Network slicing has been a key supporting technology for resource management in modern computer networks, from 5G to 6G and beyond. Network slicing creates virtual networks over one common physical infrastructure. The dynamic resource allocation for services and slices to serve demand variations caused by user mobility is a challenge in 5G networks. Based on the described scenario, this thesis developed new resource management mechanisms for the dynamic network slicing allocation problem to support mobile users in 5G/edge computing infrastructures. The contributions of this work include: 1) one evaluation of the performance of dynamic network slicing in user mobility support scenarios; 2) a mathematical model which describes the network slices reconfiguration delays; 3) two heuristic solutions for improving the network slice reconfiguration process; and 4) a network simulator for evaluating the proposed solutions. The simulator is available as free software and has been adopted as the validation environment by the scientific community. The results of this work indicate the positive impact of slices' dynamic resource allocation in dealing with demand variations triggered by mobile users. However, the frequency of those reconfigurations is crucial to the success of such a solution.

The growing adoption of drones for goods delivery has emerged as a potentially viable solution. By operating through aerial routes, drones significantly reduce delivery times and expand operational reach. However, covering large areas requires prolonged flights, leading to high battery consumption and an increased risk of collisions, particularly in densely populated regions. This study presents a Stochastic Petri Net model to evaluate drone performance, focusing on metrics such as utilization, delivery rate, mean mission time, and drop probability. Additionally, energy consumption and carbon footprint metrics were investigated to assess the environmental impact of drone operations. The model incorporates factors such as strategic recharging points and collision probability, providing insights into drone performance under high-demand scenarios.

The rise of mobile devices and growing concerns about model privacy have posed significant challenges in distributed artificial intelligence, especially due to the heterogeneity of devices, leading to model generalization and resource management issues. Federated Learning (FL), a method where machine learning models are trained collaboratively by sharing only local parameters with an aggregation server, faces challenges in model convergence, optimization, and communication overhead due to this heterogeneity. This paper introduces FedSCCS, an FL-based framework designed for such heterogeneous settings. FedSCCS clusters devices based on the similarity of their models, allowing for efficient model aggregation and improved resource utilization. Our evaluation, set against established benchmarks, shows that FedSCCS achieves superior accuracy compared to existing methods, indicating a promising direction for scalable and tailored FL solutions.

One key challenge in Adaptive Video Streaming is the ever-changing edge network conditions at the last mile of access networks. The edge environment is particularly dynamic, influenced by user locations, fluctuation in resource demands and resource capabilities, in contrast to traditional Content Delivery Network (CDN) setups, where content routing decisions are relatively known. To address the dynamism of edge computing environments and to enable applications to better exploit edge-cloud computing resources, this article focuses on content steering technology, a recent addition to adaptive video protocols such as HLS and DASH. We present AVENUE as an architecture for Content Steering Services to orchestrate video delivery dynamically across the Edge-Cloud Continuum. This work designs the principles of the Content Steering Service to create a mechanism that involves two modules - monitoring and selector: The monitoring module captures real-time context metrics, and the selector module chooses an edge server according to the Select Server Algorithm. Our study addresses three steering algorithms with different performance profiles. Numerical results demonstrate that different configurations may yield varying network performance in terms of Quality of Experience (QoE), cache hits, and request load. Moreover, the appropriate selection of heuristics in the Selector module can also have a significant impact, depending on the metric being evaluated.

Federated Learning (FL) has emerged as a privacy-preserving powerful tool in decentralized Machine Learning (ML) environments. However, real-world scenarios often face bandwidth limitations that can be overwhelmed when all clients simultaneously perform training and communicate with the server in a federated system. Consequently, selection mechanisms are critical for identifying optimal subsets of clients to participate in the federation. Traditional selection methods, however, typically do not allow clients the autonomy to decide whether or not to contribute to the federation. Therefore, this paper proposes LetsFed, a client selection framework that respects client independence throughout the training process. The LetsFed framework differentiates between participating and non-participating clients, employing targeted selection mechanisms to address system challenges effectively. Empirical results demonstrate that LestFed can outperform, in dynamic client participation environments, literature solutions by up to 40%, reducing unnecessary data transmission by as much as 29%, while also enhancing the efficacy of the selection process.

