The Galerkin method is studied for solving the boundary integral equations associated with the Laplace operator on nonsmooth domains. Convergence is established with a condition on the meshsize, which involves the local curvature on certain approximating domains. Error estimates are also proved, and the results are generalized to systems of equations.

Congestion control in signaling systems is a necessity to fulfil the requirements of a tele-communication network that aims satisfy the customers’ requirements on service quality. Heavy network load is an important source of customer dissatisfaction as congested net-works result in deteriorated service quality. Sessions of a signaling service with high real time demands which are subject to unacceptable delays may be obsolete or prematurely terminated by the customer; in either way, they are a burden to the signaling network. It would ease the load of the network and improve the performance of all sessions in progress, if such delayed sessions could be aborted as quickly as possible. By measuring the network delay on individual signals of a service session, it is possible to perform signaling network congestion control that considers the state in the entire signaling network. Under the assumption that a session comprises a sequence of signals between one originating node and an arbitrary number of destination nodes, it is possible to predict the total duration of a session. The prediction is calculated from previously completed signals using a state machine, which is defined per signaling link. The annihilation of sessions, for which the prediction exceeds a predefined time limit, is an embryo of a simple signaling network congestion control mechanism (CCM). This simple CCM increases the number of successfully completed services with several hundred percent under favorable circum-stances. The state machine approach is proven to perform well in all types of environ-ments. The robustness and stability of the proposed CCM is demonstrated in a wide range of environments. The fairness in the admission of signaling services into the network at very high loads are also shown to be good.

Congestion control in Signaling System number 7 faces new challenges as mobile com-munication systems and Intelligent Networks grow rapidly. New services change traffic patterns, add to signalling network load, and raise demands on shorter service completion times. To handle new demands, the congestion control mechanisms must foresee an over-load situation, and respond to it so that the network can maintain high throughput. With the introduction of a state machine and a memory function for each signaling link it is possible to predict the completion time of a service session and to detect an emerging congestion. If the predicted completion time of a service session is too long, the session is annihilated. This is the foundation of a congestion control mechanism that reacts fast on information supplied by the network. The congestion control mechanism enables the network to main-tain a high throughput during overload.

Congestion control in Signaling System #7 faces new challenges as mo-bile communication systems and Intelligent Networks grow rapidly. New ser-vices change traffic patterns, add to signalling network load, and raise demands on shorter service completion times. To handle new demands, the congestion control mechanisms must foresee an overload situation, and respond to it so that the network can maintain high probability for successful service comple-tion. With the introduction of a state machine and a memory function for each signaling link it is possible to predict the completion time of a service session and to detect an emerging congestion. If the predicted completion time of a service session is too long, the session is annihilated. This is the foundation of a congestion control mechanism that reacts fast on information supplied by the congested part of the network. The congestion control mechanism increases the ratio of successfully completed service sessions during congestion by several hundred percent.

Sessions of a signaling service with high real time demands which are subject to unaccept-able delays may be obsolete or prematurely terminated by the customer; in either way, they are a burden to the signaling network. It would ease the load of the network and im-prove the performance of all sessions in progress, if such delayed sessions could be abort-ed as quickly as possible. By measuring the network delay on individual signals of a service session, it is possible to perform signaling network congestion control that consid-ers the state in the entire signaling network. Under the assumption that a session comprises a sequence of signals between one originating node and an arbitrary number of destination nodes, it is possible to predict the total duration of a session. The prediction is calculated from previously completed signals using a state machine, which is defined per signaling link. The annihilation of sessions, for which the prediction exceeds a predefined time limit, is an embryo of a simple signaling network congestion control mechanism (CCM). This simple CCM increases the number of successfully completed services with a few hundred percent under favorable circumstances. The state machine approach has been proven to function well in all types of environments. The robustness and stability of the proposed CCM is demonstrated and the fairness in the admission of signaling services into the net-work at very high loads are also shown.

