Nowadays, one can observe the convergence of the Internet of Things (IoT) and Edge Computing (EC) infrastructures towards establishing a data collection and processing ecosystem in close proximity to end users. The aim is to enhance the performance of the supported applications by reducing the latency in data processing and service delivery. Various services can be employed to facilitate the execution of tasks prompted by end users or any type of external applications. Those services are mainly present at EC nodes that become the hosts of the data collected by IoT devices, the executors of the desired tasks and the intermediaries when transferring the discussed data to the Cloud back end. It is obvious that the implementation of an efficient framework for managing services across distributed edge nodes becomes imperative especially if we bear in mind that nodes are constrained devices and cannot host numerous services. In this paper, we introduce a proactive model designed to allocate the available services to core parts of the EC ecosystem based on the observed demand. This will give us the opportunity to determine ‘where’ to place any individual service putting it in locations (i.e., in EC nodes) where an increased demand is identified, while saving resources by restricting the number of nodes that become the final hosts (to avoid the flooding of the network). The paper delves into the evaluation of the proposed model, offering a comparative analysis with a baseline scheme utilizing real datasets. Through the envisioned experimental validation, the paper demonstrates that the proposed approach enhances the ability of diverse engaged edge nodes to accurately deduce the appropriate location for service placement.

Context-aware data selectivity in Edge Computing (EC) requires nodes to efficiently manage the data collected from Internet of Things (IoT) devices, e.g., sensors, for supporting real-time and data-driven pervasive analytics. Data selectivity at the network edge copes with the challenge of deciding which data should be kept at the edge for future analytics tasks under limited computational and storage resources. Our challenge is to efficiently learn the access patterns of data-driven tasks (analytics) and predict which data arerelevant, thus, being stored in nodes’ local datasets. Task patterns directly indicate which data need to be accessed and processed to support end-users’ applications. We introduce a task workload-aware mechanism which adopts one-class classification to learn and predict the relevant data requested by past tasks. The inherent uncertainty in learning task patterns, identifying inliers and eliminating outliers is handled by introducing a lightweight fuzzy inference estimator that dynamically adapts nodes’ local data filters ensuring accurate data relevance prediction. We analytically describe our mechanism and comprehensively evaluate and compare against baselines and approaches found in the literature showcasing its applicability in pervasive EC.

Reading, or the ability to infer meaning from printed words in order to correctly interpret relevant information, is the most fundamental component of education. To recognize letters, letter strings, and words, one must possess the ability to decode abstract graphemes accurately and fluently into their corresponding phonemes. Additionally, processing text requires the capacity to read and comprehend text with both fluency and accuracy. Thus, reading requires a variety of cognitive abilities, including effective processing speed, phonological awareness, syntactic processing, auditory and visual word recognition, and phonological awareness. The present study reports the outcomes of a research that evaluated the identification of students with reading disabilities using an artificial neural network. The neural network consisted of structured tasks aiming at a) reading, b) distinguishing words and pseudowords and c) reading two texts. Participants were 235 children attending grades from third to sixth class. Audio is converted into a spectrogram and students with disabilities are identified using machine learning algorithms and auditory analysis. The outcome of the present study suggests that a neural network that is comprised from three tasks can identify the reading abilities and classify the school aged children between typical achievers and reading disabled. Furthermore, including the mAP scores, the results show that the model is highly effective in identifying and classifying reading difficulties in real-time, offering a promising avenue for future research and practical applications in educational settings.

As semiconductor manufacturing technologies progress beyond the current 3nm, the demand for more compact and powerful VLSI circuits obliges on-chip power grid networks to become denser, resulting in a substantial increase in current densities. Consequently, Electromigration (EM) has emerged as a critical reliability concern since it can lead to voids on the metal wires and, consequently, large IR drops. In this paper, we present an EM/IR-aware wire sizing methodology based on the Particle Swarm Optimization (PSO) algorithm. Our methodology can be effectively applied to contemporary power grid networks to achieve the targeted lifetimes of the chip, and simultaneously resize the wires for area reduction. The advantage is that the proposed approach is able to deal with high-dimensional search spaces, which is imperative in our problem. Experimental results using the large-scale industrial IBM power grid benchmarks indicate that our new approach can increase the lifespan of the power grid up to 6.47 × while effectively reducing the area up to 65%.

Recognizing the severe health implications of dangerous pollutants emitted during a wildfire incident, we introduce a robust monitoring framework based on the Internet of Things (IoT) paradigm designed for real-time remote sensing of wildfire pollutants. The system is specifically adapted for measuring the wildfire health impact on firefighters and nearby residents. Its architecture comprises portable and stationary solutions and a Web application, enabling easy access to emission data and Air Quality Index (AQI) for interested parties and command & control centers. The portable solution benefits firefighters by providing real-time air quality information, aiding in decision-making for safety and efficient firefighting, while the stationary one contributes to mitigating the risk of potential evacuation in a region near the fire incident. Interested parties e.g., local authorities, command centers and environmental monitoring agencies, can easily and seamlessly integrate our system into their operations by initiating simple requests in our REST API.