The investigation of stateful service mobility across networking infrastructures is becoming increasingly important as applications require stateful services capable of migrating from centralized cloud data centers to edge computing infrastructures. State-of-the-art approaches propose either machine learning solutions for stateless service placement or stateful service mobility using static and inflexible state management strategies. We believe these approaches fall short of addressing the full length of the stateful service mobility problem. In this paper, we revisit an emerging concept named self-distributing systems, where a local executing application manages to detach some of its constituent (often stateful) components and place them in remote machines as a solution for stateful service mobility. In previous work, a machine learning approach to support self-distributing systems has not been thoroughly investigated. We model the distribution of stateful components across networking infrastructures as a multi-armed bandits problem and use the UCB1 algorithm to solve it as a first attempt at a flexible solution for stateful service mobility. We conclude the paper by discussing the main challenges and opportunities in this area.

As delivery companies continue to explore the use of drones, the need for efficient and safe operation in urban environments becomes increasingly critical. Market-wide versions of drone delivery services will necessarily spread many drones, especially in big cities. In this scenario, avoiding collisions with other drones or typical obstacles in urban spaces is fundamental. This paper proposes and evaluates, via simulation and analytical modeling, an aerial delivery service scenario and three autonomous geometric approaches for collision avoidance. We compare our approaches with three simple methods – DoNothing (not detouring), Random, and aviation-like Rightward – and two state-of-the-art geometric approaches. Simulation experiments consider different fleet sizes with constant and Poisson drone arrival rates and drones randomly choosing one of different altitudes for the cruise flight. Contrary to our expectations, the Random and Rightward approaches increase the collisions compared with DoNothing, making the latter our baseline. Our approaches significantly reduce collisions in all experiments and deal with more drones within the detection radius, showing that collisions are more complex to avoid. Comparing the collision rate, successful trips, and the number of flying drones reveals that the efficiency in avoiding collisions reduces the number of successful trips by increasing the number of active drones. Regardless of the expected reduction in collisions, more altitudes do not eliminate them. These results indicate the need for more sophisticated approaches to reduce or eliminate collisions. The analytical modeling using Markov Chains corroborates the simulation results by shedding some light on and helping explain the simulation results.

Federated learning (FL) is a distributed approach that enables collaborative training of a shared machine learning (ML) model for a given task. FL requires bandwidth-demanding communication between devices and a central server, which is a cause of many issues such as communication bottlenecks and scaling in the network. Therefore, we introduce the CCS (Compressed Client Selection) algorithm aimed at decreasing the overall communication costs for fitting a model in the FL environment. CCS employs a biased client selection strategy that reduces the number of devices training the ML model and the number of rounds required to reach convergence. In addition, the compression method Count Sketch is implemented to reduce the overhead in client-to-server communication. A use case on the Human Activity Recognition dataset is performed to evaluate CCS and compare it with other state-of-the-art approaches. Experimental evaluations show that CCS efficiently reduces the overall communication overhead for fitting a model and its convergence in a FL environment. In particular, CCS reduces up to 90% the communication overhead compared to literature approaches while providing good convergence even in scenarios where the data are not-independently and identically distributed among client devices.

Applications such as autonomous vehicles, virtual reality, augmented reality, and heavy machine learning-based applications are becoming popular and demanding more flexible deployment environments. The computing continuum, a hierarchical hybrid infrastructure comprehending user devices (smartphones, sensors, laptops, etc.), edge data centers, and cloud platforms, offers a wide range of deployment possibilities with a full range of varying computing resources. To take full advantage of such infrastructure, application development is faced with many challenges, the most important being the implementation of a transparent and generalized mechanism for code offloading and mobility throughout the continuum. To tackle such issues, this paper presents the Self-Distributing Systems (SDS) framework, a self-distribution framework that supports generalized code-offloading capabilities at the application level with a machine learning agent for deciding where to place components and a component-based model to enable seamless distribution of an application's components at runtime. We describe the framework, show its applicability in different application scenarios, and report our preliminary results. We conclude the paper with a list of challenges and invite the systems community to join the effort to further investigate them.

Nowadays, a crucial aspect of IoT networks management is traffic monitoring, where Machine Learning (ML) arose as a tool to perform several tasks, such as IoT device identification and network anomalies detection. However, the access to information about network traffic can affect users’ privacy, regarding devices and users identification and, consequently, violating existing privacy regulations. Thus, a solution to perform anomaly detection without compromising privacy is necessary. Within this context, this paper presents an ML solution to perform network anomaly detection based on anonymized network traffic, ensuring privacy when identifying the device and performing features selection. Experiments using a real IoT network traffic dataset indicate that it is possible to detect network anomalies with 99% accuracy while preserving privacy.