Congestion control in Signaling System #7 faces new challenges as mobile communication systems and Intelligent Networks grow rapidly. New services change traffic patterns, add to signalling network load, and raise demands on shorter service completion times. To handle new demands, the congestion control mechanisms must foresee an overload situation, and respond to it so that the net-work can maintain high probability for successful service completion. By measur-ing the time consumption for the initial Message Signalling Units of a service session it is possible to predict the duration of the service session and to detect an emerging congestion. If the predicted duration of the service session is too long, the service session is annihilated. This is the foundation of a congestion control mechanism that reacts fast and on information supplied by the congested part of the network. The congestion control mechanism increases the ratio of successfully completed services during congestion by several hundred percent.

We study real time traffic management in the perspective of a four-level model of traffic variations: cells, bursts, sessions, and links. Our interest is focused on the two latter levels: Using virtual channels (VCs) and virtual paths (VPs) as the managed entities on the two levels respectively, we put forward a simple and robust strategy for traffic management in real time: All nodes simultaneously monitor offered traffics and forward the result to a network management centre (NMC); The NMC computes a new set of optimal VPs and compares the result to the existing network; If a change appears profitable, the necessary information is sent back to the nodes and the new design is implemented. Applying our strategy to control general networks subject to traffics which in advance are known only as expectations, we develop simple methods for traffic observation, the optimal observation interval, a fast and efficient algorithm to compute VP networks, a test to determine whether to implement a new design in which costs related to lost traffic are compared to processing costs, and the optimal updating interval. Running our strategy in a simulator, we discover a considerable adaptivity, large savings in transmission capacity, and a performance far better than achieved by fixed redesign according to forecasts. Finally we discuss limits to the range of variations within which our strategy can be applied. The present work applies both to networks using SDH/SONET and/or networks based on ATM, i.e. both to present public switched telephony networks and broadband inte- grated services digital networks, and is a summary and extension of earlier works within this area [3{9].

We study dynamically rearrangable end-to-end transport networks by means of SDH/SONET and/or ATM virtual paths. The concept and its advantages are brie y reviewed and we propose a new algorithm for designing such networks. Our algorithm provides integer solutions, handles non-linear equivalent bandwidths, finds its own paths, and allows for an arbitrary design optimisation criterion. It is found to perform equal to or better than an earlier, comparable though more restricted, algorithm. We also put forward a strategy for managing virtual paths in real time. It consists of a repeated cycle of on-line measurement of offered traffics followed by the design and implementation of a new virtual path network. Methods and parameters for traffic measurements and network updating frequency are studied and optimised. Applying the strategy to rearrangeable networks subject to variable traffics in a simulator, we discover considerable gains and a robustness in our parameter settings.

We consider call admission and capacity management in a dynamically reconfigurable network that carries a multitude of traffics. Our approach is a heuristic capacity as- signment algorithm designed to maximise an arbitrary profit function. The algorithm is based on the concept of equivalent circuits, i.e. a given relationship between transmis- sion capacities and call handling capabilities expressed as effective numbers of circuits. Such relationships are examined for the two traffics of our application, voice and frame relaying. Methods and results are independent of the transfer mode used.

Traffic in ATM networks can be described by numerous parameters. On a per session basis, one may use peak rate, average rate, sustainable rate, average burst duration, average silence duration, and others. In a longer time scale, parameters like the average and peakedness of connection request interarrival times, the average and variance of session holding times, and so on are proposed. Theoretically, users should provide the former parameters and network operators the latter. In reality, however, few users can be expected to provide all sorts of statistical information about their traffic in advance, and operators do not have enough experience to prepare traffic forecasts for new services and applications. Moreover, even if the information could be provided, the lack of simple yet valid traffic models for ATM networks means that it is far from clear how such information should be used. Realising that ATM networks, which are already being built, at least for the next few years will have to operate under these uncertain conditions, we propose a robust and forgiving network design and traffic management strategy. The idea is to use only little information about offered traffics and then dynamically control resource allocations, so as to provide acceptable quality of service combined with high utilisation. The network design is based on the idea of keeping congestion at the edges of the net- work, so that the operating areas of fast congestion control mechanisms are minimised. Traffic variations are characterised by a model of six layers, each of which is assigned a corresponding layer in a traffic management model. We define the functionalities of each level in the latter modell and review possible implementations. In particular, we report on an implementation of two of these layers and discuss in some detail how a third one could be added.

