The increasing demand for real-time inference on high-volume network traffic has led to the rise of in-network machine learning, where programmable switches execute various models directly in the data-plane at line rate. Effective network management often involves multiple prediction tasks, such as predicting bit rate, flow size, or traffic class; however, existing solutions deploy separate models for each task, placing a significant burden on the data-plane and leading to substantial resource consumption when deploying multiple tasks. To address this limitation, we introduce MUTA, a novel in-network multi-task learning framework that enables concurrent inference of multiple tasks in the data-plane, without exhausting available resources. MUTA builds a multi-task neural network to share feature representations across tasks and introduces a data-plane mapping methodology to fit it within network switches. Additionally, MUTA enhances scalability by supporting distributed deployment, where different layers of a multi-task model can be offloaded across multiple switches. An orchestrator employs multi-objective optimization to determine optimal model placement in multi-path networks. MUTA is deployed on P4 hardware switches, and is shown to reduce memory requirements by ×10.5, while at the same time improving accuracy by up to 9.14% using limited training data, compared with state-of-the-art single-task learning solutions.

The need for real-time inference of large volumes of data led to the development of in-network machine learning. Programmable network switches can now execute various machine learning models in the data-plane at line rate. While a stream of data may require several prediction tasks, such as predicting bit rate, flow size, or traffic class, current solutions only support separate models for each task. This places a significant burden on the data-plane and leads to substantial resource consumption when deploying multiple tasks. To solve this problem, we introduce MUTA; a novel in-network multi-task learning solution. MUTA enables executing multiple inference tasks concurrently in the data-plane, without exhausting available resources. It introduces a data-plane mapping methodology to fit non-binarized multi-task neural networks within network switches. MUTA is deployed on P4-based hardware switches, and is shown to reduce memory requirements by × 10.5 and improve accuracy by up to 9.14% using limited training data, compared with state-of-the-art single-task learning solutions.

Intelligent data-planes (IDPs) can enhance network service performance and adaptation speed by executing one or more machine learning (ML) models directly on the served flows. The real-time ML inference enables line-speed decision-making for some traffic management functionalities. Due to the inherent scarcity of both the computational and memory resources and the strict high-speed per-packet processing demands, existing IDP deployments either realize only a limited set of ML models such as decision trees, or require substantial modifications in the switch hardware. In this paper, we propose INQ-MLT, a novel ML-based management toolbox to address the aforementioned limitations. INQ-MLT delegates the task of training various ML models to the control-plane. The latter adopts a tailored quantization-aware training process to compensate for the effect of precision loss resulting from quantization. The toolbox then employs a quantization mechanism to transform the trained ML model parameters (e.g., weights and activations) from floating-point representations to compact low-precision fixed integer values that can be easily processed and stored in the data-plane. Finally, the trained model is deployed into the IDP pipeline by restricting all its inference operations to basic arithmetic operations. To analyze the performance of INQ-MLT, we quantify the accuracy loss resulting from the quantization step through rigorous theoretical analysis. A proof-of-concept implementation of the proposed toolbox is developed using P4-based software switches. Experiments on two use-cases demonstrate that the deployed quantized models have almost no loss of accuracy when compared to their floating-point counterparts.

The frequent occurrence of network attacks highlights the criticality of developing effective intrusion detection systems (IDSs) that can promptly detect and respond to malicious flows. The proliferation of programmable devices has opened up new possibilities for integrating intelligent IDSs into the data-plane. This allows the execution of machine learning (ML)-based detection models at line-rate, meeting the low latency requirements of anomaly detection. We propose a two-stage confidence-based Intrusion Detection System (TSCIDS) that aims at early detection while considering the level of certainty of prediction. The control-plane adopts a customized transfer learning scheme, wherein two interdependent convolutional neural network (CNN) models are trained, one using the early context of flows and the other adding the later context. A post-hoc calibration method is applied to improve the performance of models. TSCIDS detects anomalous behavior in different phases of flows while allowing the latter CNN to leverage the hidden state of the early CNN. TSCIDS ensures that the two CNN models are integrated into the data-plane pipeline by building the inference steps of CNN into different modules, using switch-supported operations. Simulation results show that the calibrated model can detect more attacks in the early phase compared to the uncalibrated model. Additionally, the training scheme saves the memory consumption of running models on programmable devices.

Offloading some of the traffic management decision-making functionalities to intelligent data-planes (IDPs) can significantly enhance the accuracy and adaptation speed of network services. An IDP executes, at line-speed, one or more machine learning (ML) models for real-time inference and decision making. Unfortunately, existing IDP deployments either realize only a limited set of ML models such as decision trees or require substantial modifications in the switch hardware. These limitations can be attributed to the inherent scarcity of both the computational and memory resources and the strict high-speed per-packet processing demands. To address the aforementioned limitations, we propose a novel ML-based management framework, the in-network quantized ML architecture (INQ-MLA). First, INQ-MLA delegates the task of training and continuously optimizing the IDP ML model to the control-plane. The latter adopts a tailored quantization-aware training process to compensate for the effect of precision loss due to quantization. Second, INQ-MLA employs an efficient quantization mechanism to transform the trained ML model parameters (e.g., weights and activation functions outputs) from floating-point representations to smaller low precision fixed integer values that can be easily processed and stored in the data-plane. Finally, INQ-MLA ensures that the deployed ML model is integrated into the IDP pipeline by limiting all its execution operations to simplified arithmetic operations that are available in most switches. We developed a proof-of-concept implementation of our proposed architecture using P4-based switches. Experimental results demonstrate that INQ-MLA can achieve a high-level of accuracy at runtime.

Fast and customizable programmable data planes (PDPs) implementing new services such as multi-flow synchronization, on- and in-time delivery, and in-network caching and compression are key enablers to future applications (e.g., streamed holograms, telesurgery, and autonomous industrial systems). This paper outlines the design principles of EP4, an application-aware extended P4-based network architecture that offers hosted applications an extensible catalog of services through its control plane. The latter configures a PDP that can achieve minimal parsing and processing for fast-tracked packets as well as customized processing and forwarding for other packets. An extended parser (eParser) performs the first task, which reduces the necessary latency experienced by packets. Alternatively, adaptive processing is achieved using an enhanced processor (eProcessor) that optionally parses customized headers using just-in-time programmable parsers. It then executes selected P4 packet processing pipelines implementing different services. These programs are installed at runtime without impacting other switch functionalities. Experimental results demonstrate the architecture's enhanced performance compared to current solutions.