In this paper, we propose three multi-objective task scheduling algorithms for the cloud-fog continuum, that minimize both the makes pan and processing cost of workflows, considering the QoS requirements of the applications. Numerical results show that the scheduler based on Reinforcement Learning outperforms those based on classical optimization.

Delivering cloud-like computing facilities at the network edge provides computing services with ultra-low-latency access, yielding highly responsive computing services to application requests. The concept of fog computing has emerged as a computing paradigm that adds layers of computing nodes between the edge and the cloud, also known asmicro data centers,cloudlets, orfog nodes. Based on this premise, this article proposes a component-based service scheduler in a cloud-fog computing infrastructure comprising several layers of fog nodes between the edge and the cloud. The proposed scheduler aims to satisfy the application’s latency requirements by deciding which services components should be moved upwards in the fog-cloud hierarchy to alleviate computing workloads at the network edge. One communication-aware policy is introduced for resource allocation to enforce resource access prioritization among applications. We evaluate the proposal using the well-known iFogSim simulator. Results suggest that the proposed component-based scheduling algorithm can reduce average delays for application services with stricter latency requirements while still reducing the total network usage when applications exchange data between the components. Results have shown that our policy was able to, on average, reduce the overload impact on the network usage by approximately 11 percent compared to the best allocation policy in the literature while maintaining acceptable delays for latency-sensitive applications.

Internet access is crucial to the human society as a platform for several services to the users. Despite this importance, the Internet suffers limitations that compromise Quality of Service (QoS) guarantees. Thus, Internet Service Providers (ISPs) need to evolve, adding new technologies and management strategies to their infrastructure. A promising approach is the slicing of network resources among clients and delivered services, where reliability and elastic resource demand through the day are key issues. Within this context, this paper presents an algorithm called Reliable Reuse Encourage (R-REENC), which defines network slices based on bandwidth requirements and the desired reliability for the clients. The results suggest that the proposed algorithm allocates more suitable slices than other existing approaches.

Due to the rapid increase of IoT applications and their use in many different areas, large amounts of data have been generated to be processed and stored. In this scenario, some applications are sensitive to high latency and response times. In order to fulfil these requirements, Edge and Fog Computing appear with the objective of bringing processing and storage devices closer to applications and management mechanisms. In this context, due to limitations related to high cost, scalability and planning, several mechanisms and algorithms need to be simulated before being implemented in the real world. This paper presents a comparison between two simulation tools and their main characteristics (EdgeCloudSim and iFogSim) using a smart campus scenario deployed at the University of Campinas, where the sensors collect data from water meters and smart energy marker watches, in addition to smart public transportation and battery disposal bins. Our evaluation shows that the information processing in edge and fog can efficiently serve the applications, however, each simulation tool has its specificities, and should be used according to the researcher's objectives and needs.

In hybrid clouds, inter-cloud links play a key role in the execution of jobs with data dependencies. Insufficient available bandwidth in inter-cloud links can increase the makespan and the monetary cost to execute the application on public clouds. Imprecise information about the available bandwidth can lead to inefficient scheduling decisions. This paper attempts to evaluate the impact of imprecise information about the available bandwidth in inter-cloud links on workflow schedules, and it proposes a mechanism to cope with imprecise information about the available bandwidth and its impact on the makespan and cost estimates. The proposed mechanism applies a deflating factor on the available bandwidth value furnished as input to the scheduler. Simulation results showed that the mechanism is able to increase the number of solutions with makespans that are shorter than the defined deadline and reduce the underestimations of the makespan and cost provided by workflow schedulers.

The Internet became the primary medium for content sharing, playing a central role in our modern lives. This new paradigm changed the behavior of the user and the characteristics of the applications, such as multimedia content, real time interaction, among others. Thus, the Internet Service Providers (ISPs) need to evolve, adding new technologies and management strategies to their network infrastructures, resulting in the Future Internet Service Providers (FISPs). Within this context, this paper presents the key technical trends and challenges for the management of FISPs, considering the requirements of user's applications and the technologies to be standardized. Additionally, we introduce an architectural framework, called Framework for Integrated Management of Future Internet Service Providers (FIM), to expand the key technologies and to address the challenges in both access and edge networks, and to integrate the management of the FISP. A case study is presented to show the suitability of the proposed framework to enhance the management capacity of FISPs.

