Prediction techniques based on data are applied in a broad range of applications such as bioinformatics, disease spread, and mobile intrusion detection, just to name a few. With the rapid emergence of on-line technologies numerous techniques for collecting and storing data for prediction-based analysis have been proposed in the literature. With the growing size of global population, the spread of epidemics is increasing at an alarming rate. Consequently, public and private health care officials are in a dire need of developing technological solutions for managing epidemics. Most of the existing syndromic surveillance and disease detection systems deal with a small portion of a real dataset. From the communication network perspective, the results reported in the literature generally deal with commonly known network topologies. Scalability of a disease detection system is a real challenge when it comes to modeling and predicting disease spread across a large population or large scale networks. In this dissertation, we address this challenge by proposing a hierarchical aggregation approach that classifies a dynamic disease spread phenomena at different scalability levels. Specifically, we present a finite state model (SEIR-FSM) for predicting disease spread, the model manifests itself into three different levels of data aggregation and accordingly makes prediction of disease spread at various scales. We present experimental results of this model for different disease spread behaviors on all levels of granularity. Subsequently, we present a mechanism for mapping the population interaction network model to a wireless mobile network topology. The objective is to analyze the phenomena of malware spread based on vulnerabilities. The goal is to develop and evaluate a wireless mobile intrusion detection system that uses a Hidden Markov model in connection with the FSM disease spread model (HMM-FSM). Subsequently, we propose a software-based architecture that acts as a network function virtualization (NFV) to combat malware spread in IoT based networks. Taking advantage of the NFV infrastructure's potential to provide new security solutions for IoT environments to combat malware attacks. We propose a scalable and generalized IDS that uses a Recurrent Neural Network Long Short Term Memory (RNN-LSTM) learning model for predicting malware attacks in a timely manner for the NFV to deploy the appropriate countermeasures. The analysis utilizes the susceptible (S), exposed (E), infected (I), and resistant (R) (SEIR) model to capture the dynamics of the spread of the malware attack and subsequently provide a patching mechanism for the network. Our analysis focuses primarily on the feasibility and the performance evaluation of the NFV RNN-LSTM proposed model.

The study of shared memory concurrency is extensive. There exist many state-of-the-art strategies for dealing with fundamental concurrency problems, such as race conditions or deadlocks, to leverage massive performance boosts out of modern multiprocessors. With the introduction of blockchain technology as a popular financial tool, we observe many decades-old concurrency problems re-emerge within the context of decentralized networks. These challenges introduce additional constraints, such as the lack of hardware atomic instructions like Compare-And-Swap, or the potential for malicious clients to join the network. In this dissertation, we propose key algorithms which adapt knowledge from the domain of shared memory concurrency to solve emerging concurrency problems in decentralized networks. We propose three key algorithms which further the state of the art in decentralized networks. (1) We present Hash-Mark-Set, a concurrent algorithm for providing a read-uncommitted view of the blockchain state, enabling a higher success rate in transaction use cases where state changes frequently in relation to the block interval. (2) We propose Proof of Descriptor, a descriptor based consensus mechanism for decentralized networks. Proof of Descriptor utilizes well-known techniques from shared memory concurrent programming to create an efficient and scalable algorithm for blockchain consensus. (3) We propose a descriptor-based algorithm for concurrent execution of smart contracts that efficiently captures the concurrent execution as a graph of descriptors, enabling validators to analyze the concurrent execution and verify its results through re-execution.

No description available.