A major challenge to the anticipated large scale deployment of virtual network functions (VNFs) at the network edge (NE) is the ability to efficiently allocate and manage its scarce resources to meet these functions workload fluctuations. In this paper, we describe a novel containerized infrastructure manager (cIM) that extends current managers, such as Kubernetes, with the necessary building blocks to provide an accurate resource allocation service to containerized VNFs at scale. The proposed cIM treats the containerized VNF components (cNFCs), as atomic special purpose functions that can be rapidly deployed to form complex network services. The main component of the proposed cIM, the resource reservation manager (RRM), employs concepts of risk pooling in the insurance industry to accurately reserve the needed resources for the hosting containers and meet anticipated cNFCs demand fluctuation. The reserved quota of re-sources ensures the desired availability level of the cNFCs without over-provisioning the scarce resources of the NE. Experimental results demonstrate that our proposed cIM significantly improve the performance of the cNFCs and guarantees their availability with minimal resource reservation.

Managing and orchestrating the behaviour of virtual network functions (VNFs) remains a major challenge due to their heterogeneity and the ever-increasing resource demands of the served flows. In this paper, we propose a novel VNF manager (VNFM) architecture to automate the process of selecting appropriate VNF management actions (e.g., migration and vertical and horizontal scaling) with their corresponding configuration parameters (e.g., migration location or amount of resources needed for scaling). More precisely, we first propose a novel Markov decision process with parameterized actions to accurately describe each VNF and its permissible lifecycle management (LCM) operations. The use of parameterized actions allows us to rigorously represent the functionalities of the VNFM in order perform various operations on the VNFs. Next, we propose a two-stage reinforcement learning (RL) scheme that alternates between learning optimal LCM actions and updating their parameters selection policy. In contrast to existing schemes, the proposed work uniquely provides a holistic management platform that unifies individual efforts targeting single LCM functions such as VNF placement and scaling. Performance evaluation results demonstrate the efficiency of the proposed VNFM in maintaining the required performance level of the VNF while optimizing its resource configurations.

Fault-tolerance, survivability and resiliency in wide area networks have long been prominent research topics. With the popularity of the cloud service model and the novel software defined networking (SDN) paradigm, there is renewed interest in failure protection and restoration in network service provisioning. In this work, we propose a novel protection and restoration based virtual network management scheme to enhance fault tolerance in infrastructure-as-a-service, deployed over networked cloud infrastructure. The networked cloud infrastructure is composed of multiple geographically distributed datacenters that are interconnected with SDN. Both compute and network resources are allocated by formulating the virtual network embedding problem as an integer linear program. A shared backup virtual link protection mechanism and a reactive traffic engineering network failure restoration algorithm are proposed and integrated with the framework to provide recovery from unexpected link failures. We implemented the framework on an emulated SDN testbed and evaluated the performances of the algorithms during single and multiple link failures. Experimental results demonstrate trade-offs of the proposed approaches and their applicability in different application scenarios.

This paper proposes a Cloud Infrastructure-as-a-Service (IaaS) framework that allows customers to have their high performance computing applications hosted efficiently and Cloud Service Providers (CSPs) to use their resources profitably. The solution introduces a distributed architecture that manages geographically distributed Data Centers (Geo-Data Centers) logically grouped in regions. This framework overcomes the challenges of traditional centralized provisioning approaches: (a) efficient provisioning of IaaS demand, (b) scale with respect to the growing number of IaaS requests, (c) guarantee of the stringent Quality of Service requirements of IaaS requests, and (d) efficient use of Cloud Geo-Data Center computing resources. Our architecture incorporates two decentralized approaches, hierarchical and distributed, that use auctions instead of a pay-as-you-go pricing scheme. The two approaches use a large-scale optimization technique for the allocation of Geo-Data Centers computing resources. The results of a simulation demonstrate an efficient use of computing resources and a significant reduction in computation time. This ensures adequate scalability to meet an exponential growth of IaaS demand. The auction-based approaches are also shown to provide monetary benefits to the participants.

Named data networking (NDN) is a nascent vision for the future Internet that replaces IP addresses with content names searchable at the network layer. One challenging task for NDN routers is to manage huge forwarding information bases (FIBs) that store next-hop routes to contents. In this article, we propose a family of compressed FIB data structures that significantly reduce the required storage space within the NDN routers. Our first compressed FIB data structure is FCTree. FCTree employs a localized front-coding compression, that eliminates repeated prefixes, to buckets containing partitions of routes. These buckets are then organized in self-balancing trees to speed up the longest prefix match (LPM) operations. We propose two enhancements to FCTree, a statistically compressed FCTree (StFCTree) and a dictionary compressed FCTree (DiFCTree). Both StFCTree and DiFCTree achieve higher compression ratios for NDN FIBs and can be used for FIB updates or exchanges between the forwarding and control planes. Finally, we provide the control plane with several knobs that can be employed to achieve different target trade-offs between the lookup speed and the FIB size in each of these structures. Theoretical analysis along with experimental results demonstrate the significant space savings and performance achieved by the proposed schemes.

Named Data Networking (NDN) is a future Internet architecture that replaces IP addresses with namespaces of contents that are searchable at the network layer. A challenging task for NDN routers is to manage forwarding-information bases (FIBs) that store next-hop routes to contents using their stored usually long names or name prefixes. In this paper, we propose FCTree, a compressed FIB data structure that significantly reduces the required storage space at the router and can efficiently meet the demands of having routes that are orders of magnitude larger than IP-based ones in conventional routing tables. FCTree employs a localized front-coding compression to buckets containing partitions of the routes. The top routes in these buckets are then organized in B-ary self-balancing trees. By adjusting the size of the buckets, the router can reach an optimal tradeoff between the latency of the longest prefix matching (LPM) operation and the FIB storage space. In addition, in contrast to existing hash and bloom-filter based solutions, the proposed FCTree structure can significantly reduce the latency required for range and wildcard searches (e.g., for latency sensitive streaming applications or network-layer search engines) where up to k routes are returned if they are prefixed by a requested name. Performance evaluation results demonstrate the significant space savings achieved by FCTree compared to traditional hash-based FIBs.

Mobile Cloud Computing enables resource limited mobile devices to support rich application services. Among three types of cloud services, Infrastructure-as-a-Service (IaaS) clouds provides compute infrastructure for mobile applications on demand. In IaaS-based mobile clouds, latency and bandwidth requirements can considered as critical factors impacting Quality of Service (QoS). Opposed to centralized clouds, geographically distributed clouds realize higher QoS benefiting the proximity to the end user. In this paper, we propose an IaaS framework with regional datacenters for mobile clouds. With the benefits of software-defined networking (SDN), we address impacts on QoS during mobility by serving mobile user via the optimum datacenter. A test-bed was developed to measure the performance of service allocation and relocation in proposed framework.

