京都大学 / 0048 / 新制・課程博士 / 博士(情報学) / 甲第18530号 / 情博第534号 / 新制||情||95(附属図書館) / 31416 / 京都大学大学院情報学研究科知能情報学専攻 / (主査)教授 岡部 寿男, 教授 松山 隆司, 教授 阿久津 達也 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

In this thesis we look at mobile agent solutions to black hole search and related problems. Mobile agents are computational entities that are autonomous, mobile, and can interact with their environment and each other. The black hole search problem is for a team of these agents to work together to map or explore a graph-like network environment where some elements of the network are dangerous to the agents. Most research into black hole search has focussed on finding a single dangerous node: a black hole. We look at the problem of finding multiple black holes and, in the case of dangerous graph exploration, multiple black links as well. We look at the dangerous graph exploration problem in the network model. The network model is based on a normal static computer network modelled as a simple graph. We give an optimal solution to the dangerous graph exploration problem using agents that start scattered on nodes throughout the network. We then make the problem more difficult by allowing an adversary to delete links during the execution of the algorithm and provide a solution using scattered agents. In the last decade or two, types of networks have emerged, such as ad hoc wireless networks, that are by their nature dynamic. These networks change quickly over time and can make distributed computations difficult. We look at black hole search in one type of dynamic network described by the subway model, which we base on urban subway systems. The model allows us to look at the cost of opportunistic movement by requiring the agents to move using carriers that follow routes among the network's sites, some of which are black holes. We show that there are basic limitations on any solution to black hole search in the subway model and prove lower bounds on any solution's complexity. We then provide two optimal solutions that differ in the agents' starting locations and how they communicate with one another. Our results provide a small window into the cost of deterministic distributed computing in networks that have dynamic elements, but which are not fully random.

In this thesis we look at mobile agent solutions to black hole search and related problems. Mobile agents are computational entities that are autonomous, mobile, and can interact with their environment and each other. The black hole search problem is for a team of these agents to work together to map or explore a graph-like network environment where some elements of the network are dangerous to the agents. Most research into black hole search has focussed on finding a single dangerous node: a black hole. We look at the problem of finding multiple black holes and, in the case of dangerous graph exploration, multiple black links as well. We look at the dangerous graph exploration problem in the network model. The network model is based on a normal static computer network modelled as a simple graph. We give an optimal solution to the dangerous graph exploration problem using agents that start scattered on nodes throughout the network. We then make the problem more difficult by allowing an adversary to delete links during the execution of the algorithm and provide a solution using scattered agents. In the last decade or two, types of networks have emerged, such as ad hoc wireless networks, that are by their nature dynamic. These networks change quickly over time and can make distributed computations difficult. We look at black hole search in one type of dynamic network described by the subway model, which we base on urban subway systems. The model allows us to look at the cost of opportunistic movement by requiring the agents to move using carriers that follow routes among the network's sites, some of which are black holes. We show that there are basic limitations on any solution to black hole search in the subway model and prove lower bounds on any solution's complexity. We then provide two optimal solutions that differ in the agents' starting locations and how they communicate with one another. Our results provide a small window into the cost of deterministic distributed computing in networks that have dynamic elements, but which are not fully random.

In this thesis we look at mobile agent solutions to black hole search and related problems. Mobile agents are computational entities that are autonomous, mobile, and can interact with their environment and each other. The black hole search problem is for a team of these agents to work together to map or explore a graph-like network environment where some elements of the network are dangerous to the agents. Most research into black hole search has focussed on finding a single dangerous node: a black hole. We look at the problem of finding multiple black holes and, in the case of dangerous graph exploration, multiple black links as well. We look at the dangerous graph exploration problem in the network model. The network model is based on a normal static computer network modelled as a simple graph. We give an optimal solution to the dangerous graph exploration problem using agents that start scattered on nodes throughout the network. We then make the problem more difficult by allowing an adversary to delete links during the execution of the algorithm and provide a solution using scattered agents. In the last decade or two, types of networks have emerged, such as ad hoc wireless networks, that are by their nature dynamic. These networks change quickly over time and can make distributed computations difficult. We look at black hole search in one type of dynamic network described by the subway model, which we base on urban subway systems. The model allows us to look at the cost of opportunistic movement by requiring the agents to move using carriers that follow routes among the network's sites, some of which are black holes. We show that there are basic limitations on any solution to black hole search in the subway model and prove lower bounds on any solution's complexity. We then provide two optimal solutions that differ in the agents' starting locations and how they communicate with one another. Our results provide a small window into the cost of deterministic distributed computing in networks that have dynamic elements, but which are not fully random.