In this dissertation, we propose decentralized and scalable algorithms for Gaussian process (GP) training and prediction in multi-agent systems. The first challenge is to compute a spatial field that represents underwater acoustic communication performance from a set of measurements. We compare kriging to cokriging with vehicle range as a secondary variable using a simple approximate linear-log model of the communication performance. Next, we propose a model-based learning methodology for the prediction of underwater acoustic performance using a realistic propagation model. The methodology consists of two steps: i) estimation of the covariance matrix by evaluating candidate functions with estimated parameters; and ii) prediction of communication performance. Covariance estimation is addressed with a multi-stage iterative training method that produces unbiased and robust results with nested models. The efficiency of the framework is validated with simulations and experimental data from field trials. The second challenge is to perform predictions at unvisited locations with a team of agents and limited inter-agent information exchange. To decentralize the implementation of GP training, we employ the alternating direction method of multipliers (ADMM). A closed-form solution of the decentralized proximal ADMM is provided for the case of GP hyper-parameter training with maximum likelihood estimation. Multiple aggregation techniques for GP prediction are decentralized with the use of iterative and consensus methods. In addition, we propose a covariance-based nearest neighbor selection strategy that enables a subset of agents to perform predictions. Empirical evaluations illustrate the efficiency of the proposed methods / Doctor of Philosophy / In this dissertation, we propose decentralized and scalable algorithms for collaborative multiagent learning. Mobile robots, such as autonomous underwater vehicles (AUVs), can use predictions of communication performance to anticipate where they are likely to be connected to the communication network. The first challenge is to predict the acoustic communication performance of AUVs from a set of measurements. We compare two methodologies using a simple model of communication performance. Next, we propose a model-based learning methodology for the prediction of underwater acoustic performance using a realistic model. The methodology first estimates the covariance matrix and then predicts the communication performance. The efficiency of the framework is validated with simulations and experimental data from field trials. The second challenge regards the efficient execution of Gaussian processes using multiple agents and communicating as little as possible. We propose decentralized algorithms that facilitate local computations at the expense of inter-agent communications. Moreover, we propose a nearest neighbor selection strategy that enables a subset of agents to participate in the prediction. Illustrative examples with real world data are provided to validate the efficiency of the algorithms.

Ad hoc networks and bluetooth systems operating over the unlicensed ISM band are in-stances of decentralized wireless networks. By definition, a decentralized network is com-posed of separate transmitter-receiver pairs where there is no central controller to assign the resources to the users. As such, resource allocation must be performed locally at each node. Users are anonymous to each other, i.e., they are not aware of each other's code-books. This implies that multiuser detection is not possible and users treat each other as noise. Multiuser interference is known to be the main factor that limits the achievable rates in such networks particularly in the high Signal-to-Noise Ratio (SNR) regime. Therefore, all users must follow a distributed signaling scheme such that the destructive effect of interference on each user is minimized, while the resources are fairly shared. In chapter 2 we consider a decentralized wireless communication network with a fixed number of frequency sub-bands to be shared among several transmitter-receiver pairs. It is assumed that the number of active users is a realization of a random variable with a given probability mass function. Moreover, users are unaware of each other's codebooks and hence, no multiuser detection is possible. We propose a randomized Frequency Hopping (FH) scheme in which each transmitter randomly hops over a subset of sub-bands from transmission slot to transmission slot. Assuming all users transmit Gaussian signals, the distribution of the noise plus interference is mixed Gaussian, which makes calculation of the mutual information between the transmitted and received signals of each user intractable. We derive lower and upper bounds on the mutual information of each user and demonstrate that, for large SNR values, the two bounds coincide. This observation enables us to compute the sum