We propose a simple and robust automatic strat- egy for managing SDH/SONET and/or ATM virtual paths in real time to handle slow traffic variations. It consists of on-line measurements of offered traffics followed by the design and possi- ble implementation of a new virtual path network in a repeated cycle. Methods and parameters for traffic measurements and network updating are studied and optimised to achieve maximal traffic carrying capability at minimal cost. Applying our strat- egy to networks subject to partially unknown, variable traffics we discover a considerable adaptivity, large savings in transmis- sion capacity, and a performance far better than achieved by deterministic management.

We introduce a six layer model of traffic variations and a corresponding model of management actions. In particular, we traffic study real time traffic management on the traffic and session levels which correspond to variations on the hour and minute levels. Using virtual paths (VPs) and virtual channels (VCs) as the managed entities on the two levels respectively, we put forward a simple and robust strategy for traffic management in real time: For slow variations, i.e. on the traffic level, all nodes simultaneously monitor offered traffics and forward the result to a network management centre (NMC); The NMC computes a new set of optimal VPs compares the result to the existing network; If a change appears profitable, the necessary information is sent back to the nodes and the new design is implemented. For fast variations, i.e. on the session level, dynamic routing schemes similar to those of public switched telephone networks are employed. Applying our strategy to control general networks subject to traffics which in advance are known only as expectations, we develop simple methods for traffic observation, derive the optimal observation interval, brie y review algorithms to compute VP networks, develop a cost optimal test to determine whether to implement a new design or not, and derive the optimal updating interval. Running our strategy in a simulator, we discover a considerable adaptivity and a performance far better than obtained by alternative methods.

We study networks based on virtual paths, i.e. rearrangable end-to-end transport network. Virtual paths networks are readily implemented in any network using the synchronous digital hierarchy and/or the asynchronous transfer mode. The concept and its advantages, for example cost savings, network operation simplification and enhanced network management capabilities, are discussed. Algorithms for virtual path designs are reviewed and a new algorithm is presented which is found to compare favourably with the algorithm providing the most similar features. Applying it to a real network, we turn to operational aspects of reconfigurable networks such as methods and parameters for traffic estimation and network updating. The validity of the results is demonstrated by means of simulations of a number of networks subject to variable traffics.

We study management of transport networks based on virtual paths, i.e. rearrangable end-to-end transport network \highways". Among its main advantages are found cost savings, network op- eration simplification and enhanced network management capabilities. Virtual paths networks are readily implemented in any network using the synchronous digital hierarchy and/or the asyn- chronous transfer mode. Algorithms for virtual path designs are reviewed and found to have a few shortcomings. A new algorithm is therefore proposed and it is found to compare favourably with the algorithm providing the most similar features. Applying the proposed algorithm to a large number of networks, we discuss operational con- siderations associated with reconfigurable networks and investigate simple traffic estimation pro- cedures and their performance in this context. Finally, the choice of reconfiguration parameters such as updating frequency, and the impact of network design time are examined.

Traffic in ATM networks can be described by numerous parameters. On a per session basis, one may use peak rate, average rate, sustainable rate, average burst duration, average silence duration, and others. In a longer time scale, parameters like the average and peakedness of connection request arrivals, the average and variance of session holding times, and so on are proposed. Theoretically, users should provide the former parameters and network operators the latter. In reality, however, few users can be expected to provide all sorts of statistical information about their traffic in advance, and operators do not have enough experience to prepare traffic forecasts for new services and applications. Moreover, even if the information could be provided, the lack of simple yet valid traffic models for ATM networks means that it is far from clear how it should be used. Realising that ATM networks, which are already being built, at least for the next few years will have to operate under these uncertain conditions, we focus on robust and forgiving strategies for traffic management. The idea is to use only little information about offered traffics and then dynamically control resource allocations, so as to provide acceptable quality of service combined with high utilisation. We give some draft proposals in this direction and point at some open research issues.