In this thesis we look at mobile agent solutions to black hole search and related problems. Mobile agents are computational entities that are autonomous, mobile, and can interact with their environment and each other. The black hole search problem is for a team of these agents to work together to map or explore a graph-like network environment where some elements of the network are dangerous to the agents. Most research into black hole search has focussed on finding a single dangerous node: a black hole. We look at the problem of finding multiple black holes and, in the case of dangerous graph exploration, multiple black links as well. We look at the dangerous graph exploration problem in the network model. The network model is based on a normal static computer network modelled as a simple graph. We give an optimal solution to the dangerous graph exploration problem using agents that start scattered on nodes throughout the network. We then make the problem more difficult by allowing an adversary to delete links during the execution of the algorithm and provide a solution using scattered agents. In the last decade or two, types of networks have emerged, such as ad hoc wireless networks, that are by their nature dynamic. These networks change quickly over time and can make distributed computations difficult. We look at black hole search in one type of dynamic network described by the subway model, which we base on urban subway systems. The model allows us to look at the cost of opportunistic movement by requiring the agents to move using carriers that follow routes among the network's sites, some of which are black holes. We show that there are basic limitations on any solution to black hole search in the subway model and prove lower bounds on any solution's complexity. We then provide two optimal solutions that differ in the agents' starting locations and how they communicate with one another. Our results provide a small window into the cost of deterministic distributed computing in networks that have dynamic elements, but which are not fully random.

Rich Internet Applications (RIAs) have been widely used in the web over the last decade as they were found to be responsive and user-friendly compared to traditional web applications. RIAs use client-side scripting such as JavaScript which allows for asynchronous updates on the server-side using AJAX (Asynchronous JavaScript and XML). Due to the large size of RIAs and therefore the long time required for crawling, distributed RIA crawling has been introduced with the aim to decrease the crawling time. However, the current RIA crawling systems are not scalable, i.e. they are limited to a relatively low number of crawlers. Furthermore, they do not allow for fault tolerance in case that a failure occurs in one of their components. In this research, we address the scalability and resilience problems when crawling RIAs in a distributed environment and we explore the possibilities of designing an efficient RIA crawling system that is scalable and fault-tolerant. Our approach is to partition the search space among several storage devices (distributed databases) over a peer-to-peer (P2P) network where each database is responsible for storing only a portion of the RIA graph. This makes the distributed data structure invulnerable to a single point of failure. However, accessing the distributed data required by crawlers makes the crawling task challenging when the number of crawlers becomes high. We show by simulation results and analytical reasoning that our system is scalable and fault-tolerant. Furthermore, simulation results show that the crawling time using the P2P crawling system is significantly faster than the crawling time using both the non-distributed crawling system and the distributed crawling system using a single database.

Cette thèse explore le domaine du diagnostic appliqué à l'automobile basé sur le produit SIDIS Enterprise développé par Siemens AG. Au début, une analyse est conduite de l'importance du réseau après-vente pour les constructeurs et leurs places dans la chaine de valeur de l'automobile ainsi que les contraintes industrielles des outils d'aide au diagnostic. L'une d'elle est de limiter le coût de développement des modèles pour le diagnostic. De plus, une analyse approfondie démontre que la part des composants électroniques et des pannes d'origine électronique des véhicules à tendance à augmenter tout en complexifiant la localisation de leurs origines. Le besoin d'un outil moderne pour les constructeurs est essentiel pour la satisfaction client et pour l'image de marque. Ces travaux ont conduit au développement d'une plateforme modulaire pour le diagnostic composée : d'un module de recherche des symptômes, un système automatisé d'échange de fichier ODX, une méthode de diagnostic hybride et d'un moteur d'évaluation du retour d'expérience. Le premier module est dédié à l'interprétation et au mappage de la description des symptômes en langage naturel. Le second module permet aux auteurs des bases de données et des connaissances d'importer automatiquement les fichiers de descriptions des calculateurs électroniques. Les deux derniers sont dédiés à un algorithme de diagnostic qui combine les différentes sources de connaissances et bénéficie du retour d'expérience pour compléter automatiquement les modèles. Ces modules contribuent à l'objectif de réduction des coûts de développements des modèles pour le diagnostic et diminue significativement l'orbite des sessions de diagnostic guidé.