multiplexing gain of the system and obtain the optimum hopping strategy for maximizing this quantity. We compare the performance of the FH system with that of the Frequency Division (FD) system in terms of the following performance measures: average sum multiplexing gain and average minimum multiplexing gain per user. We show that (depending on the probability mass function of the number of active users) the FH system can offer a significant improvement in terms of the aforementioned measures. In the sequel, we consider a scenario where the transmitters are unaware of the number of active users in the network as well as the channel gains. Developing a new upper bound on the differential entropy of a mixed Gaussian random vector and using entropy power inequality, we obtain lower bounds on the maximum transmission rate per user to ensure a specified outage probability at a given SNR level. We demonstrate that the so-called outage capacity can be considerably higher in the FH scheme than in the FD scenario for reasonable distributions on the number of active users. This guarantees a higher spectral efficiency in FH compared to FD. Chapter 3 addresses spectral efficiency in decentralized wireless networks of separate transmitter-receiver pairs by generalizing the ideas developed in chapter 2. Motivated by random spreading in Code Division Multiple Access (CDMA), a signaling scheme is introduced where each user's code-book consists of two groups of codewords, referred to as signal codewords and signature codewords. Each signal codeword is a sequence of independent Gaussian random variables and each signature codeword is a sequence of independent random vectors constructed over a globally known alphabet. Using a conditional entropy power inequality and a key upper bound on the differential entropy of a mixed Gaussian random vector, we develop an inner bound on the capacity region of the decentralized network. To guarantee consistency and fairness, each user designs its signature codewords based on maximizing the average (with respect to a globally known distribution on the channel gains) of the achievable rate per user. It is demonstrated how the Sum Multiplexing Gain (SMG) in the network (regardless of the number of users) can be made arbitrarily close to the SMG of a centralized network with an orthogonal scheme such as Time Division (TD). An interesting observation is that in general the elements of the vectors in a signature codeword must not be equiprobable over the underlying alphabet in contrast to the use of binary Pseudo-random Noise (PN) signatures in randomly spread CDMA where the chip elements are +1 or -1 with equal probability. The main reason for this phenomenon is the interplay between two factors appearing in the expression of the achievable rate, i.e., multiplexing gain and the so-called interference entropy factor. In the sequel, invoking an information theoretic extremal inequality, we present an optimality result by showing that in randomized frequency hopping which is the main idea in the prevailing bluetooth devices in decentralized networks, transmission of independent signals in consecutive transmission slots is in general suboptimal regardless of the distribution of the signals. Finally, chapter 4 addresses a decentralized Gaussian interference channel consisting of two block-asynchronous transmitter-receiver pairs. We consider a scenario where the rate of data arrival at the encoders is considerably low and codewords of each user are transmitted at random instants depending on the availability of enough data for transmission. This makes the transmitted signals by each user look like scattered bursts along the time axis. Users are block-asynchronous meaning there exists a delay between their transmitted signal bursts. The proposed model for asynchrony assumes the starting point of an interference burst is uniformly distributed along the transmitted codeword of any user. There is also the possibility that each user does not experience interference on a transmitted codeword at all. Due to the randomness of delay, the channels are non-ergodic in the sense that the transmitters are unaware of the location of interference bursts along their transmitted codewords. In the proposed scheme, upon availability of enough data in its queue, each user follows a locally Randomized Masking (RM) strategy where the transmitter quits transmitting the Gaussian symbols in its codeword independently from symbol interval to symbol interval. An upper bound on the probability of outage per user is developed using entropy power inequality and a key upper bound on the differential entropy of a mixed Gaussian random variable. It is shown that by adopting the RM scheme, the probability of outage is considerably less than the case where both users transmit the Gaussian symbols in their codewords in consecutive symbol intervals, referred to as Continuous Transmission (CT).

