SP
From AEOLUS IST Project
Project AEOLUS IST/FET Project is structured into six subprojects.
SP1 will be devoted to the development of "innovative theories" to cope with new algorithmic problems that arise in Global Computing. It will study the structural properties of global/overlay computers, fundamental techniques for coping with selfishness and for achieving stability and fault tolerance, and will tackle the challenge of computing with partial (i.e., uncertain, distributed, or even incomplete) knowledge by blending theories from economics, game theory and algorithmic theory. A better understanding of these problems will have a strong impact on the ability to propose scalable, distributed and dynamic algorithms. That will also allow understanding the efficiency trade-off between undesirable centralized strategies and anticipated fully distributed strategies.
The work within this subproject focuses on the study of fundamental issues for accessing and managing communication resources in an overlay computer. Our research will address novel and challenging algorithmic issues for efficient resource discovery and querying like construction of overlay networks, query routing and execution, and for sharing critical resources like bandwidth. Our work will also include mechanism design for coping with selfish behavior when allocating resources in a distributed, uncoordinated system such as a global or overlay computer.
The work within this subproject focuses on the study of fundamental issues for organizing and accessing information in overlay computers and for controlling the computing power of their nodes. Novel algorithmic issues in distributed data management including caching and replication of primitive or more complex data (e.g., metadata), load management including load balancing and tuning and parameterization of adaptive software, and scheduling motivated by the size and the dynamic nature of overlay computers will be addressed, while we will also attempt to model intensive computations in overlay computer as processes in workflow management systems.
Given the conceptual complexity of the global computer environment, one would like to abstract away this layer of complexity and to enable developers and software architects to focus only on the application to be developed. The overlay computer will instead transparently provide the security functionalities needed for the application to be securely run in a global environment. As an added benefit, functionalities provided by the overlay computer will not have to be duplicated in each application.
For a concrete example (that will be discussed in more detail in the following sections) suppose that a distributed application for mining knowledge from several proprietary data sets is to be developed. The overlay computer will enable the application developer to concentrate on the design of the application as if privacy of the data base were not an issue. It will be the task of the overlay computers to wrap the application in such a way that privacy of the data sets is not compromised. Moreover, the security must be guaranteed even when the distributed data mining application is run concurrently with several others applications working on the same data sets. At the same time, to be of any use, the overlay computer must offer scalable and robust implementations of the functionalities. It is thus important to concentrate research efforts on the efficiency of the underlying protocols and algorithms and to develop computational and algorithmic theories that describe and model security issues in the new conceptual scenario of a global computer.
WP 4.0: "Subproject Management and Dissemination Activities"
The goal of this workpackage is to guarantee the successful progress of the subproject within the agreed time, cost and quality limits as defined by the project contract signed with the EU and the Consortium Agreement signed between the partners. This workpackage will also deal with establishing effective communication among the consortium partners, as well as the effective dissemination of subproject results, management of intellectual property rights and patent applications.
WP 4.1: "Trust Management"
One major objective is to understand the trade-off between expressibility of the language used to specify the policy and efficiency of compliance checking algorithm. The research in this workpackage will also develop links with the research conducted in SP1/WP2 (Coping with selfishness) so to blend game theoretic techniques and authorization techniques to obtain reputation-based authorization schemes.
We plan to study multi-party trust negotiation specifically addressing issues like extensions of languages for trust negotiation and deadlock detection as other specifically related to multi-party negotiations arising from highly mobile users.
We will approach the compliance checking problem from a computational point of view. We will evaluate a broad set of options between two extremes: the most general language (e.g., Turing machines or any Turing equivalent language) and policies specified using ACL.
We will study reputation-based approach under the assumptions that the users will respond to economic incentives. This calls for a blending of techniques from Game Theory and from Security. We will interact with researchers working in WP1.2 to develop a model for reputation-based authorization and to design schemes that are resilient to attacks. Also in cooperation with WP4.2, we will investigate privacy preserving mechanisms for use in trust negotiation. Ti this aim we will rely on a reference ontology, and formalize the notion of trust requirement. We will also address privacy issues allowing a subject to adopt strategies to make the set of credentials/attributes he/she is going to release privacy preserving.
We will also explore the use of “local secure distributed computation” within the context of reputation-based systems.
We aim to define an access control model suitable for large scale scenario. We envision that such model will be characterized by rich description of subjects and objects, so that high-level policies can be directly expressed in terms of subject and object properties. Ontologies and federated identity management will be exploited to support multi-domain interoperability. Context-based access decisions will also be supported by the model.
Also, we will approach the problem of multi-party trust negotiations and models and mechanism for access control.
WP 4.2: "Privacy, Identity and Anonymity"
To develop efficient, scalable and secure solutions for anonymous communications and anonymous transactions for a global computing environment.
An overview will be made of proposals for technical mechanisms for privacy, identity and anonymity; this includes both mechanisms at the network level (such as mixes and crowds) and mechanisms at the application level such as anonymous credentials.