In this work, we have introduced a fault tolerant IaaS resource management framework for networked cloud Infrastructure. Distributed multiple cloud provider sites (i.e. datacenters) are interconnected with software defined network (SDN). Both datacenter and SDN resources are allocated by formulating and solving integer linear programming based virtual network embedding problem. Two reactive traffic engineering network failure restoration algorithms have been added to the framework along with SDN-based monitoring scheme to dynamically identify and recover from unexpected link failures. We have implemented the framework on an emulated SDN testbed and evaluated the performance of aforementioned algorithms with multiple link failures. Experimental results demonstrate tradeoffs of proposed approaches and their applicability in different application scenarios.

In order to address the ossification of the traditional network, there have been many studies that show the benefits of the orthogonality offered by the principles of Software Defined Networking (SDN). Therefore, as the concept of SDN saw wide spread acceptance, it's adaptability in wireless networks began to emerge. Many proposals in the literature have addressed the issues that are related to the Mobile Ad-hoc Networks (MANETs). A computing environment formed atop a MANET that is closely linked to the rigidity of the underlying network is called Mobile Ad-hoc Cloud. In this paper we show how a seamless disruption tolerant mobile ad-hoc cloud can be maintained with the assistance of the adaptive principles offered by the SDN framework. Further, we demonstrate how a selection of mobile ad-hoc cloud composition traffic can reduce the latency in task computation and result collection in comparison with the traditional non-SDN ecosystem.

Infrastructure as a service (IaaS) has attracted significant attention from cloud research communities. While compute and storage resource management has been developed and studied to a greater extent, the use of network resource management in the context of IaaS is still in its early stages. There is a need for a comprehensive virtualization framework capable of providing users with low-level network and compute resource control while improving underlying resource utilization. We propose a Software-Defined Networking (SDN) IaaS framework that explicitly integrates network virtualization, including computing and storage, into a cloud platform. Our proposed framework abstracts data-center compute and network resources into a virtualized pool of resources, links them to logically compose virtual networks, and performs automated configurations to serve IaaS requests from users. We designed and constructed an SDN test-bed to verify the operation of the proposed framework. We successfully demonstrated our system's ability to meet our design objectives: (i) providing high degree of control over connectivity and Quality of Service (QoS), (ii) fully automated service delivery, (iii) fast and low overhead resource provisioning. We evaluated the feasibility and scalability of the proposed framework using statistics gathered from its deployment.

Cloud clients (CCs) of current distributed cloud applications are still not assured of their service quality, in particular, in terms of the experienced latency. Unfortunately, this is mainly attributed to the unpredictability of the communication links among their hosting distributed data centers. To address this problem, this article introduces a novel virtual-network-as-a-service (VNaaS) model to host these applications. In contrast to existing randomly or statically provisioned inter-data centers bandwidth sharing models, the proposed model allows CCs to accurately express their varying network resources needs, demand constraints and tolerance to the cloud latency. In turn, the model maps these requirements to create inter-data centers virtual links hosting each multiple virtual pipes with differentiated service qualities to carry the CC's various traffic flows. To aid the CCs in optimally determining their VNaaS demands, given the budget constraints of their hosted applications, we also develop a novel demand selection scheme based on a two stage-budget allocation mechanism. In the first budgeting stage, the CC calculates an optimal effective service rate for each of its virtual link along with a corresponding link budget and price index. In the second stage, the virtual link budget is distributed to purchase bandwidth for the link's virtual pipes, each with a given service quality and pricing. We then extend the proposed model to allow the CC to enforce any required virtual links' capacity constraints on the effective service rates resulting from the traffic matrix on the VNaaS. Finally, we develop corresponding differentiated VNaaS pricing and service monitoring mechanisms that can be employed by the cloud service provider (CSP) to regulate the offerings and demands of the distributed cloud services. Performance evaluation results demonstrate the significant improvement in the service quality, the higher utilization of the cloud resources and the increase in the CSP's net profit.

Current data center designs suffer from poor resource utilization. Several research works have addressed the data center resource allocation problem, and recent proposals have focused on network resource utilization as a bottleneck problem in data centers. However, little attention has been paid to manage the convergence of networking and other infrastructure resources. In this paper, we present a solution for improving data center's resource utilization. We introduce a unified cloud resource representation model and build a general resources repository using that model. We define a combined controller to manipulate infrastructure resources collected in the repository. A joint optimization model that performs the resource allocation as the main controller operation is also presented. This model represents the integration of semantic similarity and closeness centrality concepts and is formulated on a two-phase Mixed Integer Linear Programming (MILP-2P-IaaS): (i) mapping of hosting resources, and (ii) connectivity composition. Simulation results show that the (MILP-2P-IaaS) resource allocation approach improves data center's resource utilization and outperforms other benchmarks in terms of resource utilization and acceptance ratio.

This paper proposes a composition-based resource allocation approach to cloud infrastructure-as-a-service (IaaS). A number of previous proposals have addressed resource allocation for IaaS services with bandwidth guarantee as the main obstacle to accessing datacenter-based cloud resources. Less focus has been put on utilization of hosting resources, i.e. computing and storage. Shortcomings with these proposals (i) may ensue in a high blocking of IaaS requests and (ii) a less efficient use of datacenter resources, which may impact a provider's revenue. In this research paper, a Two-Phase IaaS resource allocation approach (2P-IaaS) is proposed to address these shortcomings: (i) hosting resources mapping phase, and (ii) connectivity composition phase. This approach adopts a semantic ontology model with its associated reasoning capabilities to represent, discover and assign diverse cloud resources to IaaS requests. Furthermore, the proposed approach utilizes semantic similarity and closeness centrality to define an efficient cloud IaaS topology that connects the assigned resources. In this research, experiments on various sets of IaaS requests show significant advantages when compared with selected benchmarks.

This paper presents a novel distributed Quality of Service (QoS) monitoring scheme for cellular data networks serving highly dynamic users with power- limited devices. The proposed scheme relies on candidate QoS monitoring users that can efficiently submit QoS related measurements on behalf of their neighbors. These candidate users are chosen according to their devices' residual power and transmission capabilities and their estimated remaining service lifetime. Service monitoring users are then selected from these candidates using a novel user-to-user semantic similarity matching algorithm. Simulation results demonstrate the significant gains achieved by the proposed scheme in terms of the reduced traffic overhead and overall consumed users' devices power while achieving a high monitoring accuracy.