Ad hoc networks and bluetooth systems operating over the unlicensed ISM band are in-stances of decentralized wireless networks. By definition, a decentralized network is com-posed of separate transmitter-receiver pairs where there is no central controller to assign the resources to the users. As such, resource allocation must be performed locally at each node. Users are anonymous to each other, i.e., they are not aware of each other's code-books. This implies that multiuser detection is not possible and users treat each other as noise. Multiuser interference is known to be the main factor that limits the achievable rates in such networks particularly in the high Signal-to-Noise Ratio (SNR) regime. Therefore, all users must follow a distributed signaling scheme such that the destructive effect of interference on each user is minimized, while the resources are fairly shared. In chapter 2 we consider a decentralized wireless communication network with a fixed number of frequency sub-bands to be shared among several transmitter-receiver pairs. It is assumed that the number of active users is a realization of a random variable with a given probability mass function. Moreover, users are unaware of each other's codebooks and hence, no multiuser detection is possible. We propose a randomized Frequency Hopping (FH) scheme in which each transmitter randomly hops over a subset of sub-bands from transmission slot to transmission slot. Assuming all users transmit Gaussian signals, the distribution of the noise plus interference is mixed Gaussian, which makes calculation of the mutual information between the transmitted and received signals of each user intractable. We derive lower and upper bounds on the mutual information of each user and demonstrate that, for large SNR values, the two bounds coincide. This observation enables us to compute the sum multiplexing gain of the system and obtain the optimum hopping strategy for maximizing this quantity. We compare the performance of the FH system with that of the Frequency Division (FD) system in terms of the following performance measures: average sum multiplexing gain and average minimum multiplexing gain per user. We show that (depending on the probability mass function of the number of active users) the FH system can offer a significant improvement in terms of the aforementioned measures. In the sequel, we consider a scenario where the transmitters are unaware of the number of active users in the network as well as the channel gains. Developing a new upper bound on the differential entropy of a mixed Gaussian random vector and using entropy power inequality, we obtain lower bounds on the maximum transmission rate per user to ensure a specified outage probability at a given SNR level. We demonstrate that the so-called outage capacity can be considerably higher in the FH scheme than in the FD scenario for reasonable distributions on the number of active users. This guarantees a higher spectral efficiency in FH compared to FD. Chapter 3 addresses spectral efficiency in decentralized wireless networks of separate transmitter-receiver pairs by generalizing the ideas developed in chapter 2. Motivated by random spreading in Code Division Multiple Access (CDMA), a signaling scheme is introduced where each user's code-book consists of two groups of codewords, referred to as signal codewords and signature codewords. Each signal codeword is a sequence of independent Gaussian random variables and each signature codeword is a sequence of independent random vectors constructed over a globally known alphabet. Using a conditional entropy power inequality and a key upper bound on the differential entropy of a mixed Gaussian random vector, we develop an inner bound on the capacity region of the decentralized network. To guarantee consistency and fairness, each user designs its signature codewords based on maximizing the average (with respect to a globally known distribution on the channel gains) of the achievable rate per user. It is demonstrated how the Sum Multiplexing Gain (SMG) in the network (regardless of the number of users) can be made arbitrarily close to the SMG of a centralized network with an orthogonal scheme such as Time Division (TD). An interesting observation is that in general the elements of the vectors in a signature codeword must not be equiprobable over the underlying alphabet in contrast to the use of binary Pseudo-random Noise (PN) signatures in randomly spread CDMA where the chip elements are +1 or -1 with equal probability. The main reason for this phenomenon is the interplay between two factors appearing in the expression of the achievable rate, i.e., multiplexing gain and the so-called interference entropy factor. In the sequel, invoking an information theoretic extremal inequality, we present an optimality result by showing that in randomized frequency hopping which is the main idea in the prevailing bluetooth devices in decentralized networks, transmission of independent signals in consecutive transmission slots is in general suboptimal regardless of the distribution of the signals. Finally, chapter 4 addresses a decentralized Gaussian interference channel consisting of two block-asynchronous transmitter-receiver pairs. We consider a scenario where the rate of data arrival at the encoders is considerably low and codewords of each user are transmitted at random instants depending on the availability of enough data for transmission. This makes the transmitted signals by each user look like scattered bursts along the time axis. Users are block-asynchronous meaning there exists a delay between their transmitted signal bursts. The proposed model for asynchrony assumes the starting point of an interference burst is uniformly distributed along the transmitted codeword of any user. There is also the possibility that each user does not experience interference on a transmitted codeword at all. Due to the randomness of delay, the channels are non-ergodic in the sense that the transmitters are unaware of the location of interference bursts along their transmitted codewords. In the proposed scheme, upon availability of enough data in its queue, each user follows a locally Randomized Masking (RM) strategy where the transmitter quits transmitting the Gaussian symbols in its codeword independently from symbol interval to symbol interval. An upper bound on the probability of outage per user is developed using entropy power inequality and a key upper bound on the differential entropy of a mixed Gaussian random variable. It is shown that by adopting the RM scheme, the probability of outage is considerably less than the case where both users transmit the Gaussian symbols in their codewords in consecutive symbol intervals, referred to as Continuous Transmission (CT).

