Search

Secure group communication in an ad hoc network is a largely unexplored research area. Currently available key exchange protocols were not designed to be implemented in an ad hoc network where nodes sporadically enter and leave the group. This project explores establishing secure group communication in an ad hoc network through public key infrastructure. Public key infrastructure (PKI) provides a framework for establishing and authenticating secure communication between users. A trusted certificate authority (CA) generates an identifying token, or certificate, for an authorized user. The certificate contains the user's public key and other identifying information and is digitally signed by the CA to prevent forging. This public key may then be used to initiate secure communication with the user. This project uses the tuple space distributed computing paradigm for all ad hoc group communication. A tuple space is a store of tuples, or lists of objects, from which consumers may read tuples matching filter criteria and to which producers may post new tuples. An easily made physical analogy to this concept is that of an announcement board, where people may read flyers and post new ones. Professor Alan Kaminsky's TupleBoard API is an implementation of tuple space designed for developing ad hoc distributed applications in Java. This project extends this library by adding a public key framework enabling dynamic group key exchange, public key encryption and digital signatures. To showcase the newly added security features an ad hoc music distribution application was developed in which all communication is encrypted and authenticated and users may only share or download songs authorized by certificates in their possession. Finally, a performance analysis was done to evaluate the impact of the new security features.

The tuple board is a new distributed computing paradigm based on the tuple space concept invented by Gelernter in 1985. Designed for ad hoc mobile wireless networks, tuple board is a serverless architecture that differs from the original client-server paradigm proposed in tuple space. In this project, Tuple Board, a tuple board implementation has been developed based on the M2MP protocol. Comparing to the previous implementation in Bondada's work, the following improvements have been made: (1) M2MP, instead of M2MI, is used as the underlying network protocol for data communication, (2) notification is available in the tuple board implementation for tuple status changes, (3) a JavaSpaces Entry-like approach is used rather than using Java Class/Object arrays to represent tuple objects. To further study the strength and the limitations of Tuple Board, a collaborative photo sharing application has been developed. The application allows users in ad hoc mobile networks to share selected photos and view photos posted by other users based on dynamically defined rules. Photos available in the application can be shared or withdrawn by the owner.

In this project, we introduce a new distributed computing paradigm called "tuple board" which is based on the tuple space abstraction originated by Gelernter and the blackboard architecture popular in distributed artificial intelligence. Most prominent implementations of tuple space systems including JavaSpaces and T-Spaces follow a centralized architecture where the space itself resides on a server, akin to a database server. Recently, researchers have attempted to develop decentralized architectures for tuple space systems. Some work includes LIME and PeerSpaces. In our opinion, the traditional implementation of tuple spaces is not well suited for ad hoc networks where devices frequently lose network access. The tuple space architecture attempts to provide persistent storage of tuples irrespective of the state of nodes in the network. While persistence may be easily implemented in centralized systems such as JavaSpaces, achieving this end in a purely decentralized environment would require complex replication schemes. This would place undue overhead on the resource-constrained devices that are typical for ad hoc networks. In the architecture used in this project the availability of tuples is determined by the state of devices participating in the network. At any given point of time, only the tuples contained in devices that are active in the network are available on the tuple board. As part of the project, a specific instance of the tuple board architecture was realized using the Anhinga Infrastructure. The system, called Anhinga Board, uses the M2MI protocol for the implementation of tuple board semantics. To demonstrate the Anhinga Board, a collaborative conference information system, consisting of tuple board based applications running on attendees devices, vendors devices, and the conference center s devices, was also developed.

The Generative Communication (GC) paradigm was introduced in the early 1980s in the form of the Linda coordination and communication language (Gelernter & Bernstein, 1982). The goal of the language was to provide an elegant, intuitive alternative to the state-of-the-art in parallel and distributed processing languages at that time, namely remote procedure calls; message passing systems; and shared memory systems (Gelernter, 1985). Despite the longevity of the Linda coordination language and the GC paradigm, in general, the literature is devoid of meaningful research into the construction and manipulation of dynamic, pointer-based data structures within the context of the GC paradigm. This document motivates the need for such data structures in GC. In addition, this document explores the reasons for the lack literature regarding dynamic pointer-based data structures in GC. Finally, we develop solutions to address the lack of support, within the GC paradigm, for such data structures.

A computational grid is a framework for resource sharing across multiple organizations, operating in different administrative domains. Grids are built over wide-area networks. They provide infrastructure for wide-area distributed computing platform. Grids provide secure and fine-grained access to computational resources in multiple (virtual) organizational domains. Tuple space is a network-global storage space providing a coherent-addressable shared memory abstraction. Tuple space provides a logically shared address space and enables data sharing. Grids can be used for building computing platforms that can span across wide-area networks, grid-computing platform. Tuple spaces enable distributed programming and are decoupled in space and time. Communication in tuple space is asynchronous. Tuple spaces provide data sharing ability and coordinate access to data. The purpose of this research is to explore distributed programming in context with grid-computing platform. The idea is to introduce tuple space based programming for grids. This thesis looks at the design of a grid-enabled tuple space. A prototype of grid-enabled tuple space has been implemented. Globus, an open source toolkit for grids is used as grid computing toolkit. JavaSpace is a Java-based implementation of tuple space. JavaSpace is built on grid communication infrastructure to develop a 'Grid-enabled JavaSpace'. Grid-enabled JavaSpace is the prototype for grid-enabled tuple space.