This paper focuses on the resource allocation problem in the context of Cloud Computing. More specifically, this work considers the problem of optimizing the mapping cost of Infrastructure as Cloud Service (IaaS) onto a Networked Edge Data-Centers (DCs) with respect to Quality of Service (QoS) requirements. This work proposes to dynamically partition the networked DCs resources over IaaS requests belonging to different QoS classes. In literature, a number of works have proposed IaaS mapping approaches; however their focus was mainly on the cloud hosting requirements and do not take into account the dynamics of IaaS QoS requirements. Consequently, they may not offer QoS guarantees for accepted IaaS requests which may result in a higher customer dissatisfaction ratio. The originality of our work is in the forethought and the investigation of these issues. To do so, a column generation based-formulation is proposed coupled with the Branch and Bound technique in order to solve it efficiently. Doing so, this allows the Cloud Provider to: (i) minimize IaaS mapping cost, and (ii) calculate the optimal and dynamic partitioning of DCs resources to uphold QoS guarantees for IaaS requests.

Cloud service providers (CSP) strive to effectively provision their cloud resources to ensure that their hosted distributed applications meet their performance guarantees. However, accurately provisioning the inter-data centers network resources remains a challenging problem due to the cloud hosted applications' workload fluctuation. In this paper, we propose a novel approach that enables a CSP to offer Elasticity-as-a-Service (EaaS) for inter-data centers communication in order to guarantee the performance of distributed cloud applications. The contributions of the proposed work are two fold; first, we develop an efficient approach that enables the CSP to estimate and reserve the pool of network resources needed to fulfill the demands imposed by the network workload fluctuations of applications subscribing to this service. The approach allows the CSP to offer communication EaaS at differentiated levels based on the degree of bandwidth-sensitivity of the distributed cloud applications. In order to capture the inter-data centers network activity of hosted applications, we model their workloads using Markovian modeling. The second contribution is a novel dynamic pricing mechanism for network EaaS offerings that can be employed by the CSP to maximize the expected long-term revenue, and to regulate network elastic demands. Performance evaluation results demonstrate the efficiency of our proposed approach, the higher accuracy of our prediction method, and the increase in the CSPs net profit.

Network virtualization provides more flexibility in network provisioning as it offers physical infrastructure providers (PIP) the possibility of smoothly rolling out many separate networks on top of an existing infrastructure. A major challenge is the embedding problem of mapping virtual networks (VNs) onto PIP infrastructure. In the literature, a good deal of research has focused on providing heuristic approaches to this NP-hard problem, usually with the assumption that the PIP infrastructure is operational at all times. In virtualization environment, a single physical node/link failure can result in one or more logical link failures as it effects all VNs with a mapping that spans over. Setting up a dedicated backup for each VN embedding that is not shared with others is an inefficient use of resources. To address these concerns, this paper proposes two classes of periodic VN protection against link and node failures: (a) in the physical layer, by using a path or segment$p$-cycle technique and a column generation optimization model, and (b) in the VN layer, by augmenting the topology with redundant resources and subsequently applying a column generation mapping model. Our simulations show a clear advantage of our approaches over benchmarks in terms of PIP profit, backup cost/rate and resource use.

Optimization of service compositions represents a challenging area of research in mobile networks. The use of formal semantic descriptions of service interfaces and functionalities enable automated reasoning over service compositions. The growing challenge of finding effective methods to satisfy non-functional QoS requirements of customers exacerbates the composition problem. Fuzzy reasoning provides an effective means to represent QoS concepts that are vague or approximate thus providing a flexible best-match service query scheme. This paper presents a new fuzzy-induced, semantic-similarity-based service-specific overlay network composition over a hybrid service overlay network. The presented composition method allows for efficient, accurate, and QoS-aware component service discovery, composition, and execution. We demonstrate that our approach provides better performance and reduced composition delay when compared to other service-composition solutions.

Network virtualization is a promising new resource management approach that allows customized virtual networks (VNs) to be multiplexed on a shared physical infrastructure. In this paper, our focus is on the embedding of VN resources onto this infrastructure. Since this problem is known to be NP-hard, embedding proposals in literature are heuristic-based approaches that restrict the problem space in different dimensions. Limitations of these proposals are: (1) as embedding of VN links and nodes is performed in two separate stages, it may ensue in a high blocking of VN requests and a less efficient usage of substrate resources; and (2) as pricing of embedding resources is based on linear functions, it triggers no competition among VN users in order to maximize infrastructure provider profits. These drawbacks motivate us to propose a mathematical model that makes use of large-scale optimization tools and proposes a Column Generation (CG) formulation of the problem, coupled with branch-and-bound technique or rounding-off heuristic. We also propose a periodical planning of embedding process where profitable VN requests are selected through an auction mechanism. In our experiments with different substrate network topologies and many different VN request patterns, we show a clear advantage of auction-based CG models over present benchmarks .

The inefficiency of one-to-one matching of user requirements to a single service has given rise to service composition. Service composition allows simple services to be dynamically combined into more complex services. In mobile environments, service discovery and integration involves cooperation between various service nodes where reliance on a centralized service registry introduces efficiency and single point of failure problems. Plan- and workflow-based composition solutions restrict service discovery to defined abstract models. Semantic similarity measurements play an important role in service integration. In this paper we introduce a semantic similarity and semantic nearness service composition method. The objective of our work is to free service nodes from adherence to restrictive composition plans. Our work also enables compositions to succeed in dynamic environments through a decentralized solution even when nodes are unaware of nearby service nodes. This paper presents a service composition solution that semantically advances composition paths towards users’ needs with each service hop while guaranteeing a user-acceptable QoS level. Simulation results showcase the effectiveness of the presented solution and its compatibility with underlying service discovery methods.

This paper considers the problem of intra-interference in a Wireless Body Area Network (WBAN). The problem arises mainly because each bio-sensor collects different parameters with different data rate and alternation. Another source of interference is related to normal patient movement. Proposals in the literature usually assume that the interference can be handled using time multiplexing or by listening before transmission to avoid collision. However, these adaptive approaches, given the high-occupancy channels, bring with them major problems of collision and extra energy consumption. One solution could be a power control mechanism. Nevertheless, techniques of that kind are challenging in that they require periodic information on the condition of the wireless channels, conditions that are difficult to estimate. To address these concerns, a tree-based WBAN topology using a set of relay nodes with stable communication called a "virtual backbone" is proposed. As bio-sensors report data mainly in uplink traffic, it is assumed that they share a small number of wireless channels using the TDMA technique. Relay nodes, on the other hand, share the most number of channels in order to improve the fluidity of data across the WBAN. To mitigate co-channel interference among relays, two techniques from the literature called the adaptive data rate and the adaptive duty cycle are used. Simulation experiments showed that the proposed architecture, when combined with intra-interference mitigation techniques, improves energy efficiency and increases data rate.