Recently, context-aware data management becomes the focus of many research efforts placed at the intersection between the Internet of Things (IoT) and Edge Computing (EC). Huge volumes of data can be collected by IoT devices being ‘connected’ with EC environments transferring data towards the Cloud. EC nodes undertake the responsibility of managing the collected data, however, they are characterized by limited storage and computational resources compared to Cloud. Evidently, this makes imperative the introduction of data selectivity methods to keep locally only the data requested by end users or applications for current and future analytics services. In this paper, we study an EC environment where nodes rely on data selectivity and decide the allocation of newly received data to peers, or Cloud when these data are not conformed with local data filters. Data filters are the means for determining local data selectivity by keeping only data that statistically match the needs of nodes (e.g., match the already present data or requests for processing defined by incoming tasks). We contribute with data selectivity and filtering models that support intelligent decisions on when and where incoming data should be allocated. We intent to ‘postpone’ the transfer of data to the Cloud by keeping them close to end users. Our approach concludes a data map of an EC environment nominating every node as the owner of specific data (sub)spaces facilitating the placement of future processing tasks. We evaluate and compare our models and algorithms against schemes found in the literature showcasing their applicability and efficiency in pervasive edge computing environments.

Distributed Predictive Analytics (DPA) refers to constructing predictive models based on data distributed across nodes. DPA reduces the need for data centralization, thus, alleviating concerns about data privacy, decreasing the load on central servers, and minimizing communication overhead . However, data collected by nodes are inherently different; each node can have different distributions, volumes, access patterns, and features space . This heterogeneity hinders the development of accurate models in a distributed fashion. Many state-of-the-art methods adopt random node selection as a straightforward approach. Such method is particularly ineffective when dealing with data and access pattern heterogeneity, as it increases the likelihood of selecting nodes with low-quality or irrelevant data for DPA. Consequently, it is only after training models over randomly selected nodes that the most suitable ones can be identified based on the predictive performance . This results in more time and resource consumption, and increased network load. In this work, holistic knowledge of nodes’ data characteristics and access patterns is crucial. Such knowledge enables the successful selection of a subset of suitable nodes for each DPA task (query) before model training. Our method engages the most suitable nodes by predicting their relevant distributed data and learning predictive models per query. We introduce a novel DPA query-centric mechanism for node and relevant data selection. We contribute with (i) predictive selection mechanisms based on the availability and relevance of data per DPA query and (ii) various distributed machine learning mechanisms that engage the most suitable nodes for model training. We evaluate the efficiency of our mechanism and provide a comparative assessment with other methods found in the literature. Our experiments showcase that our mechanism significantly outperforms other approaches being applicable in DPA.

Abstract Fraud detection and prevention has received a lot of attention from the research community due to its high impact on financial institutions’ revenues and reputation. The increased use of the web and the provision of online services open up the pathway for exposing these systems to numerous threats and jeopardizing their effective functioning. Naturally, financial frauds are increased in number and form imposing various requirements for their efficient and immediate detection. These requirements are related to the performance of the adopted models as well as the timely response of the decision-making mechanism. Machine learning and data mining are two research domains that can provide a number of techniques/algorithms for fraud detection and setup the road for mitigation actions. However, these methods still need to be improved with respect to the detection of unknown fraud patterns and the incorporation of big data processing mechanisms. This paper presents our attempt to build a hybrid system, i.e., a sequential scheme for combining two deep learning models and efficiently detecting potential financial frauds. We elaborate on the combination of an autoencoder and a Long Short-Term Memory Recurrent Neural Network trained upon datasets which are processed through the use of an oversampling technique. Oversampling is adopted to handle heavily imbalanced datasets which is the ‘natural’ scenario due to the limited number of frauds compared to the humongous volumes of transactions. The proposed approach tends to capture much more fraud events in comparison with other conventional ML techniques. Our experimental evaluation exposes that our model exhibits a good performance in terms of recall and precision.

Distributed Machine Learning (DML) at the edge of the network involves model learning and inference across networking nodes over distributed data. One type of model learning could be the delivery of predictive analytics services to formulate intelligent environments, however, those environments heavily rely on real-time inference and are significantly influenced by changes in the underlying data (concept drifts). Moreover, the quality of service and availability in DML environments are directly tied to each node’s reliability, since such environments are highly susceptible to the impact of node failures. Even if such challenges can be tackled with distributed resilience mechanisms, their effectiveness and efficiency, due to concept drifts, should be maintained to ensure continuous and sustained quality of service. DML systems operate in dynamic environments, thus, they require their models to be updated according to the novel trends embedded in the new data they encounter. We, therefore, introduce several model maintenance mechanisms to ensure resilient DML systems in the long term when concept drifts emerge. We provide a comprehensive experimental evaluation of our resilience maintenance mechanisms over synthetic and real data showcasing their importance and applicability in edge learning environments.