Models of bursty arrival processes are reviewed and compared with re- spect to their in uence on the performance of a single server system. Considerable variations are found between different models fitted to the same traffic, though some models exhibit similar behaviour. Comparing the ability of the models to reproduce the performance of an explicitly given parcel they all tend to fail.

Traffic models are an essential component of performance analyses of telecommunication networks. This paper investigates the modeling of bursty traffic streams in both single source and multiple source environments. Using a ranking scheme it is shown that, for the single server case, the switched Poisson process models performed the best and that for the multiple source case, models based on the switched Poisson process, or its discrete version (switched Bernoulli) together with the threshold model may give the closest match to the performance of the “real data“. The results also seem to suggest that for traffic for which there is no known, ideal model, a traffic-oriented fitting method performs better than a model oriented one. This study has concentrated on a particular set of “real data” and it has not tried to consider the many different types of bursty traffic, since, at the time of commencing this study, there was little real data available

Routing problems are often encountered when designing and managing telecommuni- cation networks. Today, routing problems are affected by the growing sizes of networks, which increase the complexity, and by introduction of new services and technologies, which rise the demands. Rather than resorting to entirely heuristic algorithms and/or large data bases with off-line precomputed routing information for various situations, we propose a a new decomposition method whereby any routing algorithm is speeded up considerably, thus permitting the deployment of well founded routing algorithms even for real time purposes. In our paper, we present this new method in formal terms, and apply it to a real routing problem. Finally, investigating the performance of our fast implementation by comparing it to the optimal, non-accelerated solution, we find that considerable time savings can be made at a limited cost in terms of non-optimality of the final solution. It is also emphasised that in real-time applications with non-constant traffics, this nominal degradation might be more than compensated for by the prompt delivery.

Markov modulated models are often used to model traffic in a network. How well the processes really capture the essentials of real traffic is less known. We review some of these models and apply a number of associated fitting methods to an Ethernet traffic in different timescales. The fitted models are then used as input to a buffer where buffer occupancies are used as a measure of the goodness of the models. Also, the ability of a fitting method to correctly model the traffic when applied to traffic generated from a model that the fitting method was intended for is examined.

Current developments in technologies and markets stress the importance of exible and robust performance optimisation and congestion control. In intelligent networks, congestion control has traditionally taken a purely technical view and focused on pro- tecting individual nodes from harmful overloads. We take a profit oriented, network wide view and develop a congestion control mechanism to ensure real time performance under profit maximisation. Numerical studies in a simulator show that the proposed mechanism is robust and, in particular in overloaded states, provides a substantial improvement in comparison to conventional load control mechanisms.

Current developments in technologies and markets stress the importance of exible and robust performance optimisation and congestion control. In intelligent networks, conges- tion control has traditionally taken a mainly technical view and focused on protecting individual nodes from harmful overloads. We take a profit oriented, network wide view and develop a congestion control mechanism to ensure profit maximisation under real time performance. Numerical studies in a simulator show that the proposed mechanism is robust and, in particular in overloaded states, provides a substantial improvement in comparison to conventional load control mechanisms.

Simulation is a tool that can be used to assess functionality and performance of communication networks and protocols. However, efficient simulation of complex communication systems is not a trivial task. In this paper, we discuss modeling and simulation of bus-based communication networks and present the results of modeling and simulation of a multigigabit/s LAN. We used parallel simulation techniques to reduce the simulation time of the LAN and implemented both an optimistic and a conservative parallel simulation scheme. Our experimental results on a shared memory multiprocessor indicate that the conservative parallel simulation scheme is superior to the optimistic one for this specific application. The parallel simulator based on the conservative scheme demonstates a linear speedup for large networks.