Nous étudions dans cette thèse le problème de l’exploration parallèle d’un graphe à l’aide des multiples, synchronisés et mobiles agents. Chaque agent est une entité individuelle qui peut, indépendamment des autres agents, visiter les sommets du graphe ou parcourir ses arêtes. Le but de ensemble des agents est de visiter tous les sommets de graphe.Nous étudions d’abord l’exploration du graphe dans un modèle où chaque agent est équipé de mémoire interne, mais les noeuds n’ont pas de mémoire. Dans ce modèle les agents sont autorisés à communiquer entre eux en échangeant des messages. Nous présentons des algorithmes qui s’exécutent dans un minimum de temps possible pour polynomiale nombre d’agents (polynomiale en nombre de sommets du graphe). Nous étudions aussi quelle est l’impact de différentes méthodes des communications. Nous étudions des algorithmes où les agents peuvent se communiquer à distance arbitraire,mais aussi où communication est possible seulement entre les agents situés dans le même sommet. Dans les deux cas nous présentons des algorithmes efficaces. Nous avons aussi obtenu des limites inférieures qui correspondent bien à la performance des algorithmes.Nous considérons également l’exploration de graphe en supposant que les mouvements des agents sont déterminés par le soi-disant rotor-router mécanisme. Du point de vue d’un sommet fixé, le rotor- router envoie des agents qui visitent les sommet voisins dans un mode round-robin. Nous étudions l’accélération défini comme la proportion entre le pire des cas de l’exploration d’un agent unique et des plusieurs agents. Pour générales graphes, nous montrerons que le gain de vitesse en cas de multi-agent rotor-router est toujours entre fonction logarithmique et linéaire du nombre d’agents. Nous présentons également des résultats optimaux sur l’accélération de multi-agent rotor-router pour cycles, expanseurs, graphes aléatoires, cliques, tores de dimension fixé et une analyse presque optimale pour hypercubes.Finalement nous considérons l’exploration sans collision, où chaque agent doit explorer le graphe de manière indépendante avec la contrainte supplémentaire que deux agents ne peuvent pas occuper le même sommet. Dans le cas où les agents sont donnés le plan de graphe, on présente un algorithme optimal pour les arbres et un algorithme asymptotiquement optimal pour générales graphes. Nous présentons aussi des algorithmes dans le cas de l’exploration sans collision des arbres et des générales graphes dans la situation où les agents ne connaissent pas le graphe. Nous fermons la thèse par des observations finales et une discussion de problèmes ouverts liés dans le domaine de l’exploration des graphes. / In this thesis we study the problem of parallel graph exploration using multiple synchronized mobile agents. Each mobile agent is an entity that can, independently of other agents, visit vertices of the graph and traverse its edges. The goal of the agents is to visit all vertices of the graph. We first study graph exploration in the model where agents are equipped with internal memory but no memory is available at the nodes. Agents in this model are also allowed to communicate between each other by exchanging messages. We present algorithms working in a minimal possible time for a team of polynomial size (in the number of vertices of the graph). We also study the impact of the available range of communication by analysing algorithms for agents which can communicate at arbitrary distance, or only with other agents located at the same node. We present efficient algorithms and lower bounds that almost match our positive results in both communication models. We also consider graph exploration when movements of agents are determined according to the so-called rotor-router mechanism. From the perspective of a fixed node, the rotor-router sends out agents which visit the node along its outgoing edges, ina round-robin fashion. We study the speedup which is the ratio between the worst-case exploration of a single agent and of multiple agents. We first show that the speed up for general graphs for the multi-agent rotor-router is always between logarithmic and linear in the number of agents. We also present a tight analysis of the speedup for the multi-agent rotor-router for cycles, expanders, random graphs, cliques, constant dimensional tori and an almost-tight analysis for hypercubes. Finally we consider collision-free exploration, where each agent has to explore the graph independently with the additional constraint that no two agents can occupy the same node at the same time. In the case when agents are given the map of the graph, we show an optimal algorithm for trees and an asymptotically optimal algorithm for general graphs. We also present algorithms for collision-free exploration of trees and general graphs in the case when agents have no initial knowledge about the graph. We close the thesis with concluding remarks and a discussion of related open problems in the area of graph exploration.

Computing with mobile agents is rapidly becoming a topic of mainstream research in the theory of distributed computing. The main research questions undertaken in this study concern the feasibility of solving fundamental tasks in an anonymous network, subject to limitations on the resources available to the agent. The considered challenges include: exploring a graph by means of an agent with limited memory, discovery of the network topology, and attempting to meet with another agent in another network (rendezvous). The constraints imposed on the agent include the number of moves which the agent is allowed to perform in the network, the amount of state memory available to the agent, the ability of the agent to communicate with other agents, as well as its a priori knowledge of the network topology or of global parameters.