Nowadays, the Internet is part of our lives, as an essential way of communication and information. However, it suffers from limitations that prevent guarantees of Quality of Service (QoS) to the users. Thus, Internet Service Providers (ISPs) need to evolve, adding new technologies and management strategies, such as Network Virtualization (NV), Software Defined Network (SDN) and Network Function Virtualization (NFV), resulting in the Future Internet Service Providers (FISPs). FISPs slice their network resources among clients and delivered services, where an algorithm to define the structure of each slice is crucial to support Quality of Service (QoS) and resource utilization. Therefore, FISPs should consider the elastic resource utilization of the clients within periods of time through the day. In this context, this paper presents the Per Time Calculation (PETIC) algorithm to define slices based on the daytime bandwidth requirements of the clients. Numerical results suggest that the proposed algorithm defines more suitable slices than existing approaches.

Fog computing provides a low latency access to resources at the edge of the network for resource-constrained devices. The high mobility of some of these devices, such as vehicles, brings great challenges related to resource allocation and management. In order to improve the management of computing resources utilized by mobile users connected to the Fog infrastructure, this paper proposes a virtual machine placement and migration decision model based on mobility prediction. Simulations have shown that moving the virtual machine to a Fog node ahead of the user's route using the proposed approach can decrease by almost 50% the number of migrations needed by the user. The Fog architecture provides an average latency of about 15 milliseconds for the users' applications and the proposed approach presents a lower latency compared to a greedy approach for the VM placement problem.

Imprecise input data imposes special challenges to workflow scheduling. This paper introduces a robust scheduler based on particle swarm optimisation, called RobWE, which considers uncertainties of available bandwidth when producing schedules for workflow ensembles. The proposed scheduler is also a flexible scheduler since it allows the replacement of its objective function according to the user's needs. The effectiveness of the proposed RobWE scheduler is compared to a non-robust scheduler that does not consider the presence of such uncertainties. Results of simulations considering diverse scenarios, based on several degrees of uncertainty in available bandwidth estimates, characteristics of bandwidth estimations and workflow applications, demonstrate the advantages of the proposed RobWE scheduler.

This paper presents the contributions of the phd thesis “Mechanisms for management of SLA for Virtual Software Defined Networks based on QoS Classes”. Over the years the Internet has become the primary means of communication, where the users expect to access the Internet anytime, anywhere, and with a certain quality level. As a consequence, in the past few years the traffic demand to Internet access has increased. However, since the current Internet does not guarantee Quality of Service (QoS) and Quality of Experience (QoE), the companies apply a Service Level Agreements (SLA) with Internet Service Providers (ISPs). In this way, the Internet service providers aim to improve utilization of network resources and service delivery to the users through the deployment of Virtual Networks (VNs) over Software Defined Networks (SDNs), called virtual software defined networks (VSDNs). However, the management of this type of environment still presents open issues. In this context, a set of mechanisms to manage VSDN was developed. The objectives of the developed mechanisms are to negotiate, to deploy and to adapt the VSDNs according to the current state of the network infrastructure and the SLA definition.

As multimedia traffic provides the dominating data flows on Internet channels nowadays, on-demand video streaming has also grown in popularity. Users watching on-demand streams are interested in receiving their video streams without interruption and at low cost. To achieve this, video providers need to build computational infrastructure that can adapt to changes in demand. This paper presents an architecture to build synergy between Peer-to-Peer and Cloud Computing systems to achieve the necessary throughput for video-on-demand providers with reduced costs. We propose heuristics that overcome lack of stability in P2P systems by the inclusion of cloud servers into the pool of video streaming servers, avoiding disruptions (stalls) even when the total number of servers are reduced. Simulation results show that the proposed heuristic strategies can help reduce costs while maintaining quality of service.

Even after decades, the Internet does not provide Quality of Service (QoS) guarantee for the users. In order to soften such situation, clients negotiate Service Level Agreements (SLAs) with their Internet Service Providers (ISPs), where reliability and available bandwidth are key requirements for Internet services. On the other hand, the ISPs aim to decrease the energy consumption to deliver their services and provide a green environment. The mixing of Software Defined Networks (SDNs) and Virtual Networks (VNs), called Virtual Software Defined Network (VSDN), arises as a realizable option to deal with both client and provider perspectives. Within this context, this paper proposes two algorithms to allocate VSDNs combining reliability, bandwidth, and energy efficiency, considering the current network state to adjust the importance of each of these aspects during the allocation process. The results show the effectiveness of the proposed algorithms in allocating reliable VSDNs, while improving the energy efficiency and bandwidth usage of the ISP.