Service Overlay Networks (SONs) provide new complex services in the Internet without requiring major changes to underlying physical networks. A SON is an overlay network made up of virtual nodes and links on top of the existing infrastructure. Whenever a client requests a specific service, such as streaming a video on his mobile device, the SON creates a path on the fly to deliver the video stream from the server to the client. This path is called a Service Specific Overlay Network (SSON) and consists of nodes that meet the Quality of Service (QoS), Quality of Experience (QoE), and technical requirements of the user. Unfortunately, the highly dynamic nature of overlay networks makes it challenging to keep video quality at the required levels. Errors and delays due mainly to congestion cause packets to be dropped or queued for a long period of time in intermediate nodes. Video quality is greatly affected by such impairments. In this paper, we propose a utility function to predict QoE of video delivered over SSONs. The proposed function is based on application-level statistical information, namely frame delay. It allows user QoE to be monitored in real-time without incurring additional overhead on the network. When degradation in the utility is detected, appropriate adaptation schemes can be used to restore the QoE to acceptable levels. We show the effectiveness and flexibility of the proposed scheme via mathematical proofs and simulation results.

In this paper, we propose a reliable topology design and provisioning approach for Wireless Body Area Networks (named RTDP-WBAN) that takes into account the mobility of the patient while guaranteeing a reliable data delivery required to support healthcare applications' needs. To do so, we first propose a 3D coordinate system able to calculate the coordinates of relay-sensor nodes in different body postures and movements. This system uses a 3D-model of a standard human body and a specific set of node positions with stable communication links, forming a virtual backbone. Next, we investigate the optimal relay nodes positioning jointly with the reliable and cost-effective data routing for different body postures and movements. Therefore, we use an Integer Linear Programming (ILP) model, that is able to find the optimal number and locations of relay nodes and calculate the optimal data routing from sensors and relays towards the sink, minimizing both the network setup cost and the energy consumption. We solve the model in dynamic WBAN (Stand, Sit and Walk) scenarios, and compare its performance to other relaying approaches. Experiment results showed that our realistic and dynamic WBAN design approach significantly improves results obtained in the literature, in terms of reliability, energy-consumption and number of relays deployed on the body.

Our focus, in this paper is on the embedding problem which consists on the mapping of Virtual Network (VN) resources onto physical infrastructure network. In relevant literature, number of works have been proposed to solve this challenging problem. However, few proposals have been dedicated to provide backup mapping mechanism for failing physical links. In virtualization environment a failing link will affect all VNs that span over this resource. To address these concerns, we propose in this work a resilient VN embedding (RVNE) approach that: (a) uses an adaptation of a Column Generation-based technique developed in our previous work, to calculate in a first stage a cost-efficient VN mapping while minimizing the affects of a single link failure in VN layer, and (b) proposes in a second stage a p-Cycle based VN protection approach that minimizes the backup resources while providing a full protection scheme. Experiments on large mix of VN requests show a clear advantage of resilient VN embedding model over benchmarks in terms of VN revenue, VN protection cost and resources utilization.

Network virtualization techniques aim at efficiently allocating the underlying substrate network (SN) resources to the hosted virtual networks (VNs). Unfortunately, over time, and due to the frequent initiation and termination of VNs, the available and utilized SN resources become fragmented. This in turn, gradually degrades the performance of these techniques. In this paper, we propose a novel proactive SN resource re-optimization technique that efficiently overcomes the fragmentation problem by performing appropriate re-arrangement, or house cleaning, for the available and utilized SN resources. To minimize the incurred computational overhead, the invocation of this technique is only triggered by certain events such as the departure of an expired VN. The contributions of the proposed work are two fold; first, we develop an efficient technique for the selection and re-allocation of VN portions that are contributing to the fragmentation problem. The technique takes into consideration the trade-off between the benefit from increasing the SN utilization and the cost incurred by the VN migration. The second contribution is novel VN live migration techniques that significantly reduce the service interruption time during migration. Simulation experiments demonstrate the achieved gain in the SN resource utilization as well as in the VN acceptance ratio and the net revenue.

Recent work in network management has led to the incorporation of Autonomic Computing (AC) concepts in management tasks to reduce cost and complexity. IBM was the first to propose the concept of AC for systems management [1]. Entities called Autonomic Managers (AMs) run an Autonomic Control Loop (ACL) to perform self-management functions. Every AM is responsible for monitoring one or more elements. All nodes of an overlay network are autonomic entities but only a subset is active in the management process depending on available resources. This paper proposes an efficient self-organizing mechanism for the autonomic management of overlay networks that reduces delays and error ratios of sensed data. A game theory approach to solve the AM selection problem is used. The game is based on a modified regret matching scheme. Advantages are that the approach is highly distributed and incurs limited network overhead. More importantly, it adapts to the dynamic network conditions. Simulation results show that our scheme achieves high gains in terms of delays and error rates compared to a straightforward scheme where nodes report to the closest AM.

Resource discovery aims to assist the Virtual Network Provider (VNP) in selecting the best Infrastructure Provider (InP) which has the best matching resources, for a Virtual Network (VN) request. It provides information about substrate resources that are advertised by InPs. In the current literature, resource discovery focuses mainly on static (functional) attributes of physical infrastructures with an emphasis on efficient organization of discovered data. In this paper we are proposing an aggregation based approach, called Aggregation-based Discovery for Virtual Network Environments (ADVNE). ADVNE allows VNP to make use of information on dynamic (nonfunctional) attributes to enhance the efficiency of resource discovery process while minimizing the continuous monitoring overhead. We analyze the performance of ADVNE and show its importance in VN embedding.

In future Internet architectures, Virtual Network Providers (VNP) must be able to compose virtual networks in a way that balances security with other priorities as expressed by Service Providers (SP). In this paper, we outline a framework in which a VNP can assess the security and assurance (robustness) properties of virtual networks and use these to help select an SP-appropriate topology. The topology selection algorithm is based on the Analytic Hierarchy Process (AHP) and the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) and uses cost and robustness as selection constraints. Our framework allows the VNP to identify the customer's needs for security and balance these with other priorities such as quality of service and cost. We show that this approach fits naturally into the business model of the VNP and takes advantage of risk management activities that may already be performed by SPs. Because security concerns can dampen enthusiasm for new ways of doing business, addressing the challenges above can help ease the transition to the future Internet.

