration of a DAC on the costs and service structures in the network. The example shows that different complex operating strategies can be modelled with TRENO. By applying the method, the specific effects of the operating strategies on the defined key figures of the network are made transparent. This enables to draw new conclusions from a rail transport planning perspective regarding the choice of operating strategies during a migration. Furthermore, the results show the high quality of the predictions by the classification and regression models. When testing four classification and five regression algorithms, models based on gradient boosting achieve the best results. For classification, the model yields correct predictions in 94% of the cases. The regression model achieves an average coefficient of determination of 93% across all key figures and can thereby explain a large part of the variance in the data. TRENO thus represents an applicable contribution to significantly automate and accelerate the manual process for planning consolidation networks in rail freight transport, especially for the case of the introduction of innovations. The model-based approach provides a more objective decision support compared to the status quo and enables to study the influence of innovations on the structures of service networks over numerous scenarios. Hereby, the dissertation expands the methodological spectrum of rail transport planning by linking it with methods from operations research and machine learning.

Student Number : 9812990V - MSc project report - School of Electrical Engineering - Faculty of Engineering and the Built Environment / The NGN is a multi-service network which inter-works with the Public Switched Telephone Network (PSTN), the voice network and the data network provided by Internet. Through network independent APIs such as OSA-Parlay, the NGN slowly migrates and converges Telecoms and IT networks, voice and Internet, into a common packet infrastructure. The OSA/Parlay group defines a softswitch architecture which provides network independent APIs or SCFs that enable cross network application development The Parlay softswitch provides connectivity to underlying transport networks for application providers. The standard specifies the interaction between application providers and the softswitch. However, the standard does not specify an interface to regulate the interaction between service providers and the consumer/end user domain. This means that applications housed in the service provider domain have no defined interfaces to manage service delivery to the consumer domain. For most service providers, the lack of a non-standardized API set impedes efforts to decrease application creation and deployment time. This research investigates the design and implementation of a standard consumer interface which can be used by application providers within an OSA/Parlay system to deliver service content to end users. The main objectives with regard to the functionality provided by the interface include the integration of facilities which will assist application providers to manage end user access and authentication (to enable users to establish a secure context for service usage), subscription (to handle the subscription life cycle), and service usage management (to enable the initiation and termination of services). The TINA-Consortium (TINA-C) has developed a service architecture to support the creation and provisioning of services in the NGN. The TINA architecture offers a comprehensive set of concepts and principles that can be used in the design of NGN services. The architecture consists of a set of reusable and interoperable service components encapsulating a rich and well defined set of APIs aimed at supporting the interaction between application providers and consumers. TINA’s session concepts, information structures, interfaces and service components can be used to support the design of a consumer premises end user interface for OSA/Parlay. This research also aims to explore the feasibility of using the TINA API within an OSA/Parlay system to support consumer domain service delivery. In order to implement the consumer interface for Parlay applications, the ability of the TINA service architecture to provide Access and Authentication management; Subscription and Profile management; and Service Usage management was investigated. The report documents the design and implementation of an OSA/Parlay consumer interface utilizing TINA service components and interfaces.

In the distribution network of a large retailer, shipments can either be transported by the retailer's own trucks or outsourced to third-party logistics (3PL) companies. In the former case, shipments are consolidated and transported from their origins through an intermediate facility, namely a crossdock. At a crossdock, shipments are unloaded, sorted, re-consolidated, loaded and transported to their destinations. The consolidation process offers economies of scale that reduce the transportation costs. At the same time, it increases travel distances and incurs handling costs at a crossdock. For this reason, consolidation is uneconomic for a shipment in which origin and destination are located close to one other, especially through a distant crossdock. It is cheaper to outsource transportation of such a shipment to 3PL companies. This shipping decision raises a series of questions. Should a shipment be consolidated through a crossdock or outsourced to 3PL companies? How do facility locations, the operational cost of a crossdock and mode of shipments influence the shipping decision? Can the robustness and potential growth of a crossdock be measured? How does outsourcing affect the robustness and potential growth of a crossdock? We formulate a strategic model of a retailer's distribution network as an economic trade-off between consolidated shipments through a crossdock and outsourced shipments to 3PL companies. We study the locus of facility locations where the costs of a consolidated shipment and an outsourced shipment are equal and discover that the trade-off can be modeled by classical geometric curves, particularly an ellipse, a hyperbola, a limacon and a Cartesian oval. These curves can be developed into a preliminary routing and locating tool. We also observe interesting connections between the single hub crossdocking network and other fields of geometric study, such as Voronoi diagrams and geometric inversion. In addition, the area bounded by these curves represents the likelihood in which a particular shipment is consolidated through a crossdock. We expand this concept to multiple vendor-store pairs and suggest an index that measures robustness and potential growth of a particular crossdock. This asymptotic-probability index explains economic driving factors of consolidation and outsourcing. Although the derivation of the index is limited by the dimension and spatial distribution of facilities, its numerical value can be determined by a computer simulation. Therefore, we use Monte Carlo simulation to compute the proposed index to explain the outsourcing and the interaction between TL threshold0.1 and mode of shipments. The analysis and computer simulation suggest that outsourcing may cause an adverse effect in a single hub crossdocking network, resulting in the abrupt reduction of consolidated shipments in the network. Furthermore, we propose transportation planning to alleviate this effect and compare them to the optimal allocation. The routing and locating application of the model is illustrated using the Home Depot distribution network. Our model predicts 5.5% and additional 1.0% savings in transportation cost by re-allocation of shipments and re-location of crossdocks, respectively. The empirical study shows that the adverse effect of outsourcing can be eliminated by limiting the number of crossdocks used by each store.