For the network level techniques, quantitative statistical models will be developed and implemented to model and analyze the behavior of the mechanisms and to improve new building blocks. This will require statistical analysis for more complex mix models. The goal is to achieve a deep insight in the trade-off between quantitative measures of anonymity and cost and quality of service. The security analysis will be improved by considering more realistic attack models. In addition, more flexible and feature-rich techniques will be developed, that include forward secrecy, congestion control and adaptivity. Part of this work will be undertaken in close collaboration with SP2.
For anonymous credentials, research will be performed into more efficient and scalable solutions, e.g. by building on solutions and techniques developed in WP4.3 and by integrating these into other protocols. This will focus on a comparison of different cryptographic techniques and on the development of novel credential based mechanisms based on well established and on novel cryptographic problems. Moreover, theoretical limitations (lower bounds) will be investigated.
WP 4.3: "Secure Distributed Computation"
To develop efficient and proactively secure protocols for applications relevant to global computing such as privacy-preserving data mining, secure algorithmic mechanisms for route discovery in networks and electronic voting.
Thanks to the completeness results for secure distributed protocols, it is known how to compute in a secure way any efficiently computable function; however, the protocols obtained through the application of the completeness result are not practical and are only secure if executed as stand-alone protocols.
The research of this workpackage is articulated along several axes. Along a foundational line of research, we will extend the theory of secure distributed protocols so to be able to formally describe and analyze secure distributed protocols for global computers. Possible lines of research include the design of communication-efficient protocols for computing any (efficiently) computable function that can be concurrently composed or even universally composed (currently, similar general results are only known under strong set-up assumptions or for stand-alone security).
Along a more immediate line of thought, we will design efficient secure protocols for important applications among which we list privacy-preserving data mining (see also WP4.2), secure algorithmic mechanisms for route discovery in networks and electronic voting. Within the context of proactivly secure protocols we will adopt the graph-theoretic model of the pursuit-evasion distributed game and develop efficient distributed strategy to efficiently locate all the clones of a worm that is attacking a network. This entails both modeling the spreading of the worm and designing efficient distributed algorithms to coordinate the work of the antivirus mobile agents. For the more efficient protocols, prototype implementations will be provided.
At the intersection of the two lines of research lies the problem of defining universally composable security for time-stamping schemes and whether there are efficient and practical constructions for universally composable time-stamping schemes.
The main objective of the subproject is to provide practically efficient algorithmic solutions for high quality, reliable, stable end-users services to heterogeneous wireless mobile networks; this will be necessarily accomplished by appropriately abstracting over the common characteristics of the diverse component devices. In turn, the provision of such algorithmic solutions requires facing a number of strongly related issues that can be grouped in the following categories, directly corresponding to the workpackages of this subproject.
WP 5.0: "Subproject Management and Dissemination Activities"
The goal of this workpackage is to guarantee the successful progress of the subproject within the agreed time, cost and quality limits as defined by the project contract signed with the EU and the Consortium Agreement signed between the partners. This workpackage will also deal with establishing effective communication among the consortium partners, as well as the effective dissemination of subproject results, management of intellectual property rights and patent applications.
WP 5.1: "Resource Management and Quality-of-Service (QoS)"
We will study range assignment problems with specific properties. We also aim improving the adherence of existing energy cost models to real features of wireless networks. We will investigate the existence of feasible, approximation truthful mechanisms for range assignment problems, that is, truthful mechanisms that compute in polynomial-time an approximate solution. Another important goal is the design of efficient collision avoidance protocols in sensor networks. Energy efficient routing shall also be considered during the first 18 months.
WP 5.2: "Dynamical Aspects of Network Design and Topology Control"
The main goals of the first 18 months are (i) to analyze and handle topologies as basic network structures for overlay computing; (ii) to well understand the impact of the shape of the transmissions on the communications; (iii) to improve the understanding of the collisions. Concerning the shape of the transmissions, in [BLR05] a stochastic method [RM51] was implemented in order to compute numerically some critical parameters of the networks to ensure a particular task to be possible. For example, considering the process of localization in wireless networks, it is numerically shown that increasing the angle of emission increases the chance of success of the process. The most relevant observation is the existence of a critical value of the angle of emission under which the process almost surely fails and above which almost surely succeeds. The process then shows a phase transition when the angle of emission increases.
WP 5.3: "Mobility and Fault Tolerance"
We plan here to model and characterize mobility and fault tolerance in scalable, heterogeneous wireless networks.
The work in this subproject will serve as a "proof-of-concept" for the whole project. Our main goal is to develop a programmable Overlay Computing Platform based on the scientific advances of the project; this will be an overlay computer built on top of a global computer consisting of Internet-connected nodes. Among the functionalities that will be separately investigated in the other subprojects, the Overlay Computing Platform will include those which are important for the execution of distributed applications with intensive requirements for efficient use of computing, communication and information resources and security. To demonstrate the programmability of the platform, we also plan to implement an application on top of it. In parallel, for the validation of the scalability of both the platform and the application, we will also develop and setup an appropriate testbed environment.