In this paper, we present some interesting issues of resource allocation in network virtualization. We also propose RAiNV, a resource allocation framework to handle those issues. Sharing of resources is in the heart of network virtualization. The key idea of network virtualization is to replace conventional Internet service provider (ISP) with two separate business entities: infrastructure providers (InP) and service providers (SP). The InPs are the owners of resources. They are responsible for maintaining a certain level of quality of service (QoS) at the physical layer. The SPs lease resources from InPs. They offer various services to end users. It is more realistic to think about a scenario with multiple InPs and multiple SPs. A single service provider (SP) can integrate resources from multiple InPs. Similarly, a single InP can distribute resources to multiple SPs. As a result, optimal resource sharing becomes a challenging task. Resource allocation covers all the virtual network (VN) provisioning matters. Major parts include resource advertisement, request, negotiation, etc. The RAiNV aims to ease dealing with multiple InPs. Its performance is evaluated using obtained numerical results through simulations.

In this paper, our focus is on the embedding problem which consists on the mapping of VN resources onto physical infrastructure network. More specifically, we consider the problem of optimizing the Physical Infrastructure Provider's (PIP) profit while minimizing the dissatisfaction of VN customers. We propose to dynamically partition the PIP resources over VN requests belonging to different Quality of Service (QoS) classes using periodical auction mechanism. We formulate the dynamic embedding problem as an Integer Linear Program (ILP) that allows us to: (i) maximize the PIP profit, and (ii) calculate the optimal embedding scheme of VN requests without disruption of those previously accepted in order to uphold QoS guarantees.

Internet Protocol Television (IPTV) over broadband networks offers more services and flexibility, such as time-shifted TV and video-on-demand (VoD), than the traditional broadcast TV. Unfortunately, IP offers best effort service and Quality of Service (QoS) is not guaranteed. In order to provide QoS guarantees over the Internet, overlay networks are used. Impairments happening at the network layer such as delays and packet losses significantly affect the quality of the IPTV stream. Hence, a reliable mechanism to monitor the Quality of Experience (QoE) of IPTV users is essential to an IPTV service provider in order to avoid customer churn. It is essential that an IPTV provider be able to monitor the QoE with minimal overhead. Therefore, we propose a utility based QoE monitoring approach based on statistical losses at the application layer. This paper also presents mathematical proofs and simulation results that confirm the suitability and effectiveness of our proposed method.

Network virtualization is a concept in which a Virtual Network Provider constructs logical virtual networks for various clients on a common, virtualized infrastructure substrate. However, there is currently no general framework or benchmark for assessing the security properties of these logical networks within the context of network virtualization. In this paper, we describe a virtual network security assessment process in which a preference model is constructed over a select set of network element attributes. This preference model reflects the knowledge and experience of one or more security experts. The relevant attribute values are exposed during virtual network composition. Our process answers the question: “how does the security of my virtual network compare to an equivalent topology whose attribute values are most preferred by security experts?”

Wireless Sensor Networks (WSNs) are nowadays widely used in building decision support systems for better monitoring. One of the most interesting fields having an increasing need in decision support systems is agriculture. Inefficient and wasteful methods of agricultural monitoring lead to extra time and cost loss for farmers. This paper presents the iFarm framework system, an easy-to-use and expandable agricultural monitoring solution to enhance land productivity by better managing water, improving the socio-economic factor of farmers and their awareness, predicting and planning the crop yields. The iFarm system proposes WSNs as a promising mechanism to agricultural resources optimization, decision making, and land monitoring. WSNs make it possible to know at any time information about the land and crop conditions, so that farmers can be assisted with various notifications and suggestions during their farming tasks. It addresses the advantage of the precision agriculture approach to help making valuable decisions which could not only improve the land productivity but also optimize the use of resources. The paper gives a description of the precision agriculture monitoring approach that provides meaningful services to farmers.

Security is becoming one of the major issues that mobile users face. With their powerful devices, they are willing to take advantage of the wide range of services provided in mobile environments. Therefore, in this paper, we propose a solution to provide mobile users and service providers with security at different levels. Indeed, software policies enable each user/service provider to specify the rules for what they consider safe transactions. We also show that the addition of context information enhances security.

Context based dynamic adaptation of autonomous services and applications require the acquisition of an array of contextual information. Context dissemination mechanisms may result in network flooding, unrestricted access to private context information, and the inability of consumers to limit or personalize received context. Adapting context information disseminated to users is linked to direct contextual requests made by users. This paper describes policy- and ontology-based, context-aware system architecture. A Context Level Agreements is proposed to help deliver contextual information to users according to their needs. The paper illustrates use of the negotiation protocol through design and implementation of context-aware system architecture capable of acquiring, modeling, reasoning and disseminating context through ontologies.

In service-oriented environments and distributed systems, service composition allows simple services to be dynamically combined into new, more complex services. Service composition techniques are usually designed as an extension to service discovery. Traditional techniques try to match a user’s requirements, often complex, with the available services. However, one-to-one matching is inefficient; it is preferable to meet the request from available services even when one of the basic services is not present. Separating composition and discovery has also led to inefficiency, especially in a highly dynamic environment. With the heterogeneity of networks, users, and applications having multiple sources, constructing service-specific overlays in large distributed networks is challenging. In this article, we propose a new service composition algorithm to deal with the problem of composing multiple autonomic elements to achieve system-wide goals. Using a self-organizing approach, autonomic entities are dynamically and seamlessly composed into service-specific overlay networks. The algorithm combines composition and service discovery into one step, thereby achieving more efficiency and less latency. The decentralized and self-organizing nature of the algorithm allows it to respond rapidly to system changes. Extensive simulation results validate the effectiveness of the approach when it is compared to other solutions.

Building efficient general purpose sensor network infrastructure that could be leveraged by upper layer protocols is still an open research problem. In this paper, mobility is exploited to organize sensor nodes into generic efficient infrastructure. We propose layered infrastructure protocol (LIP) that allows mobile robots to organize the network nodes into co centric circular layers. Once the network is organized, the mobile robots are assigned layers to serve called home service layers where they act as moving probes to access the data and monitor the layers. Access positions are selected dynamically at each layer to provide anchors for the probes to visit in their home service layers. Probes cooperate to perform the application requests by executing a communication plan that is provided by the upper layer applications. The protocol is greatly able to cope with failures and requires only local updates for maintenance. We show that the proposed protocol provides a flexible infrastructure that keeps the nodes proximity and could be leveraged by upper layer protocols. To evaluate the performance of the proposed infrastructure some upper layer applications are implemented and built over the proposed infrastructure. Simulation-based results show the robustness and efficiency of the implemented applications.