The coexistence of the Internet of Things and Edge Computing aims to offer a processing infrastructure close to end users that will improve the performance of applications and limit the latency in the provision of services. Services are adopted to assist in the execution of tasks imposed by the requests of end users/applications. The implementation of an effective framework for services management in the distributed edge nodes is necessary to achieve the aforementioned goals. The discussed framework ought to address the trade-off between overheads related to services migration/replication and data transmission. In this paper, we propose a proactive statistical model for allocating the available services upon the observed demand and supporting edge nodes to decide when and where it is necessary to migrate/replicate them. Our aim is to place services at locations where an increased demand is observed, however, under the uncertainty about the future evolution of the incoming requests. We elaborate on the evaluation of the proposed model and provide a comparative assessment with relevant schemes adopting real datasets. Our experimental validation demonstrates that our approach reinforces the heterogeneous engaged edge nodes to correctly infer the time instance and the location when/where services should be migrated/replicated to meet the dynamics of their demand. The interesting is that the proposed model achieves encouraging outcomes when it is adopted to cope with the mobility of users.

To cope with the challenge of managing numerous computing devices, humongous data volumes and models in Internet-of-Things environments, Edge Computing (EC) has emerged to serve latency-sensitive and compute-intensive applications. Although EC paradigm significantly eliminates latency for predictive analytics tasks by deploying computation on edge nodes’ vicinity, the large scale of EC infrastructure still has huge inescapable burdens on the required resources. This paper introduces a novel paradigm where edge nodes effectively reuse local completed computations (e.g., trained models) at the network edge, coined as knowledge reuse. Such paradigm releases the burden from individual nodes, where they can save resources by relying on reusing models for various predictive analytics tasks (e.g., regression and classification). We study the feasibility of our paradigm by involving pair-wise (dis)similarity metrics among datasets over nodes based on statistical learning techniques (kernel-based Maximum Mean Discrepancy and eigenspace Cosine Dissimilarity). Our paradigm is enhanced with computationally lightweight monitoring mechanisms, which rely on Holt-Winters to forecast future violations and updates of the reused models. Such mechanisms predict when ‘borrowed’ models are insufficient for being reused, triggering a new process of finding more appropriate models to be reused at the network edge. We provide comprehensive performance evaluation and comparative assessment of our algorithms over different experimental scenarios using real and synthetic datasets. Our findings showcase the ability and robustness of our paradigm to maintain up-to-date reused models at the edge trading off quality of analytics and resource utilization.

Internet of Things offers the infrastructure for smooth functioning of autonomous context-aware devices being connected towards the Cloud. Edge Computing (EC) relies between the IoT and Cloud providing significant advantages. One advantage is to perform local data processing (limited latency, bandwidth preservation) with real time communication among IoT devices, while multiple nodes become hosts of the collected data (reported by IoT devices). In this work, we provide a mechanism for the exchange of data synopses (summaries of extracted knowledge) among EC nodes that are necessary to give the knowledge on the data present in EC environments. The overarching aim is to intelligently decide on when nodes should exchange data synopses in light of efficient execution of tasks. We enhance such a decision with a stochastic optimization model based on the Theory of Optimal Stopping. We provide the fundamentals of our model and the relevant formulations on the optimal time to disseminate data synopses to network edge nodes. We report a comprehensive experimental evaluation and comparative assessment related to the optimality achieved by our model and the positive effects on EC.

The combination of the Internet of Things (IoT) and Edge Computing (EC) can support intelligent pervasive applications that meet the needs of end users. A challenge is to provide efficient inference models for supporting collaborative activities. EC nodes can interact with IoT devices and each other to conclude those activities producing knowledge. In this paper, we propose a proactive scheme to decide upon the efficient management of services and tasks present/reported at EC nodes. Services can be processing modules applied upon local data while being required for the execution of tasks. We monitor the demand for the available services and reason upon their management, i.e., for their local presence/invocation as the demand is updated by the requested processing activities. For each incoming task, an inference process is fired to proactively meet the strategic targets of the envisioned model. We propose a statistical inference process upon the demand for services and the contextual performance data of nodes combining it with a utility aware decision making model. Instead of exclusively focusing on services migration or tasks offloading as other relevant efforts do, we elaborate on the decision making for the selection of one of the aforementioned activities (the most appropriate at a specific time instance). We present our model and evaluate it through a high number of simulations to expose its pros and cons placing it in the respective literature as one of the first attempts to proactively decide the presence of services to an ecosystem of processing nodes.