The emergence of Big Data has had profound impacts on how data are stored and processed. As technologies created to process continuous streams of data with low latency, Complex Event Processing (CEP) and Stream Processing (SP) have often been related to the Big Data velocity dimension and used in this context. Many modern CEP and SP systems leverage cloud environments to provide the low latency and scalability required by Big Data applications, yet validating these systems at the required scale is a research problem per se . Cloud computing simulators have been used as a tool to facilitate reproducible and repeatable experiments in clouds. Nevertheless, existing simulators are mostly based on simple application and simulation models that are not appropriate for CEP or for SP. This article presents CEPSim , a simulator for CEP and SP systems in cloud environments. CEPSim proposes a query model based on Directed Acyclic Graphs (DAGs) and introduces a simulation algorithm based on a novel abstraction called event sets. CEPSim is highly customizable and can be used to analyse the performance and scalability of user-defined queries and to evaluate the effects of various query processing strategies. Experimental results show that CEPSim can simulate existing systems in large Big Data scenarios with accuracy and precision.

The use of Complex Event Processing (CEP) and Stream Processing (SP) systems to process high-volume, high-velocity Big Data has renewed interest in procedures for managing these systems. In particular, self-management and adaptation of runtime platforms have been common research themes, as most of these systems run under dynamic conditions. Nevertheless, the research landscape in this area is still young and fragmented. Most research is performed in the context of specific systems, and it is difficult to generalize the results obtained to other contexts. To enable generic and reusable CEP/SP system management procedures and self-management policies, this research introduces the Attributed Graph Rewriting for Complex Event Processing Management ( AGeCEP ) formalism. AGeCEP represents queries in a language- and technology-agnostic fashion using attributed graphs. Query reconfiguration capabilities are expressed through standardized attributes, which are defined based on a novel classification of CEP query operators. By leveraging this representation, AGeCEP also proposes graph rewriting rules to define consistent reconfigurations of queries. To demonstrate AGeCEP feasibility, this research has used it to design an autonomic manager and to define a selected set of self-management policies. Finally, experiments demonstrate that AGeCEP can indeed be used to develop algorithms that can be integrated into diverse CEP systems.

It is a consensus that the Internet suffers from architectural limitations, including resilience, scalability, and manageability, among others. Therefore, companies access the Internet by establishing a service level agreement (SLA) in the attempt to ensure quality of service (QoS) for users. To address the current limitations of the Internet, researchers have recently proposed the edge as a service (EaaS) paradigm as a suitable solution to improve the access capacity of edge networks. EaaS uses network virtualization and software-defined networks to expand flexibility and manageability of access to edge network resources. Moreover, to maintain QoS assurance for users, EaaS addresses network events (such as traffic overload, failures, etc.) that can potentially affect QoS. Within this context, this paper proposes a software-defined management of EaaS environment, called SDM-EaaS. The proposal enhances the QoS for the end user, while improving the utilization of network resources in dynamic scenarios allowing for unpredictable changes in traffic demands, network infrastructure availability, and customer characteristics. Experiments based on emulation as well as real testbed demonstrate the effectiveness of the SDM-EaaS strategy.

The Internet is now present in our personal and professional life, where reliability and green communications are still open issues to be addressed. Usually, clients establish Service Level Agreements (SLAs) with their Internet Service Providers (ISPs), including parameters of reliability and bandwidth. On the other hand, the ISPs aim to maximize their profit by establishing as many SLAs as possible and decreasing the energy consumption to provide their services. In this context, this paper presents an algorithm for Virtual Software Defined Network (VSDN) allocation that considers both reliability and energy efficiency. The algorithm deploys the VSDN aiming to comply with the parameters defined in SLAs, as well as to improve the energy efficiency of the ISP. Experiments using a real network topology show the effectiveness of the algorithm to deploy the VSDN with a performance 15% better than existing algorithms.