Resource sharing is a common collaborative strategy used in practice. It has the potential to create synergistic value and leads to higher system efficiency. However, realizing this synergistic value can be challenging given the prevalence of decentralization in practice, where individual operators manage resources based on their own benefits. Hence, optimizing a decentralized system requires understanding not only the optimal operational strategy in terms of the overall system efficiency, but also the implementation of the strategy through proper management of individual incentives. However, traditional network optimization approaches typically assume a centralized perspective. The classic game theory framework, on the other hand, addresses incentive issues of decentralized decision makers, but mainly takes a high-level, economic perspective that does not fully capture the operational complexity involved in optimizing systems with resource sharing. The purpose of this thesis is to bridge this gap between practice and theory by studying the design of tools to manage and optimize the operations in decentralized systems with resource sharing using approaches that combine optimization and game theory. In particular, we focus on decentralized network systems and analyze two research streams in two application domains: (i) implementation of environmental legislation, and (ii) managing collaborative transportation systems. These applications are characterized by their decentralized multi-stakeholder nature where the conflicts and tension between the heterogeneous individual perspectives make system management very challenging. The main methodology used in this thesis is to adopt game theory models where individual decisions are endogenized as the solutions to network optimization problems that reflect their incentives. Such an approach allows us to capture the connection between the operational features of the system (e.g., capacity configuration, network structure, synergy level from resource sharing) and the individual incentives thus the effectiveness of the management tools, which is a main research contribution of this thesis. In the first research stream, we consider designing effective, efficient and practical implementation of electronic waste take-back legislation based on the widely-adopted Extended Producer Responsibility (EPR) concept that mandates the financial responsibility of post-use treatment of their products. Typical implementations of EPR are collective, and allocate the resulting operating cost to involved producers. In this thesis, we demonstrate the complexity of collective EPR implementation due to the tension among different stakeholder perspectives, based on a case analysis of the Washington implementation. We then perform analytical studies of the two prominent challenges identified in current implementations: (i) developing cost allocation mechanisms that induce the voluntary participation of all producers in a collective system, thus promoting implementation efficiency; and (ii) designing collective EPR so as to encourage environmentally-friendly product design, thus promoting implementation effectiveness. Specifically, we prescribe new cost allocation methods to address the first challenge, and demonstrate the practicality and economic impact of the results using implementation data from the state of Washington. We then analyze the tensions between design incentives, efficiency and the effectiveness of the cost allocation to induce voluntary participation under collective EPR implementation. We show there exists a tradeoff among the three dimensions, driven by the network effects inherent in a collective system. The main contribution of this research stream is to demonstrate how the implementation outcomes of an environmental policy is influenced by the way that the policy ``filters' through operational-level factors, and to propose novel and implementation mechanisms to achieve efficient and effective EPR implementation. Hence, our study has the potential to provide guidance for practice and influence policy-making. In the second research stream, motivated by the practice of transportation alliances, we focus on a decentralized network setting where the individual entities make independent decisions regarding the routing of their own demand and the management of their own capacity, driven by their own benefits. We study the use of market-based exchange mechanisms to motivate and regulate capacity sharing so as to achieve the optimal overall routing efficiency in a general multicommodity network. We focus on the design of capacity pricing strategies in the presence of several practical operational complexities, including multiple ownership of the same capacity, uncertainty in network specifications, and information asymmetry between the central coordinator and individual operators. Our study in this research stream produces two sets of results. First, we demonstrate the impact of the underlying network structure on the effectiveness of using market-based exchange mechanisms to coordinate resource sharing and to allocate the resulting synergistic benefit, and characterize the network properties that matter. Second, we propose efficient and effective pricing policies and other mechanism design strategies to address different operational complexities. Specifically, we develop duality-based pricing algorithms, and evaluate different pricing strategies such as commodity-based price discrimination, which is shown to have an advantage in coordinating networks under uncertainty.