Providing seamless handoff in session mobility is a challenging, and vital, issue relating to the heterogeneous next generation all-IP networks. The research work outlined by this paper describes a solution to support seamless video handoff in session mobility over IP multimedia subsystem (IMS) networks. Our system is built on top of an IMS network, which we believe to be the best platform for session mobility deployment, as it merges mobility and Internet services. The system combines two techniques to achieve seamless video handoff: It uses the SIP REFER method, with a proposed prediction handoff manager that predicts the required handoff time to transfer the session completely. The proposed system is implemented using PJSIP framework, an open source of SIP. Results show effective seamless handover in session mobility over IMS networks.

This paper proposes a new fault-resilient service overlay for MediaPort resource discovery that allows services to be efficiently and accurately located. MediaPorts are network-side functions used in the path between the source (media server) and the sink (media client). MediaPorts enable the adaptation of media content by providing value-added services such as caching, synchronization, and special routing functions. Our new approach addresses the problems of inefficiency and large message overhead, both typical of traditional approaches to resource discovery. The approach is based on a widely studied family of chordal rings, called the optimal chordal ring. Our solution is based on the types of services offered and also on the geographical locations of nodes. Extensive simulation results are presented to validate the effectiveness of the new approach, when compared to several other service-discovery solutions.

Overlay management systems face the challenges of increased complexity and heterogeneity because of the many elements involved in providing overlay services. In dynamic networks, the challenge and complexity is increased. Service composition allows simple services to be dynamically combined into more complex services in order to provide new services. In this paper, we deal with the problem of composing multiple autonomic elements to achieve system wide goals. Using a self-organizing approach, autonomic entities are dynamically and seamlessly composed into service-specific overlay networks. The paper describes and discusses the details of the composition approach. Results of extensive simulation are presented.

This paper investigates the efficiency of diagnosing network anomalies using concepts of statistical analysis and evidential reasoning. A bi-cycle of auto-regression is first applied to model increments in the values of network monitoring variables to accurately detect network anomalies. To classify the rootcause of the detected anomalies, concepts of evidential reasoning of Dempster-Shafer theory are employed; the root-cause of a network failure is inferred by gathering pieces of evidence concerning different groups of candidate failures obtained from a training set of detected anomalies and their corresponding root-causes. These groups are then refined to infer the exact cause of failure when evidence accumulates using the Dempster rule of combinations. To handle cases of imbalanced training sets, two new approaches for assigning belief values to different anomaly classes are also proposed. Performance analysis and results demonstrate the accuracy of the proposed scheme in detecting anomalies using real data.

This paper proposes a context-aware policy-based architecture that automates the management of overlay networks. The objective of this architecture is to form a policy layer above the overlay layer. The policy layer consists of a set of overlay policy enforcement points and an overlay policy decision point that cooperate to dynamically adapt the behavior of the overlay network to the preferences of the user, the network and the service providers. With this contribution to overlay networks, policies are generated dynamically at run time from the available context information and enforced on the fly. This paper describes our novel architecture, its advantages, and the simulation results that varify its effectiveness.

In this paper, we introduce a hybrid scalable approach for data-gathering and dissemination in sensor networks. The approach maintains the desirable on-demand communication feature of the pull technique by synchronizing its underling push and pull components. At the same time, scalability is achieved by allowing sinks to cooperatively embed virtual network sectorizing. The mechanism has two phases. First, virtual boards are constructed to form virtual closed sectors where queries are announced. Second, nodes located outside these boards initiate local bridges to carry their active events to the first board they encounter. Bridges are only created when boards are constructed. We validate our proposed solution with a simple analytical model and extensive software simulation. The initial investigation shows that our integrated protocol promises an efficient and scalable paradigm for data-gathering and dissemination in sensor networks that reduces the communication cost and leads to load balancing.

This paper describes PACMAN, a policy-based architecture for context management in ambient networks. The proposed architecture is a middleware that interacts with entities responsible for discovering and gathering various context in the execution environment and provides a unified mechanism for the delivery of appropriate contextual information to services and applications. PACMAN utilizes a novel context model based on policy structures to provide a coherent methodology for the representation, dissemination of both simple and complex context. A set of control policies is utilized to realize core management functionalities such as context processing (e.g., aggregation and filtering), storage, access and dissemination. Through the dynamic generation of policies, PACMAN evolves to accommodate new context types as well as emerging applications requirements.

Abstract One of the main focuses of Grid computing is solving resource‐sharing problems in multi‐institutional virtual organizations. In such heterogeneous and distributed environments, selecting the best resource to run a job is a complex task. The solutions currently employed still present numerous challenges and one of them is how to let users know when a job will finish. Consequently, reserve in advance remains unavailable. This article presents a new approach, which makes predictions for job execution time in Grid by applying the case‐based reasoning paradigm. The work includes the development of a new case retrieval algorithm involving relevance sequence and similarity degree calculations. The prediction model is part of a multi‐agent system that selects the best resource of a computational Grid to run a job. Agents representing candidate resources for job execution make predictions in a distributed and parallel manner. The technique presented here can be used in Grid environments at operation time to assist users with batch job submissions. Experimental results validate the prediction accuracy of the proposed mechanisms, and the performance of our case retrieval algorithm. Copyright © 2006 John Wiley & Sons, Ltd.

This paper presents a new approach for autonomic management of heterogeneous communication systems by featuring awareness of the surrounding environment's context. The contribution of this work is two-fold; the first is the development of an overall management infrastructure which identifies the necessary components for future autonomic management systems, namely, a context plane, an adaptive policy plane, a logic plane and an event management plane. The second contribution is a new model of context that is based on event calculus and a new methodology for context inference based on circumscriptive reasoning. The model lends itself easily to an automated deduction process on primitive context knowledge to derive more complex context that can be used to efficiently guide the automated management operations.

This paper presents a novel framework for user mobility prediction that can accurately predict the traveling trajectory and destination using knowledge of user's preferences, goals, and analyzed spatial information without imposing any assumptions about the availability of users' movements history. Using concepts of evidential reasoning of Dempster-Shafer's theory, the user's navigation behavior is captured by gathering pieces of evidence concerning different groups of candidate future locations. These groups are then refined to predict the user's future location when evidence accumulate using Dempster rule of combination. Simulation results are presented to demonstrate the performance of the proposed framework.