Measurement systems are of increasing importance for manufacturing, due to high automation level of production processes. Although most measurement systems have much in common and are expensive to construct, these systems are often developed from scratch, hardly reusing the available designs and implementations. To address this, we have designed and implemented an object-oriented framework for the domain of measurement systems that can be used as the core of measurement systems. Evaluations of the framework show that it captures the main concepts in the domain and that the required extensions for individual applications are limited. In this paper, a number of example framework instantiations are presented. The lessons we learned during the framework design and an evaluation of the object-oriented modelling paradigm are presented.

Design patterns have proven to be useful for the design of object-oriented systems. The power of a design pattern lies in its ability to provide generic solutions that can be specialised for particular situations. The implementation of design patterns has received only little attention and we have identified two relevant problems associated with the implementation. First, the traceability of a design pattern in the implementation is often insufficient; often the design pattern is `lost'. Second, implementing design patterns may present significant implementation overhead for the software engineer. Often, a, potentially large, number of simple methods has to be implemented with trivial behaviour, e.g. forwarding a message to another object. In this paper, the layered object model (LayOM) is presented. LayOM provides language support for the explicit representation of design patterns in the programming language. LayOM is an extended object-oriented language in that it contains several components that are not pa rt of the conventional object model, such as states, categories and layers. Layers are used to represent design patterns at the level of the programming language and example layer types for four design patterns are presented. LayOM is supported by a development environment that translates LayOM code into C++. The generated C++ code can be used as any C++ code for the development of applications. An important aspect of LayOM is that the language itself is extensible. This allows new design patterns to be added to the language model.

Design patterns have proven to be useful for the design of object-oriented systems. The power of a design pattern lies in its ability to provide generic solutions that can be specialised for particular situations. The implementation of design patterns has received only little attention and we have identified two relevant problems associated with the implementation. First, the traceability of a design pattern in the implementation is often insufficient; often the design pattern is `lost'. Second, implementing design patterns may present significant implementation overhead for the software engineer. Often, a, potentially large, number of simple methods has to be implemented with trivial behaviour, e.g. forwarding a message to another object. In this paper, the layered object model (LayOM) is presented. LayOM provides language support for the explicit representation of design patterns in the programming language. LayOM is an extended object-oriented language in that it contains several components that are not part of the conventional object model, such as states, categories and layers. Layers are used to represent design patterns at the level of the programming language and example layer types for four design patterns are presented. LayOM is supported by a development environment that translates LayOM code into C++. The generated C++ code can be used as any C++ code for the development of applications. An important aspect of LayOM is that the language itself is extensible. This allows new design patterns to be added to the language model.

Large object-oriented systems have, at least, four characteristics that complicate object communication, i.e the system is distributed and contains large numbers, e.g. thousands, of objects, objects need to be reallocated at run-time and objects can be replaced by other objects in order to adapt to the dynamic changes in the system. Traditional object communication is based on sending a message to a receiver object known to the sender of the message. At linking or instantiation time, an object establishes its acquaintances through name/class-based binding and uses these objects through its life time. If this is too rigid, the software engineer has to implement the binding of objects manually using pointers. In our experiments we found the traditional acquaintance communication semantics too limited and we identified several problems, related to the reusability of objects and selection mechanisms, understandability and expressiveness. We recognised that it is important to separate a class or object's requirements on its acquaintances from the way an object selects and binds its acquaintances in actual systems. Based on this observation, we studied the required expressiveness for acquaintance handling and identified four relevant aspects: type and duration of binding, conditions for binding, number of selected objects and selection region for binding. To implement these aspects, we defined acquaintance layers as part of the layered object model. Acquaintance layers uniformly extend the traditional object-oriented acquaintance handling semantics and allow for the first-class representation of acquaintance selection and binding, thereby increasing traceability and reusability.

Reuse of software has been one of the main goals of software engineering for decades. Reusing software is not simple and most efforts resulted in small reusable, black-box components. With the emergence of the object-oriented paradigm, the enabling technology for reuse of larger components became available and resulted in the definition of object-oriented frameworks. Frameworks attracted attention from many researchers and software engineers and frameworks have been defined for a large variety of domains. The claimed advantages of frameworks are, among others, increased reusability and reduced time to market for applications. Although several examples have shown these advantages to exist, there are problems and hindrances associated with frameworks that may not appear before their usage in real projects. The authors have been involved in the design, maintenance and usage of several object-oriented frameworks and based on the experiences from these projects, a number of problems related to frameworks are descr ibed. The problems are organised according to four categories, i.e. framework development, usage, composition and maintenance. For each category, the most relevant problems and experiences are presented. This paper may help software engineers to avoid the described problems, whereas researchers may try to address these topics in their research.