Nowadays, the multimedia content accessed by users represents 55% of the Internet traffic and it is forecast to expand to 92% by 2020. This fact increased the resource demand in access networks across the world, since multimedia traffic has high network requirements. To deal with this scenario, Internet Service Providers (ISPs) tend to apply the Edge as a Service (EaaS) approach. EaaS aims to use Network Virtualization and Network Programmability to expand the flexibility of access to edge network resources. In this context, this paper presents the Bitrate-Aware Virtual Resources Adjustment (BEAVER) mechanism, which increases the Quality of Experience (QoE) support for multimedia clients in EaaS environments. Experiments using real traffic demonstrate the effectiveness of the proposed mechanism to ensure the user's QoE.

Cities with a large number of people are currently facing urban mobility problems, especially the problem of traffic congestions. This not only has an adverse effect on the economy of the city, but also impairs the quality of life of its citizens. One measure that can be adopted to mitigate these problems is the use of systems that help identify, reduce, and/or avoid these traffic jams, such as intelligent transport systems. In this context, we propose an intelligent traffic information system called UCONDES, which is based on inter-vehicle communications and can be applied to detect and reduce congestion in urban centers. Simulation results shows that, when compared to original vehicular mobility trace, our solution reduces the average trip time, and the overall CO2 emission and fuel consumption. More specifically, the average travel time for drivers was reduced by approximately 26%, resulting in a reduction of fuel consumption by 23% and the CO2 emission by 25%.

The improvement in virtualisation technologies, as well as higher data transmission rates and the popularisation of the Internet contributed to strengthen the cloud computing. This paradigm attracts more and more users to consume, and providers to offer services through the pay-as-you-go model. The users can benefit from the clouds by receiving high availability, cost reduction, load balancing, and better fault tolerance when deploying cloud services across several cloud providers instead of using a single service provider. However, the variety of charging models, quality of service (QoS), and monetary costs of virtual machines, makes it difficult to the customer to choose the best resource charging models to deploy applications. In this paper we propose a redundancy strategy based on the tenancy level among requests in order to decrease the scheduling cost and maintain QoS levels along different IaaS charging models. The strategy is modelled as an integer linear program (ILP) to compute the lowest cost when utilising virtual machines. Simulations show that the proposed approach computes schedules with smaller costs than alternative approaches.

Currently, the Internet is the main communication medium, however it does not guarantee Quality of Service. To contour this problem, companies establish Service Level Agreements with their Internet Service Providers, including parameters like bandwidth and reliability of the network. This paper presents an algorithm for virtual network allocation in a reliability context, where a feasible bandwidth approach and reliability calculation are used to deploy the virtual network according to the client's requirement. Simulations using a real network topology show the effectiveness of the algorithm to deploy the virtual network in a scenario of multiple cumulative requests.

In hybrid clouds, the available bandwidth in inter-cloud links is quite variable. Overestimating the available bandwidth on theses channels at scheduling time can enlarge the makespan and cause deadline misses. In this paper, we propose a procedure for deflating the estimated available bandwidth used as input to cloud schedulers since schedulers are not usually designed to cope with inaccurate information on available bandwidth. The procedure is based on a multiple linear regression procedure which utilizes historical information of previous executions of workflows. Results showed that the proposed procedure can increase the number of valid schedules without increasing the makespan and cost estimations, regardless the variability in the available bandwidth during the execution of an application workflow.

To access the Internet, companies define a Service Level Agreement (SLA) with Internet Service Providers (ISPs). Nevertheless, the current Internet does not assure Quality of Service (QoS), what points toward the concept of network virtualization and software defined network (SDN) to support the Future Internet. Moreover, the initial parameters defined in SLAs may not withstand traffic changes, resulting in a lower quality experienced by users and a waste of resources for ISPs. Within this context, this paper proposes an architecture to deal with resource usage variations. The proposal aims to enhance the QoS provided by the ISPs, while improving the usage of network resources in dynamic scenarios. Experiments using real traffic demonstrate the effectiveness of the proposed architecture.

To access the Internet, companies define a Service Level Agreement (SLA) with Internet Service Providers (ISPs). Nevertheless, the current Internet does not assure Quality of Service (QoS), what points toward the concept of virtual networks (VNs) and software defined network (SDN) to support the Future Internet. Moreover, the VN and SDN approaches can be mixed creating the Virtual Software Defined Network (VSDN), which enhances the customization capacity of the networks. Based on this context, this work proposes a similarity metric and a fuzzy decision-making model (FDMM) to support the VSDN negotiation. Experiments show the effectiveness of the proposal in providing the clients with a mechanism to compare protocols offered by providers, allowing them to identify which one best fulfills the requirements.