Data management at the edge of the network is a significant research subject. Devices being active at the Internet of Things (IoT) can collect data and transfer them to a set of edge nodes for further processing. There, various activities can be realized. Among them, of great importance it is the detection of anomalies in the incoming data and their preparation to be the subject of advanced processing tasks. In this paper, we propose an ensemble scheme for data anomalies detection and elaborate on the use of an extended sliding window approach. We differentiate from the state of the art solutions and argue on the concept of potential anomalies confirming their presence by incorporating more data into our decision mechanism. The performance of the proposed scheme is evaluated by a set of experimental scenarios being also exposed by numerical results.

Workflow management is a widely studied research subject due to its criticality for the efficient execution of various processing activities towards concluding innovative applications. The ultimate goal is to eliminate the required time for delivering the final outcome considering the dependencies between workflow’s tasks. In this paper, we enhance the decision making of a scheduler with a batch oriented approach to deal with multiple workflows. A probabilistic data oriented approach combined with an infrastructure oriented scheme is provided to pay attention on dynamic environments where the underlying data are continuously updated trying to minimize the network overhead for migrating data. Workflows are mapped to the available datasets according to their data requirements, then, we combine the outcome with an optimization model upon the time and cost requirements of every placement. The performance of our model is revealed by a high number of experiments depicting the advantages in the network overhead.

Tasks management is a very interesting research topic for various application domains. Tasks may have the form of analytics or any other processing activities over the available data. One of the main concerns is to efficiently allocate and execute tasks to produce meaningful results that will facilitate any decision making. The advent of the Internet of Things (IoT) and Edge Computing (EC) defines new requirements for tasks management. Such requirements are related to the dynamic environment where IoT devices and EC nodes act and process the collected data. The statistics of data and the status of IoT/EC nodes are continuously updated. In this paper, we propose a demand- and uncertainty-driven tasks management scheme with the target to allocate the computational burden to the appropriate places. As the proper place, we consider the local execution of a task in an EC node or its offloading to a peer node. We provide the description of the problem and give details for its solution. The proposed mechanism models the demand for each task and efficiently selects the place where it will be executed. We adopt statistical learning and fuzzy logic to support the appropriate decision when tasks' execution is requested by EC nodes. Our experimental evaluation involves extensive simulations for a set of parameters defined in our model. We provide numerical results and reveal that the proposed scheme is capable of deciding on the fly while concluding the most efficient allocation.

Edge Computing (EC) offers an infrastructure that acts as the mediator between the Cloud and the Internet of Things (IoT). The goal is to reduce the latency that we face when relying on Cloud. IoT devices interact with their environment to collect data relaying them towards the Cloud through the EC. Various services can be provided at the EC for the immediate management of the collected data. One significant task is the management of missing values. In this paper, we propose an ensemble based approach for data imputation that takes into consideration the spatio-temporal aspect of the collected data and the reporting devices. We propose to rely on the group of IoT devices that resemble to the device reporting missing data and enhance its data imputation process. We continuously reason on the correlation of the reported streams and efficiently combine the available data. Our aim is to `aggregate' the local view on the appropriate replacement with the `opinion' of the group. We adopt widely known similarity techniques and a statistical modelling methodology to deliver the final outcome. We provide the description of our model and evaluate it through a high number of simulations adopting various experimental scenarios.

The increased adoption of the Internet of Things (IoT) for the delivery of intelligent applications over huge volumes of data opens new opportunities to draw conclusions from data and support efficient decision making. For this reason many applications have been developed for data collection and processing. A large part of them are aligned with the requirements of the vast infrastructure of IoT. However, one of the biggest problems occurring at real-time applications is that they are prone to missing values. Missing values can negatively affect the outcomes of any processing activity, thus, they can limit the performance of IoT applications. In this paper, we depart from the relevant literature and propose a data imputation model that is based on the correlation of data reported by different IoT devices. Our aim is to support data imputation using the ‘knowledge’ of a team of IoT devices over their reports for various phenomena. Our scheme adopts a continuous correlation detection methodology applied at real time reports of the involved devices. Hence, any missing value can be replaced by the aggregated outcome of data reported by correlated devices. We provide the description of our approach and evaluate it through a high number of simulations adopting various experimental scenarios.

The Cloud infrastructure offers to end users a broad set of heterogenous computational resources using the pay-as-you -go model. These virtualized resources can be provisioned using different pricing models like the unreliable model where resources are provided at a fraction of the cost but with no guarantee for an uninterrupted processing. However, the enormous gamut of opportunities comes with a great caveat as resource management and scheduling decisions are increasingly complicated. Moreover, the presented uncertainty in optimally selecting resources has also a negatively impact on the quality of solutions delivered by scheduling algorithms. In this paper, we present a dynamic scheduling algorithm (i.e., the Uncertainty-Driven Scheduling - UDS algorithm) for the management of scientific workflows in Cloud. Our model minimizes both the makespan and the monetary cost by dynamically selecting reliable or unreliable virtualized resources. For covering the uncertainty in decision making, we adopt a Fuzzy Logic Controller (FLC) to derive the pricing model of the resources that will host every task. We evaluate the performance of the proposed algorithm using real workflow applications being tested under the assumption of different probabilities regarding the revocation of unreliable resources. Numerical results depict the performance of the proposed approach and a comparative assessment reveals the position of the paper in the relevant literature.