In this paper, a number of problems of conventional automated software engineering support environments are described. These problems are related to the functional approach these environments take, the lack of initiative these systems exhibit and the causal connection gap between design information and the operational entity. As an alternative, we suggest an autonomous agent approach. Each artifact is modelled as an autonomous agent that is responsible for its own development. Artifacts are organised in specialisation hierarchies which can be traversed by agents. As a consequence, the methodology needs to be inverted, i.e. specified from the artifact's perspective rather than from the software engineer's perspective.

Object-oriented software architectures, such as design patterns and frameworks, have found extensive use in software industry. As a design technique, these architectural structures have proven themselves very valuable. Their implementation in traditional object-oriented programming languages, however, is often difficult, since these languages provide no support for the explicit specification of software architectures. In this paper, we develop the notion of architectural fragments, i.e. reusable architectural specifications that describe a design pattern or a framework architecture. An architectural fragment describes the structure of an architecture in terms of its components (roles), the architecture-specific behaviour of the components and the architecture-specific interaction between the components. Fragments can be composed with each other and with reusable components. To illustrate the use of fragments, we describe the role and architecture language constructs in the layered object model (LayOM) and present examples.

During the last years, one can recognise a development towards application domain languages and extensible language models. Due to their extended expressiveness, these language models have considerable advantages over rigid general purpose languages. However, a complicating factor in the use of extensible language models are the conventional compiler construction techniques. Compilers constructed using these techniques often are large entities that are highly complex, difficult to maintain and hard to reuse. As we have experienced, these characteristics clearly complicate extending existing compilers. As a solution to this, we developed an alternative approach to compiler construction is proposed, based on object-oriented principles. The approach is based on delegating compiler objects (dcos) that provide a structural decomposition of compilers in addition to the conventional functional decomposition. The dco approach supports modularisation and reuse of compiler specifications, such as lexer and parser speci fications. We constructed an integrated tool set, letos, implementing the functionality of delegating compiler objects.

A wavelet-based tool is reported for the analysis of Long-Range Dependence (LRD) traffic to allow for semiparametric estimation of the Hurst (H) parameter. Validation has been also done, by using fBm and fGn models, and the obtained estimations show excellent agreements with the theoretical results.

Study of long-range dependence (LRD) properties in real traffic has received an increasing attention in traffic analysis. A wavelet-based tool for the analysis of LRD is presented in this paper together with a semi-parametric estimator of the Hurst parameter. The estimator has been proved to be unbiased under very general conditions and efficient under Gaussian assumptions. Analysis of the Bellcore Ethernet traces as well as of some VBR video traces using the wavelet-based estimator is reported.

This contribution deals with a formula for the capacity an ATM multiplexer must at least have to accomodate the loss probability demands of all connections. So it can be used for connection admission control as well as for network resource management purposes. The formula is based on the bufferless fluid flow multiplexer model. It allows for a more exact capacity evaluation than if equivalent bandwidths are used on a per-connection basis. On the other hand, it merely requires a computational effort which is comparable to evaluating the mean of a given distribution. Indeed, the bottleneck is the convolution of probability ensity functions, on which the formula operates. Two steps to reduce the computational effort are proposed: a framework for convolution operations, consisting of pre-computed probability density functions, and a suitable truncation of the state space.

VP reconfiguration is a powerful and flexible tool to cope with traffic changes and/or equipment failures in ATM networks. In the paper we present an application of a stochastic optimization algorithm called Simulated Allocation to the problem of VP reconfiguration in response to traffic shifts. The considered optimization task takes into account the cost of VPs reconfiguration imposed by changes in VP routing tables, rearrangement and possible loss of some calls in progress. Numerical results illustrating the effectiveness of the Simulated Allocation algorithm are given.