本研究主要是欲從1980年代福利國家危機所發展出來的因應對策“福利多元化”的理念出發，依此理念所發展出來的「少年福利多元化」的政策主張，來探討我國少年福利在福利多元化政策上的施行現況、成效與困境。因此，對於目前建構少年福利服務網絡的少年福利各相關服務組織，包括政府單位、及民間社會福利機構，在服務網絡上如何進行協調聯繫，及服務項目整合的情形，即瞭解目前少年服務機構的服務提供、與少年服務網絡聯繫建構現況，並藉由探討歸納少年服務網絡建構的影響因素，以尋求最佳少年服務網絡模式的建立。 因此，本研究的研究架構，主要有3個概念即：少年福利各相關服務組織特性—組織進行協調聯繫，與服務整合現況—建構最佳少年服務網絡。而在研究方法的採取上，則主要是採質化的研究方法，並以文獻探討法，及質的訪談法2者相輔。文獻探討的部分主要是包括：整體少年福利施行現況，及高雄市少年福利相關文獻資料的蒐集。而質的訪談法部分，則是兼採開放式與封閉式的訪談型式來深入探討蒐集資料。另外，在訪談對象的選取，則是先採立意取樣的方式，選取資料最多、經驗最豐富的青少年福利服務中心，作為基礎，再由訪談的青少年福利服務中心，以滾雪球的方式確定有少年業務聯繫關係的民間社會福利機構對象。最後再針對這些民間社會福利機構，進行約談訪問。 本研究的研究發現則主要有以下幾點： 1.資源不足，服務提供有限：不論是針對一般青少年的休閒娛樂服務方面，亦或是不幸少年的寄養、與安置服務上，各少年服務機構皆無法完全配合少年的服務需求來提供服務。 2.人力不足，經費編列有限：不論是政府單位、青少年福利服務中心，亦或是民間社會福利機構，皆是普遍會存在的服務機構問題，且受訪的機構服務工作者皆反應機構目前在人力上有呈現不足、缺乏的困境，多數機構的社工員必須身兼數職，工作負擔很重。 3.資源尋求，促進服務聯繫：研究發現：促進服務機構進行少年服務協調聯繫的主要動機，不論是在政府單位，亦或是民間社會福利機構彼此間，促進其進行服務聯繫的動機，不外乎以資源相互尋求、與相互支援為主。 4.個案整合，機構合作無間：在高雄市少年服務網絡的建構中，主管機關社會局在建立少年服務資源網絡上，除與其所屬的青少年福利服務中心相互聯繫外，在少年服務的提供上亦多以公辦民營、協助補助、個案委託、方案委託等方式，交由民間機構來辦理少年福利服務，尤其在公設民營、與委託業務上更是蓬勃。另外，在機構所進行服務整合的項目，則是以少年個案整合為主。 5.本位主義，網絡建構難題：不論是青少年福利服務中心，亦或是民間社會福利機構的工作者皆反應，在機構間的協調聯繫成效方面，皆有遇到與教育機構、單位溝通協調困難的問題，而主要的原因則是專業機構本位主義所致。 6.定位不明，認知未達共識：服務機構工作者反應政府社政單位在與民間社會福利機構，進行委辦、承辦、協辦等不同業務時，經常由於彼此間對相互的角色、定位的認知不同，而產生協調上的問題。 7.網絡建構，默契須再培養：雖然目前整個少年服務網絡在建構上，各專業服務機構彼此間在聯繫的動機、及整合情形，皆有助於整個少年服務網絡體系的建構與維繫，但彼此間在少年服務工作的團隊默契上，仍須再培養。 8.資源競爭，機構相互排擠：少年福利的民間資源不足，使得少年在老人、婦女、兒童等弱勢族群中，又是更為弱勢的一群。且機構工作者表示：在資源缺乏下，同性質的服務機構彼此間會相互競爭資源、相互排擠，如此使得小型、新成立的服務機構，難以生存的現象。 最後在研究限制方面，則有2部分，即：研究方法的限制，與研究時間、人力的限制。研究方法的限制，則是包括：文獻研究法、與質的訪談法等2種資料蒐集方法的缺點，也就是在研究方法上的限制。而在研究時間、與人力的限制方面，則為研究範圍上的限制、及訪談對象數量上的限制等此2部分，為本研究的研究方面的限制。