Abstract Lack of knowledge in the underlying data distribution in distributed large-scale data can be an obstacle when issuing analytics & predictive modelling queries. Analysts find themselves having a hard time finding analytics/exploration queries that satisfy their needs. In this paper, we study how exploration query results can be predicted in order to avoid the execution of ‘bad’/non-informative queries that waste network, storage, financial resources, and time in a distributed computing environment. The proposed methodology involves clustering of a training set of exploration queries along with the cardinality of the results (score) they retrieved and then using query-centroid representatives to proceed with predictions. After the training phase, we propose a novel refinement process to increase the reliability of predicting the score of new unseen queries based on the refined query representatives. Comprehensive experimentation with real datasets shows that more reliable predictions are acquired after the proposed refinement method, which increases the reliability of the closest centroid and improves predictability under the right circumstances.

Analysts wishing to explore multivariate data spaces, typically issue queries involving selection operators, i.e., range or equality predicates, which define data subspaces of potential interest. Then, they use aggregation functions, the results of which determine a subspace’s interestingness for further exploration and deeper analysis. However, Aggregate Query (AQ) results are scalars and convey limited information and explainability about the queried subspaces for enhanced exploratory analysis. Analysts have no way of identifying how these results are derived or how they change w.r.t query (input) parameter values. We address this shortcoming by aiding analysts to explore and understand data subspaces by contributing a novel explanation mechanism based on machine learning. We explain AQ results using functions obtained by a three-fold joint optimization problem which assume the form of explainable piecewise-linear regression functions. A key feature of the proposed solution is that the explanation functions are estimated using past executed queries. These queries provide a coarse grained overview of the underlying aggregate function (generating the AQ results) to be learned. Explanations for future, previously unseen AQs can be computed without accessing the underlying data and can be used to further explore the queried data subspaces, without issuing more queries to the backend analytics engine. We evaluate the explanation accuracy and efficiency through theoretically grounded metrics over real-world and synthetic datasets and query workloads.

The combination of Internet of Things (IoT) and Edge Computing (EC) can assist in the delivery of novel applications that will facilitate end-users’ activities. Data collected by numerous devices present in the IoT infrastructure can be hosted into a set of EC nodes becoming the subject of processing tasks for the provision of analytics. Analytics are derived as the result of various queries defined by end-users or applications. Such queries can be executed in the available EC nodes to limit the latency in the provision of responses. In this article, we propose a meta-ensemble learning scheme that supports the decision making for the allocation of queries to the appropriate EC nodes. Our learning model decides over queries’ and nodes’ characteristics. We provide the description of a matching process between queries and nodes after concluding the contextual information for each envisioned characteristic adopted in our meta-ensemble scheme. We rely on widely known ensemble models, combine them, and offer an additional processing layer to increase the performance. The aim is to result a subset of EC nodes that will host each incoming query. Apart from the description of the proposed model, we report on its evaluation and the corresponding results. Through a large set of experiments and a numerical analysis, we aim at revealing the pros and cons of the proposed scheme.

Nowadays, the increased amount of users' devices produce huge volumes of data that should be efficiently managed by modern applications. Streams are adopted to deliver data that, usually, are stored into a number of partitions. Splitting the data offers a lot of advantages as applications can process them in parallel, thus, they increase the speed of processing. Progressive analytics are also adopted to deliver partial responses, during processing, thus, saving time in the execution of applications. Data exploration and analytics queries are very significant for future applications. Usually, such queries demand for an ordered set of objects as a response and require intelligent predictive schemes to deliver the responses on top of the partial results retrieved by the distributed data partitions. A finite set of query processors are adopted to produce these partial results. Processors are placed in front of each partition and report progressive analytics to a central entity. In this paper, we envision the query controller (QC) as the central entity that collects progressive analytics and return the final response to users/applications. The QC receives partial ordered sets of objects and aggregates them to derive the final outcome. We focus on a QC that applies time-optimized techniques and aggregation operators to deliver every response, i.e., ordered sets, over streams of partial ordered subsets. We perform a comprehensive performance assessment with synthetic data and report on the performance of the QC. Our experimental evaluation reveals the pros and cons of the proposed model and a comparison assessment places this paper in the respective literature.