We show that the OQT is a precursor to geometric quantization

This book is devoted to a symplectic approach of classical and quantum mechanics

We define a Maslov index for symplectic paths by using the properties of Leray's index for pairs of Lagrangian paths. Our constructions are purely topological, and the index we define satisfies a simple system of five axioms. The fifth axiom establishes a relation between the spectral flow of a family of symmetric matrices and the Maslov index

We compare the indices of Robbin, salöamon, and McDuff with the cohomological index defined by Leray and extended by the author

We study the relation between the complete Maslov index defined by Leray and the author, and the Lagrangian path intersection index defined by Robbin and Salamon, and used by McDuff and Salamon in their study of symplectic topology.

In shared-memory multiprocessors, caches are attached to the processors in order to reduce the memory access latency. To keep the memory consistent, a cache coherence protocol is needed. A well known approach is to record which caches have copies of a memory block in a directory and only notify the caches having a copy when a processor modifies the block. Such a protocol is called a directory-based cache coherence protocol. This thesis, which is a summary of seven papers, identifies three problems in a directory-based protocol, and evaluates implementation and performance aspects of some design alternatives. The evaluation methodology is based on program-driven simulation. The write-invalidate policy, which is used in the baseline protocol, forces all other copies of a block to be invalidated when a processor modifies the block. This leads to a cache miss each time a processor accesses an invalidated block. To reduce the number of cache misses, a competitive-update policy is proposed in this thesis. The competitive-update policy is shown to reduce both the read stall and execution times as compared to write-invalidate under a relaxed memory consistency model. However, update-based policies need more buffering and hardware support in the caches. In the baseline protocol, the implementation cost of the directory is proportional to the number of caches. To reduce this cost, an alternative directory organization is proposed which distributes the directory information among the caches sharing the same memory block. To achieve a low write latency, the caches sharing a block are organized in a tree. The caches are linked into the tree in parallel with application execution to achieve a low read latency. The hardware-implemented directory controller in the baseline protocol may lead to high design complexity and implementation cost. This thesis evaluates a design alternative where the controller is implemented using software handlers executed on the compute processor. By using efficient strategies and proper architectural support, this design alternative is shown to be competitive with the baseline protocol. However, the performance of this alternative is more sensitive to other design choices, e.g., block size and latency tolerating techniques, than the baseline protocol.

The solution for multi-frequency plane waves propagating through a dissipative and nonlinear medium is shown for some examples of periodic conditions. The expression may for any given condition be expressed analytically as a ratio of Fourier series with Bessel function coefficients. In the examples are shown how the final appearance of any initial wave always is a pure periodic wave in the lowest frequency existing in the problem - the period of the condition.

Fault handling in Multi-Agent Systems (MAS) is not much addressed in current research. Normally, it is considered difficult to address in detail and often well covered by traditional methods, relying on the underlying communication and operating system. In this paper it is shown that this is not necessarily true, at least not with the assumptions on applications we have made. These assumptions are a massive distribution of computing components, a heterogeneous underlying infrastructure (in terms of hardware, software and communication methods), an emerging configuration, possibly different parties in control of sub-systems, and real-time demands in parts of the system. The key problem is that while a MAS is modular and therefore should be a good platform for building fault tolerant systems, it is also non-deterministic, making it difficult to guarantee a specific behaviour, especially in fault situations. Our proposal is to introduce sentinels to guard certain functionality and to protect from undesired states. The sentinels form a control structure to the MAS, and through the semantic addressing scheme they can monitor communication, build models of other agents, and intervene according to given guidelines. As sentinels are agents themselves, they interact with other agents through agent communication. The sentinel approach allows system developers to first implement the functionality (by programming the agents) and then add on a control system (the sentinels). The control system can be modified on the fly with no or minimal disturbance to the rest of the system. The present work is conducted in cooperation with Sydkraft, a major Swedish power distribution company. Examples are taken from that venture, and it is shown how problems can be solved by programming DA-SoC agents, developed here.