The dissertation investigates the form of network governance in the context of U.S. homeless service networks (namely continuum of care programs; CoCs). This research examines CoC homeless service networks by applying the institutional collective action (ICA) perspectives to understand the forms of network governance as a reflection of network context. The ICA perspective has been applied to understand the rational behavior of network members for the network governance form to mitigate the collective action problems. The ICA perspective helps understand why network members accept specific governance structures with their expectation to maximize the benefits and minimize the costs and uncertainty in their process of collaboration. This dissertation uses the data of CoC networks and point in time data from the U.S. Department of Housing and Urban Development in 2019 and Census. The data developed for this study offers the homeless incidences, geographical characteristics, and governance structure based on the contact information. For an in-depth understanding, interview by CoC leaders was integrated. This dissertation consists of four essays about 1) Literature review on network governance and the theoretical argument in the ICA framework, 2) Background and network governance of the U.S. homeless service networks, 3) Factors affecting the choice of network administrative organization (NAO) form, and 4) Interviews by the representatives of CoC networks. The findings inform us of the governance structure for the effective service provisions and coordination of actions of network members and about why and how network organizations choose a form of network governance.

This dissertation examines the factors and combinations of factors that affect the performance of continuum of care homeless service networks, applying the configurational approach of contingency theory to data sets drawn from federal sources. The study accepts the two key assumptions from the theory: (1) that there are multiple paths to high performance and (2) key conditions have a joint influence on network performance. The data analysis in this study has two parts. The first employs OLS regression to examine the causal relationship between independent variables and the performance of permanent supportive housing (PSH) programs. This study also applies fuzzy set qualitative comparative analysis (fsQCA) to identify multiple combinations of factors that influence the performance of PSH programs. The results identify key factors and multiple combinations of factors that lead to high and low network performance. Federal CoC funding emerges as a core condition for high and low performance, but all relevant conditions can be critical factors depending on how they interact with other relevant conditions. This analysis helps expand the utility of existing contingency theory by using it to explain the dynamic interactions between factors in the context of public service networks. Ultimately, this dissertation confirms that fsQCA can be a useful method for evaluating the performance of public service networks and helping them provide more services more effectively.

Cette thèse se place dans le cadre de services web en dépassant leur description pour considérer leur structuration en réseaux (réseaux d'interaction et réseaux de similitude). Nous proposons des méthodes basées sur les motifs, la modélisation probabiliste et l'analyse des concepts formels, pour améliorer la qualité des services découverts. Trois contributions sont alors proposées: découverte de services diversifiés, recommandation de services et cohérence des communautés de services détectées. Nous structurons d'abord les services sous forme de réseaux. Afin de diversifier les résultats de la découverte, nous proposons une méthode probabiliste qui se base à la fois sur la pertinence, la diversité et la densité des services. Dans le cas de requêtes complexes, nous exploitons le réseau d'interaction de services construit et la notion de diversité dans les graphes pour identifier les services web qui sont susceptibles d'être composables. Nous proposons également un système de recommandation hybride basé sur le contenu et le filtrage collaboratif. L'originalité de la méthode proposée vient de la combinaison des modèles thématiques et les motifs fréquents pour capturer la sémantique commune maximale d'un ensemble de services. Enfin, au lieu de ne traiter que des services individuels, nous considérons aussi un ensemble de services regroupés sous forme de communautés de services pour la recommandation. Nous proposons dans ce contexte, une méthode qui combine la sémantique et la topologie dans les réseaux afin d'évaluer la qualité et la cohérence sémantique des communautés détectées, et classer également les algorithmes de détection de communautés. / As a part of this thesis, we exceed the description of web services to consider their structure as networks (i.e. similarity and interaction web service networks). We propose methods based on patterns, topic models and formal concept analysis, to improve the quality of discovered services. Three contributions are then proposed: (1) diversified services discovery, (2) services recommendation and (3) consistency of detected communities. Firstly, we propose modeling the space of web services through networks. To discover the diversified services corresponding to a given query, we propose a probabilistic method to diversify the discovery results based on relevancy, diversity and service density. In case of complex requests, it is necessary to combine multiple web services to fulfill this kind of requests. In this regard, we use the interaction web service network and the diversity notion in graphs to identify all possible services compositions. We also propose a new hybrid recommendation system based on both content and collaborative filtering. Its originality comes from the combination of probabilistic topic models and pattern mining to capture the maximal common semantic of a set of services. Finally, instead of processing individual services, we consider a set of services grouped into service communities for the recommendation. We propose in this context, a new method combining both topology and semantics to evaluate the quality and the semantic consistency of detected communities, and also rank the detection communities algorithms.