The Internet of Things involves a huge number of devices that collect data and deliver them to the Cloud. The processing of data at the Cloud is characterized by increased latency in providing responses to analytics queries defined by analysts or applications. Hence, Edge Computing (EC) comes into the scene to provide data processing close to the source. The collected data can be stored in edge devices and queries can be executed there to reduce latency. In this paper, we envision a case where entities located in the Cloud undertake the responsibility of receiving analytics queries and decide on the most appropriate edge nodes for queries execution. The decision is based on statistical signatures of the datasets of nodes and the statistical matching between statistics and analytics queries. Edge nodes regularly update their statistical signatures to support such decision process. Our performance evaluation shows the advantages and the shortcomings of our proposed schema in edge computing environments.

This work studies a sequential decision making methodology of when to update machine learning models in Edge Computing environments given underlying changes in the contextual data distribution. The proposed model focuses on updates scheduling and takes into consideration the optimal decision time for minimizing the network overhead. At the same time it preserves the prediction accuracy of models based on the principles of the Optimal Stopping Theory (OST). The paper reports on a comparative analysis between the proposed approach and other policies proposed in the respective literature while providing an evaluation of the performances using linear and support vector regression models. Our evaluation process is realized over real contextual data streams to reveal the strengths and weaknesses of the proposed strategy.

Humongous contextual data are produced by sensing and computing devices (nodes) in distributed computing environments supporting inferential/predictive analytics. Nodes locally process and execute analytics tasks over contextual data. Demanding inferential analytics are crucial for supporting local real-time applications, however, they deplete nodes' resources. We contribute with a distributed methodology that pushes the task allocation decision at the network edge by intelligently scheduling and distributing analytics tasks among nodes. Each node autonomously decides whether the tasks are conditionally executed locally, or in networked neighboring nodes, or delegated to the Cloud based on the current nodes' context and statistical data relevance. We comprehensively evaluate our methodology demonstrating its applicability in edge computing environments.

Current networking applications involve the definition and utilization of multiple virtualized resources on top of the available infrastructure. Software defined networks increase the performance compared with legacy systems as any functionality is managed through software. Securing the quality of service in such environments is significant to support novel applications that deliver their results in real time. In this paper, we propose a monitoring mechanism that observes the performance of the virtualized resources and identifies possible quality of service violations. Our model can be applied to any application domain, however, it is adapted to virtualized resources. We rely on a simple model that collects performance data, focuses on multiple parts of a virtualized functions chain and immediately concludes potential violations in real time. The proposed mechanism is incorporated in an SDN controller that is responsible to manage the virtualized resources. We provide an analytical description of the model and through a large set of simulations, we reveal its performance. Our results exhibit the timely identification of quality of service violations even in very dynamic environments where the performance of the network changes continuously.

We focus on Internet of Things (IoT) environments where a network of sensing and computing devices are responsible to locally process contextual data, reason and collaboratively infer the appearance of a specific phenomenon (event). Pushing processing and knowledge inference to the edge of the IoT network allows the complexity of the event reasoning process to be distributed into many manageable pieces and to be physically located at the source of the contextual information. This enables a huge amount of rich data streams to be processed in real time that would be prohibitively complex and costly to deliver on a traditional centralized Cloud system. We propose a lightweight, energy-efficient, distributed, adaptive, multiple-context perspective event reasoning model under uncertainty on each IoT device (sensor/actuator). Each device senses and processes context data and infers events based on different local context perspectives: (i) expert knowledge on event representation, (ii) outliers inference, and (iii) deviation from locally predicted context. Such novel approximate reasoning paradigm is achieved through a contextualized, collaborative belief-driven clustering process, where clusters of devices are formed according to their belief on the presence of events. Our distributed and federated intelligence model efficiently identifies any localized abnormality on the contextual data in light of event reasoning through aggregating local degrees of belief, updates, and adjusts its knowledge to contextual data outliers and novelty detection. We provide comprehensive experimental and comparison assessment of our model over real contextual data with other localized and centralized event detection models and show the benefits stemmed from its adoption by achieving up to three orders of magnitude less energy consumption and high quality of inference.

Software Defined Networking (SDN) is the key technology for combining networking and Cloud solutions to provide novel applications. SDN offers a number of advantages as the existing resources can be virtualized and orchestrated to provide new services to the end users. Such a technology should be accompanied by powerful mechanisms that ensure the end-to-end quality of service at high levels, thus, enabling support for complex applications that satisfy end users needs. In this paper, we propose an intelligent mechanism that agglomerates the benefits of SDNs with real-time “Big Data” forecasting analytics. The proposed mechanism, as part of the SDN controller, supports predictive intelligence by monitoring a set of network performance parameters, forecasting their future values, and deriving indications on potential service quality violations. By treating the performance measurements as time-series, our mechanism employs a novel ensemble forecasting methodology to estimate their future values. Such predictions are fed to a Type-2 Fuzzy Logic system to deliver, in real-time, decisions related to service quality violations. Such decisions proactively assist the SDN controller for providing the best possible orchestration of the virtualized resources. We evaluate the proposed mechanism w.r.t. precision and recall metrics over synthetic data.