In many application domains the design of complex systems must cope with rapid and unforeseeable change, requiring an elaborate integration of execution and communication methods. This is especially the case when real-time constraints are present. In this paper a Multi-Agent Systems approach to distributed computing is presented. It has a programmable model of agent interaction, and the interaction is tightly integrated with the mechanisms that control the individual agent behaviour. We introduce the concepts of tailored interaction and semantic addressing which support abstraction, flexibility and robustness in open and emergent distributed systems. We then show that it enables the system to adapt to a number of frequent types of change in the environment. The automation of power distribution is a domain where new services are introduced and new infrastructures are exploited. DA-SoC is an agent architecture developed for this domain, and it exemplifies the need for adaptability and the use of semantic addressing to accomplish this.

The agent metaphor has become increasingly popular in distributed computing systems. As often with a new paradigm, it is not very clear what the exact meaning is. The meaning and the development of the agent paradigm is therefore discussed in this paper. One motivation for using agents in distributed systems is that, due to the rapid technical development of computing and communication facilities, these systems become larger, more complex, and with more heterogeneous components. The change rate for such systems will also increase. Another, but related, motivation is that, in many cases, the systems are open and emergent in new respects: Different parties will be in control of parts of the total system, and the change of the system may not be altogether predictable. It is simply not possible to describe the total system from any one point of focus or at any one point in time, meaning that traditional methods for describing and implementing the systems are not sufficient. Industrial applications (e.g. manufacturing systems, process control systems, or resource management systems) show many of the characteristics mentioned above. Specifically, there is a demand for integrating quite dispersed applications, for example systems for process control, market analysis, and economic management. There is also a stronger demand for communication and interacting with sub-contractors, partners, and customers. This is emphasized with the development of a common communication medium like the Internet. In the ISES project, the agent paradigm is studied and developed for use in Power Distribution Auto-mation. Examples of this work are given here, and the applicability of agent technology is shown for applications that concern Distribution Automation, Demand Side Management, and Home Automa-tion. It is also argued for the applicability of agent technology in other types of industrial applica-tions, based on similarities between problem domains.

Distributed computing, where the underlying network is complex, heterogeneous and un-reliable, is hard to realize with traditional approaches. Energy Distribution Automation (DA), where the power line is used for communication, and the processing is highly distributed, is our application domain for assessments of new technologies for this kind of system. In this paper we propose a Multi-Agent approach with a large expected number of heavily distributed, medium grain agents. The concept of Agent-Oriented Programming (AOP) is adopted, and an architecture, the DA-SoC, is developed. The language DAAL is used for pro-gramming the agents, and the most important feature is a novel programmable Model of Inter-action (MoI), which allows the user to tailor agent interaction and the social behaviour of agents in the society. The semantic addressing is a powerful abstraction mechanism, that facilitates run-time network management. The emphasis of the paper is on describing and analysing the architecture, the language and the programmable MoI. An example from the DA application domain illustrates how the DA-SoC can be used. The research reported here is a cooperation between the authors’ affiliations mentioned above, and it is a part of the larger projects, Intelligent Distribution Automation (IDA) at Syd-kraft, Malmoe, Sweden, and Societies of Computation (SoC) at the University of Karlskrona/Ronneby, Sweden.

As computing systems are being applied to ever more demanding and complex domains, the infeasibility of constructing a single, monolithic problem solver becomes more apparent. Furthermore, important applications such as different types of management control systems for power generation and distribution are inherently distributed. A promising novel research direction to overcome complexity barriers in the design and maintenance of complex distributed applications is based on the view of such systems as societies of cooperating agents. In this paper we describe an agent oriented design for load man-agement in an automated power distribution system and a testbed for implementations of such systems. We also present an agent with some novel and important features. The research reported here is a cooperation between the authors’ affiliations mentioned above, and it is a part of the larger projects, Intelligent Distribution Automation (IDA) at Sydkraft, Sweden, and Societies of Computation (SoC) at the University of Karlskrona/Ronneby, Sweden.