Blockchains, smart contracts and cryptocurrencies are reaching further global adoption each day and are becoming more prominent to be the foundation for a new technological era and to be used in transactions globally. The technology has strong coherent properties, including a high level of security, decentralization, and its use of smart contracts to obviate intermediaries. These technologies offer the possibility to do any transaction without a centralized bank controlling, denying, or reversing the transaction. This report aims to shed light on blockchain technologies possible impact on society and if the current centralized-based system can be replaced or if it is deemed as necessary. What the potential outcome would be if these centralized authorities sees a decrease in power and how blockchains, smart contracts and stablecoins can be used as an everyday payment and transaction system, excluding all third parties. To do this, a literature review, a quantitative survey, and qualitative interviews were conducted. The literature review, to lay the ground for the questions in the survey, interviews, and additionally, to determine what blockchain and what kind of stablecoin is most suited for global adoption. The result from the qualitative interview were to acquire knowledge from more experienced individuals working or owning a company that is based on top of blockchains. Lastly, considering blockchain technology and web 3.0 is still under development and yet to become globally accepted, the questionnaire survey was conducted to retrieve the general consensus from individuals inside crypto communities. The report overall, came out to be successful, indicating that blockchain technologies has a bright future and that the decentralization it adds to society can benefit the individual in multiple of different ways, specifically in the financial sector. However, although the conclusion that blockchain technologies can be used as a global payment system were deemed successful, objective, and subjective opinions were discussed and reported relevant, whereas one individual might view the exclusion of intermediaries as necessary and beneficial, while the other – as something negative. Scalability issues in blockchains and smart contracts controversy, such as its complexity and its immutability aspect are also analyzed and discussed as a potential hindrance for further global adoption.

El presente artículo explora la disputa vigente sobre el principio queindica que se requiere el consentimiento libre, previo e informado(CLPI) de un pueblo indígena para autorizar la extracción industrialen su territorio. A partir de los aportes de la obra de Tsing acerca delas conexiones globales, el trabajo analiza cómo es que los actoresinterconectados en redes se empeñan en llevar adelante ampliosproyectos de colaboración (como el reconocimiento de los derechosindígenas) empleando estrategias de persuasión. Se discutenlos esfuerzos realizados por el movimiento indígena transnacionalpara promover el concepto del CLPI, así como tres ejemplos en losque diferentes actores buscan apropiarse y recaracterizar el CLPIpara que calce en sus propias metas.En este trabajo propongo examinar cómo los proyectos gubernamentalesglobales rivales son promovidos y disputados por las redesdescentralizadas que unen a actores que operan a diferentes escalas.Sostengo que la noción de Tsing de «paquetes itinerantes» ofreceuna manera útil de conceptualizar los medios por los cuales loselementos de estos proyectos son difundidos, traducidos, acogidosy adaptados en diferentes localidades alrededor del mundo. Analizoestas dinámicas en relación con el cuestionamiento al modelo degobernanza basado en el principio de que se necesita el consentimientolibre, previo e informado (CLPI) de un pueblo indígenapara autorizar acciones que puedan impactar sobre un territorio o derechos indígenas. A través de la promoción de diferentes versionesde CLPI, los actores interconectados en red están disputandola naturaleza y la forma de la responsabilidad social empresarial,la autoridad del Estado y la relevancia de la gobernanza indígena.Propongo explorar las implicaciones de las diferentes estrategias deempaquetamiento para la disputa entre modelos rivales de gobernanzay para su propensión a ser acogidos en los sitios locales. / In this paper, I propose to examine how rival global governmentalprojects are asserted and contested by decentralized networks thatlink actors operating at different scales. I argue that Tsing’s notionof «travelling packages» provides a useful way of conceptualizingthe means by which elements of these projects are diffused, translated,taken up, and adapted into different localities around theworld. I explore these dynamics in relation to the contestation of agovernance model based on the principle that the free, prior andinformed consent (FPIC) of an indigenous people is required toauthorize actions that may affect upon indigenous territory or indigenousrights. Through the assertion of different versions of FPIC,networked actors are contesting the nature and shape of corporatesocial responsibility, the authority of the state, and the significanceof indigenous governance. I propose to explore the implicationsof different packaging strategies on the contestation between rivalgovernance models and on their propensity for uptake in local sites.