We focus on Internet of Things (IoT) environments where sensing and computing devices (nodes) are responsible to observe, reason, report, and react to a specific phenomenon. Each node (e.g., an unmanned vehicle or an autonomous device) captures context from data streams and reasons on the presence of an event. We propose a distributed predictive analytics scheme for localized context reasoning under uncertainty. Such reasoning is achieved through a contextualized, knowledge-driven clustering process, where the clusters of nodes are formed according to their belief on the presence of the phenomenon. Each cluster enhances its localized opinion about the presence of an event through consensus realized under the principles of fuzzy logic (FL). The proposed FL-driven consensus process is further enhanced with semantics adopting type-2 fuzzy sets to handle the uncertainty related to the identification of an event. We provide a comprehensive experimental evaluation and comparison assessment with other schemes over real data and report on the benefits stemmed from its adoption in IoT environments.

Data stream monitoring provides the basis for building intelligent context-aware applications over contextual data streams. A number of wireless sensors could be spread in a specific area and monitor contextual parameters for identifying various phenomena, e.g., fire or flood. A back-end system receives measurements and derives decisions for possible abnormalities related to negative effects. We propose a mechanism which, based on multivariate sensors data streams, provides real-time identification of phenomena. The proposed framework performs contextual information fusion over consensus theory for the efficient measurements aggregation while time-series prediction is adopted to result future insights on the aggregated values. The unanimous fused and predicted pieces of context are fed into a type-2 fuzzy inference system to derive highly accurate identification of events. The type-2 inference process offers reasoning capabilities under the uncertainty of the phenomena identification. We provide comprehensive experimental evaluation over real contextual data and report on the advantages and disadvantages of the proposed mechanism. Our mechanism is further compared with type-1 fuzzy inference and other mechanisms to demonstrate its false alarms minimization capability.

The new era of Web and Internet of Things (IoT) paradigm is being enabled by the proliferation of various devices like RFIDs, sensors, and actuators. Smart devices (devices having significant computational capabilities, transforming them to `smart things') are embedded in the environment to monitor and collect ambient information. In a city, this leads to Smart City frameworks. Intelligent services could be offered on top of such information related to any aspect of humans' activities. A typical example of services offered in the framework of Smart Cities is IoT-enabled waste management. Waste management involves not only the collection of the waste in the field but also the transport and disposal to the appropriate locations. In this paper, we present a comprehensive and thorough survey of ICT-enabled waste management models. Specifically, we focus on the adoption of smart devices as a key enabling technology in contemporary waste management. We report on the strengths and weaknesses of various models to reveal their characteristics. This survey sets up the basis for delivering new models in the domain as it reveals the needs for defining novel frameworks for waste management.

We present a robust, dynamic scheme for the automatic self-deployment and relocation of mobile sensor nodes (e.g., unmanned ground vehicles, robots) around areas where phenomena take place. Our scheme aims (i) to sense environmental contextual parameters and accurately capture the spatiotemporal evolution of a certain phenomenon (e.g., fire, air contamination) and (ii) to fully automate the deployment process by letting nodes relocate, self-organize (and self-reorganize), and optimally cover the focus area. Our intention is to “opportunistically” modify the previous placement of nodes to attain high-quality phenomenon monitoring. The required intelligence is fully distributed within the mobile sensor network so the deployment algorithm is executed incrementally by different nodes. The presented algorithm adopts the Particle Swarm Optimization technique, which yields very promising results as reported in the article (performance assessment). Our findings show that the proposed algorithm captures a certain phenomenon with very high accuracy while maintaining the networkwide energy expenditure at low levels. Random occurrences of similar phenomena put stress upon the algorithm which manages to react promptly and efficiently manage the available sensing resources in the broader setting.

Electronic markets are virtual frameworks where entities not known in advance have the opportunity to interact for the trading of products or services. Usually, a negotiation is necessary for the conclusion of the transaction. The conclusion is either positive (agreement) or negative (conflict). An efficient reasoning mechanism is necessary for players participating in negotiations. In this paper, we focus on the buyer side and propose two decision models based on the optimal stopping theory (OST). OST is proved to be very efficient in cases where an entity tries to find the time to stop a process with the aim of maximizing her utility. The outcome of the proposed decision method indicates whether the buyer stops a negotiation either by accepting the offer or continuing in the negotiation by rejecting it. In our models, we assume zero knowledge on the players' characteristics. Our proposed decision models do not require any complex modeling or any information provided by experts. Experimental results reveal the efficiency of each model and provide a comparison assessment with other research efforts.

