Ellen Munthe-Kaas

A Data Sharing Facility for Mobile Ad-Hoc Emergency and Rescue Applications

Efficient information sharing is very important for emergency and rescue operations. Mobile ad-hoc Networks (MANETs) are often the only network environment for such operations. We have developed the MIDAS data space (MDS) to transparently share information among rescue applications in such environments. To achieve the required level of availability for important information, MDS performs optimistic replication. The problems caused by optimistic replication, like consistency management, are not solved by standard solutions; instead we employ tailor-made solutions for emergency and rescue applications.

A Conceptual Service Discovery Architecture for Semantic Web Services in Dynamic Environments

Web Services technology is being used for increasingly different environments than it was designed for. To facilitate discovery of Web Services in dynamic environments, both service description and distribution of descriptions must be improved. Several research efforts target semantic description of services. However, Semantic Web Service discovery in peer-to-peer-like, dynamic environments where services and registries are transient cannot be based on current mechanisms for distribution of Web Service descriptions. Based on a set of generic requirements, we introduce a conceptual architecture that aims to solve many of the problems related to Semantic Web Service discovery in dynamic environments. The architecture is based on a distributed multi-registry topology and aliveness information. We present several research problems we have identified during our initial work on this architecture.

Metadata management for ad-hoc infoware – a rescue and emergency use case for mobile ad-hoc scenarios

In this paper we present a multi-tier data dictionary approach to metadata management in mobile ad-hoc networks (MANETs). Our work is part of the Ad-Hoc InfoWare project, aiming to find middleware solutions to information sharing in MANETs used in rescue and emergency scenarios. Such scenarios are characterised by highly dynamic and time critical issues, and pose new challenges to solutions for information sharing and integration. We present the rescue and emergency scenario, the results of the requirements analysis, and some use cases resulting from this analysis. We describe the design of a knowledge manager component and how it may provide solutions, the main focus being on metadata management and the use of global and local data dictionaries. We present the status of our work, including detailed design and ongoing implementation of a data dictionary manager.

A Top Down Approach to the Formal Specification of SCI Cache Coherence

The paper gives an introduction to an ongoing effort of formally specifying and verifying the cache coherence protocol of the new IEEE interconnect standard called the Scalable Coherent Interface. We first give the the most abstract (top level) specification of a memory system. We then introduce a private cache for each processor, and specify the notion of cache coherence. We refine the specifications of the memory operations for use with caches, and finally outline the more complex bottom layers where directory structures and concurrency are introduced.

Detecting communities in sparse MANETs

In sparse mobile ad hoc networks, placement of services and data is crucial to assure their availability to all nodes because sparse population of nodes can lead to (frequent) network partitions. If these dynamic networks display a fairly stable cluster structure, it is possible to utilize this structure to improve service and data availability. However, clustering in a dynamic network is a very challenging task due to the ever-changing topology and irregular density of such a network. In this paper, we investigate clustering of dynamic networks with the help of community detection mechanisms, using only topology information from the local routing table. The main aim of our approach is to reduce to “zero” the communication overhead needed for cluster management and to dynamically adapt to the size and layout of the network. We have performed extensive experiments to evaluate the consistency, quality, and stability of the clustering returned by our algorithms. The results show that our nonintrusive clustering indeed discovers temporary groups of nodes that form stable clusters in the network. Moreover, even though the local routing tables in general reflect slightly diverging topologies, our results still show only small differences between the communities detected at different nodes.

Building resource aware middleware services over MANET for rescue and emergency applications

In the ad-hoc InfoWare project we investigate middleware solutions for mobile ad-hoc networks (MANETs) used in emergency and rescue situations. One goal in the project is to increase availability of services and information through resource awareness. In this context we regard the ability to predict with a certain confidence the future resource availability as very helpful. The work described in this paper, presents our current progress towards prediction of future connectivity between nodes using information about a nodes neighborhood extracted from the routing protocol. The simple idea is that if two nodes have been close to each other for a certain time, there is a high probability that they will be close also in the immediate future. The advantage with our approach is that the nodes do not need location information. We are using simulation to analyze for how far into the future we can make valid predictions

Incremental fault-tolerant design in an object-oriented setting

With the increasing emphasis on dependability in complex, distributed systems, it is essential that system development can be done gradually and at different levels of detail. We propose an incremental treatment of faults as a refinement process on object-oriented system specifications. An intolerant system specification is a natural abstraction from which a fault-tolerant system can evolve. With each refinement step a fault and its treatment are introduced, so the fault-tolerance of the system increases during the design process. Different kinds of faults are identified and captured by separate refinement relations according to how the tolerant system relates to abstract properties of the intolerant one in terms of safety, and liveness. The specification language utilized is object-oriented and based upon first-order predicates on communication traces. Fault-tolerance refinement relations are formalized within this framework.

Supporting multiple subscription languages by a single event notification overlay in sparse MANETs

The subscription language is an important design decision for distributed event notification services (DENS). In order to minimize resource consumption and enable applications to use rich and complex subscription languages only when they are really needed, we have developed a DENS that separates the concerns of delivering subscriptions and notifications from the subscription specification and event filtering, i.e., the subscription language. To resolve the conflict between subscription language independence in DENS and a strict decoupling of publishers and subscribers through the DENS, we request that for each new subscription language three language specific plug-ins are provided. In this paper, we present the technical details of this solution and describe our proof-of-concept implementation that supports a simple attribute-value based subscription language and a fuzzy concept-based language.

Middleware Services for Information Sharing in Mobile Ad-Hoc Networks

Information sharing is a mission critical key element in rescue and emergency operations. Mobile ad-hoc networks (MANETs) could provide a useful infrastructure to support information sharing, but appropriate applications are needed. To facilitate efficient application development for this type of infrastructure, middleware support is needed. In the Ad-Hoc InfoWare project, we are currently developing corresponding middleware services. In this paper, we discuss the application requirements that are imposed onto the middleware services, and we outline our technical approach to address the corresponding challenges. The architecture we propose comprises five main building blocks, namely knowledge management, a local and a distributed event notification service, resource management, and security and privacy management. We indicate design alternatives for these building blocks, identify open problems and relate our approach to the state-of-the-art.

A linked list cache coherence protocol: verifying the bottom layer

High performance and correctness are crucially important in hardware development. This paper discusses aspects of correctness of a linked list cache coherence protocol. The protocol is tailored to execute as many operations as possible in parallel, achieving speed and avoiding the memory bottleneck. This makes it hard to verify that the protocol is correct. In the paper, a bottom layer of the protocol is identified and it is shown how the correctness of this layer can be established.<<ETX>>