Les communautés virtuelles sont des systèmes sociotechniques dans lesquels des entités (humaines et/ou artificielles) répartis à travers le monde se réunissent autour d’intérêts et/ou d’objectifs communs. Afin de réaliser ces objectifs, les membres de la communauté doivent collaborer en partageant leurs ressources et/ou connaissances. Or, toute collaboration comporte une part de risque dans la mesure où les membres peuvent se comporter de manière non coopérative ou malveillante. Dans de tels contextes, où les mécanismes de sécurité standard ne suffissent plus, la confiance est rapidement devenue un facteur déterminant lors de la prise de décision. Le travail présenté dans cette thèse s’attaque à la problématique de la gestion de la confiance dans les communautés virtuelles ouvertes et décentralisées. Pour cela, nous avons proposé une infrastructure de gestion de la confiance adaptative et conforme socialement (ASC-TMS). L’aspect novateur de ce système réside dans sa faculté à exhiber des propriétés sociales et adaptatives. L’aspect social du ASC-TMS fait référence à la capacité de notre système à prendre des décisions qui soient sûres non seulement pour l’individu mais également et surtout pour les autres membres de la communauté. Par ailleurs, l’aspect adaptatif du système fait référence à la capacité du système à prendre des décisions qui soient en parfaite adéquation avec l’environnement dans lequel ces décisions sont prises. Ainsi, cette thèse constitue une nouvelle étape vers l’automatisation de l’évaluation de la confiance en assistant les membres des communautés virtuelles ouvertes et décentralisées dans leur prise de décision. Le système a été implémenté et déployé en utilisant la plateforme de développement multi-agent JaCaMo. Nous avons également illustré l’applicabilité de notre approche sur un scénario réel de communauté virtuelle d’innovation ouverte. Enfin, nous avons évalué notre système expérimentalement en utilisant la plateforme de simulation multi-agent Repast. Les résultats obtenus montrent que l’utilisation de notre système avait un impact positif sur la dynamique des communautés dans lesquels il est a été utilisé. / Virtual communities (VCs) are socio-technical systems wherein distributed individuals (human and/or artificial) are grouped together around common objectives and goals. In such systems, participants are massively collaborating with each other’s by sharing their private resources and knowledge. A collaboration always bears the risk that one partner exhibits uncooperative or malicious behaviour. Thus, trust is a critical issue for the success of such systems. The work presented in this dissertation addresses the problem of trust management in open and decentralised virtual communities (VCs). To address this problem, we proposed an Adaptive and Socially-Compliant Trust Management System (ASC-TMS). The novelty of ASC-TMS lies in its ability to exhibit social-awareness and context-awareness features. Social-awareness refers to the ability of the trust management system (TMS) to handle the social nature of VCs by making trust evaluations that are collectively harmful, while context-awareness refers to the ability of the system to handle the dynamic nature of VCs by making trust evaluations that are always in adequacy with the context in which these evaluations are undertaken. Thus, the contributions made in this thesis constitute an additional step towards the automation of trust assessment. We provided accordingly a novel trust management system that assists members of open and decentralised virtual communities in their trust decisions. The system has been implemented and deployed using the JaCaMo multi-agent platform. We illustrated also the applicability of on a real life open innovation virtual community scenario. Finally, the ASC-TMS has been experimentally evaluated using the multi-agent based Repast simulation platform. The preliminary results show that the use of our system significantly improves the stability of the virtual communities in which it has been deployed.