With the ever-increasing application demands, the execution of applications may require more resources than locally available. In this scenario, resources from multiple IaaS cloud providers can be leased to fulfill application requirements. In this paper we deal with the problem of scheduling workflow applications in multiple IaaS providers, where the workflow scheduler must determine on which computational resource each component of a workflow should be allocated in order to minimize the involved monetary costs. We propose the use of different levels of discrete-time intervals in linear programming to schedule workflows with deadline constraints in multiple cloud providers. Simulations shown that increasing the granularity level of time-discretization decreases the scheduler running time, although yet achieving good solutions.

The Internet has become a primary means of communication, even though it currently does not guarantee Quality of Service (QoS). In order to tackle with this problem, companies establish Service Level Agreements (SLA) with their providers, including the reliability of services. This paper shows an algorithm for virtual network allocation in a reliability context, where some techniques of disjoint path and reliability calculation are used to deploy the virtual network according to the client's specification. Simulations using real network topologies show the capacity of the algorithm to deploy the virtual network according to the client requirements.

The so-called hybrid cloud is the composition of an infrastructure that comprises private resources as well as public resources leased from public clouds. Hybrid clouds can be utilized for the execution of applications composed of dependent jobs, usually modeled as workflows. In this scenario, a scheduler must distribute the components of the workflow onto available resources considering the communication demands and the available bandwidth in network links. However, such information can be imprecise, and consequently decisions on resource allocation can be ineffective. In this paper, we evaluate scheduling algorithms in the face of imprecise information on the availability of communication channels. Results showed that schedules are negatively affected by the unforeseen variations in bandwidth during the execution of the application.

Cloud computing is being used to avoid maintenance costs and upfront investment, while providing elasticity to the available computational power in a pay-per-use basis. Customers can make use of the cloud as a software (SaaS), platform (PaaS), or infrastructure (IaaS) provider. When one customer utilizes an environment provided by a SaaS cloud, she is unaware of any details about the computational infrastructure where her requests are being processed. Therefore, such infrastructure can be composed of computational resources from a datacenter owned by the SaaS or its resources can be leased from a cloud infrastructure provider. In this paper we present an integer linear program (ILP) formulation for the problem of scheduling SaaS customer's workflows into multiple IaaS providers where SLA exists at two levels. In addition, we present heuristics to solve the relaxed version of the presented ILP. Simulation results show that the proposed ILP is able to find low-cost solutions for short deadlines, while the proposed heuristics are effective when deadlines are larger.

Cloud computing has recently emerged as a convergence of concepts such as cluster computing, grid computing, utility computing, and virtualization. In hybrid clouds, the user has its private cloud available for use, but she can also request new resources to public clouds in a pay-per-use basis when there is an increase in demand. In this scenario it is important to decide when and how to request these new resources to satisfy deadlines and/or to get a reasonable execution time, while minimizing the monetary costs involved. In this paper we propose a strategy to schedule service workflows in a hybrid cloud. The strategy aims at determining which services should use paid resources and what kind of resource should be requested to the cloud in order to minimize costs and meet deadlines. Experiments suggest that the strategy can decrease the execution costs while maintaining reasonable execution times.

Among the numerous DAG scheduling heuristics suitable for heterogeneous systems, the Heterogeneous Earliest Finish Time (HEFT) heuristic is known to give good results in short time. In this paper, we propose an improvement of HEFT, where the locally optimal decisions made by the heuristic do not rely on estimates of a single task only, but also look ahead in the schedule and take into account information about the impact of this decision to the children of the task being allocated. Preliminary simulation results indicate that the lookahead variation of HEFT can effectively reduce the makespan of the schedule in most cases without making the algorithm's execution time prohibitively high.

Abstract A scheduler must consider the heterogeneity and communication delays when scheduling dependent tasks on a grid. The task‐scheduling problem is NP‐Complete in general, which led us to the development of a heuristic for the associated optimization problem. In this work we present a dynamic adaptive approach to schedule dependent tasks onto a grid based on the Xavantes grid middleware. The developed dynamic approach is applied to the Path Clustering Heuristic, and introduces the concept of rounds, which take turns sending tasks to execution and evaluating the performance of the resources. The adaptive extension changes the size of rounds during the process execution, taking task attributes and resources performance as parameters, and it can be adopted in other task schedulers. The experiments show that the dynamic round‐based and adaptive schedule can minimize the effects of performance losses while executing processes on the grid. Copyright © 2007 John Wiley & Sons, Ltd.