Computing and networking technologies are becoming ever more pervasive in order to support their users' increasingly dynamic lifestyles. These technologies aim to increase the networks' awareness of their users' requirements. At the same time, they strive to reduce the amount of manual reconfiguration and explicit interaction between the users and the network resources. The main problem is the extreme difficulty in reconfiguring the network's resources at runtime and in adapting the network's behaviour automatically according to the changes happening in the surroundings. To address these issues, we investigated the use of the emerging semantic Web technologies, policies, and context information. We also developed an ontology-based reasoning machinery to address the problem of automated adaptability. In this paper, we provide an analysis of the ontologies we developed and the reasoning machinery required increasing the network's awareness of its context. We report on the experiments we conducted to evaluate the techniques we are proposing.

This paper presents a novel paradigm to approach the issue of autonomous policy-based management of wired/wireless differentiated communication systems. In contrast to existing management approaches which require static a priori policy configurations, policies are created dynamically. The proposed framework addresses the management issue from a new perspective through posing it as a problem of learning from current system behavior, while creating new policies at runtime in response to changing requirements. A hierarchical policy model is used to capture users and administrators' higher level goals into network level objectives. Given sets of network objectives and constraints, policies are assembled at runtime. The new approach gives more flexibility to users and applications to dynamically change their quality-of-service (QoS) requirements while maintaining a smooth delivery of QoS through network monitors feedback. Simulation results demonstrate the performance of the proposed work.

User Mobility prediction represents a key component in assisting handoff management, resource reservation, and service preconfiguration. However, most of the existing approaches presume that the user travels in an a priori known pattern with some regularity; an assumption that may not always hold. This paper presents a novel framework for user mobility prediction that can accurately predict the traveling trajectory and destination using knowledge of user's preferences, goals, and analyzed spatial information without imposing any assumptions about the availability of users' movements history. This framework thus incorporates the notion of combining user context and spatial conceptual maps in the prediction process. The main objective of this notion is to circumvent the difficulties that arise in predicting the user's future location when adequate knowledge about the history of user's traveling patterns is not available. Using concepts of evidential reasoning of Dempster-Shafer's theory, the user's navigation behavior is captured by gathering pieces of evidence concerning different groups of candidate future locations. These groups are then refined to predict the user's future location when evidence accumulates using the Dempster rule of combination. Simulation results are presented to demonstrate the performance of the proposed framework.

This paper proposes an agent-based service provisioning system for mobile users. It describes a set of cooperative agents distributed over different sites that work together to provide personalized services for mobile users over the Internet. Users moving outside the office are able to maintain an off ice-like environment at home, or at temporary locations such as a meeting at another company, a business trip, or a hotel. Agents representing the end-users and the system agents engage in a negotiation process to facilitate access to personalized services at other sites. This access is obtained in accordance with the users' home policies as well as those at their current location. An Adaptive Service Presentation agent is used to adapt the service presentation to the capabilities of the users' workstations, laptops, phones, PDAs, or other devices. This work is conducted in Canada as part of the Mobile Agent Alliance project involving the University of Ottawa, the National Research Council, and the Mitel Corporation.

The work presented is part of a Web mining agent (WMA) system under development at our Multimedia and Mobile Agent Research Laboratory. The purpose of this system is to automatically extract specific information from Web pages and appropriately format the extracted information for further use. This requires resolving problems related to the disorganized nature of the Web that may result from ill-formatted HTML-based Web pages. The desired information is extracted from the Web documents by applying a sequence of filters to these documents. Each of the filters has a specific role. We discuss the filter that is used to convert Web documents into well-formed XML documents. This conversion involves the following operations: (i) syntactic mapping of HTML to XML, (ii) resolving ambiguity introduced by HTML tagging rules, and (iii) handling errors that may occur due to improper usage of HTML by the authors. The paper presents an overview of the Web mining agent system, then gives the motivations for the conversion into XML and finally, discusses in detail the transformation process performed on the Web documents.

A new class of protocols like RTP, RTSP and RSVP are tailoring today's packet networks for streaming real time multimedia data that once carried only non real time traffic. Multimedia networking is now possible with the existing network infrastructure. We describe a real time multimedia presentation system for media on demand over the Internet. It uses RTP and is capable of real time video/audio transmission. Our system consists of servers that maintain video/audio clips and clients that retrieve and present them to the users. It provides VCR like control features for user interactivity during playback. The paper describes the system architecture, and the communication protocols, along with media synchronization and dynamic QoS control mechanisms.

We describe a broadband multimedia telelearning system under development in the Multimedia Information Research Laboratory at the University of Ottawa. The system aims at providing a seamless multimedia telelearning platform using the latest telecommunication and multimedia information processing technologies. The main technology components of the system are: media production center, courseware author system, multimedia database, courseware navigator, and on-line facilitator. All these components are distributed over an ATM network and work together to offer a multimedia interactive courseware service. An MHEG-based model is developed to achieve the interchange of reusable information between these components.

We discuss a broadband multimedia telelearning system under development in the Multimedia Information Research Laboratory at the University of Ottawa. The system aims at providing a seamless environment for telelearning using the latest telecommunication and multimedia information processing technology. It basically consists of a media production center, a courseware author site, a courseware database, a courseware user site, and an online facilitator site. All these components are distributed over an ATM network and work together to offer a multimedia interactive courseware service. An MHEG-based model is exploited in designing the system architecture to achieve the real-time, interactive, and reusable information interchange through heterogeneous platforms. The system architecture, courseware processing strategies, courseware document models are presented.

Multimedia information management and communications systems for medical applications have been undergoing design and development. The aim of these activities has been the development of distributed computer systems providing storage, processing, and communication services required by the medical community. One of the main critical issues of such systems is the handling of multimedia information (i.e., text, images, graphics, and voice) in a uniform way and the fast access to images through the network. We present the design architecture and concepts used in our multimedia database system, along with its performance evaluation. The evaluation focuses on the response times of a multimedia medical database server, as seen by remote workstations for various retrieval and image transfer requests. For this purpose, a queueing network model was developed and simulated. Communications between the workstations and the database server are governed by the client-server model using TCP/IP and appropriate application protocols on a 10 Mb/s Ethernet LAN.>

The authors present a study whose main objective has been to explore the use of high-bandwidth multimedia communications technology for medical applications physically distributed among hospitals, that is to say an interhospital multimedia communications system. The design and development issues of a prototype are described, as well as the results of an in-field trial. From the results of this trial, a performance study has been developed for evaluating the expected communications requirements of a large-scale configuration of the system.>

The problem of communicating the radiologist's findings to the attending physician is introduced, and it is shown why it is essentially a multimedia communication problem. A multimedia communications system that was designed with a focus on database and medical reports organization and architecture is described. A report is presented of some of the key results of an in-house trial linking the Department of Radiological Sciences and the Emergency Department at the Ottawa Civic Hospital, Canada, a 950-bed tertiary-care teaching hospital.>