We study the performance improvement of Location Based Services through the identification and subsequent use of groups of mobile nodes. In our scheme we exploit the formation of nodes into groups in order to reduce the computation load incurred in back-end systems (e.g., Location Servers) and the associated network overhead. The back-end systems track the position and communicate with the Group Leader (GL). The GL, in turn, passes the received information to the members of the group (e.g., through short-range communications). The formation of mobile groups is validated over time to avoid misinterpreted temporary groupings which could endanger the adoption of the reduced load/overhead scheme. A time scheduling scheme based on the Optimal Stopping Theory assists in the finalization of the group validity. Metrics like group compactness are thoroughly assessed in line with the optimal stopping time scheme to increase confidence on group validity and persistence. Performance assessment reveals significant benefits for the considered location based services system.

Future Web business models involve virtual environments where entities interact in order to sell or buy information goods. Such environments are known as Information Markets (IMs). Intelligent agents are used in IMs for representing buyers or information providers (sellers). We focus on the decisions taken by the buyer in the purchase negotiation process with sellers. We propose a reasoning mechanism on the offers (prices of information goods) issued by sellers based on fuzzy logic. The buyer’s knowledge on the negotiation process is modeled through fuzzy sets. We propose a fuzzy inference engine dealing with the decisions that the buyer takes on each stage of the negotiation process. The outcome of the proposed reasoning method indicates whether the buyer should accept or reject the sellers’ offers. Our findings are very promising for the efficiency of automated transactions undertaken by intelligent agents.

Software Agents can assume the responsibility of finding and negotiating products on behalf of their owners in an electronic marketplace. In such cases, Fuzzy Logic can provide an efficient reasoning mechanism especially for the buyer side. Agents representing buyers can rely on a fuzzy rule base in order to reason for their next action at every round of the interaction process with sellers. In this paper, we describe a model where the buyer builds its fuzzy knowledge base using algorithms for automatic fuzzy rules generation based on data provided by experts and compare a set of such algorithms. Owing to such algorithms, agent developers spend less time and effort for the definition of the underlying rule base. Moreover, the rule base is efficiently created through the use of the dataset indicating the behaviour of the buyer and, thus, representing its line of actions in the electronic marketplace. In our work, we use such algorithms for the definition of the buyer behaviour and we provide critical insides for every algorithm describing their advantages and disadvantages. Moreover, we present numerical results for every basic parameter of the interaction process, such as the time required for the rule base generation, the Joint Utility of the interaction process or the value of the acceptance degree that each algorithm results.

Electronic markets are places where entities not known in advance can negotiate and agree upon the exchange of products. Intelligent agents can be proved very advantageous when representing entities in markets. Mostly, such entities are based on reputation models in order to conclude a transaction. However, reputation is not the only parameter that they could be based on. In this work, we deal with the problem of how and on which entity a buyer should be rely upon in order to conclude a transaction. Reinforcement learning techniques are used for these purposes. More specifically, the Q-learning algorithm is used for the calculation of the reward that the buyer will take for every action in the market environment. Actions represent the selection of specific entities for the negotiation of products. The most important is that the reward values are calculated based on a number of parameters such as the price, the delivery time, etc. The result is a more efficient model that is not based only on the reputation of each entity. Finally, we extend the Q-learning algorithm and propose a methodology for the dynamic Q-table creation which results reduced time for its construction and respectively limited time for the purchase action. Simulations show that this model indicates a significant time reduction in the purchase process in conjunction with the best solution according to the characteristics of products.

Intelligent agents can help users in finding and retrieving goods from electronic marketplaces. Additionally, agents can represent providers in such places facilitating the automatic negotiation about the purchase of products. In this paper, we describe a finite horizon bargaining model between buyers and sellers and we focus on the seller's side. Seller agents are a good example of an autonomous decentralized system. We present a method for the dasiabargainingpsila deadline calculation based on fuzzy-logic (FL). Such deadline indicates the time for which it is profitable for a seller to participate in the bargaining procedure. We provide methods for automatic fuzzy rules generation. These rules result the deadline values at each interaction and are based on data provided by experts. We compare results taken from a fuzzy controller based on such automatic methods with results taken by previous research efforts.

Present and future Web business models involve the trading of information goods. Information marketplaces can be considered as places where users search and retrieve information goods. Such places appear to be very interesting information retrieval models. Furthermore, software agent technology could help users and providers to work in such open environments providing a variety of advantages. Users as well as information providers could be represented by intelligent agents that work autonomously. The representatives of users assume the role of information buyers while the representatives of information sources could be referred to as sellers. In this paper, we examine a scenario where agents representing entities involved in an information marketplace bargain over the prices of information goods. Bargaining originates in game theory (GT). The rationale is that some entities contest to gain as much profit as possible in an open environment. We study the sellerspsila side. Sellers involved in a number of games with buyers, are trying to achieve as greater prices as possible in order to gain more profit from each game. We present a theoretical model of deadline computation for which sellers are participating in the game. Over this time limit it is useless for sellers to continue the game while buyers reject the proposed prices.

Information marketplaces are places where users search and retrieve information goods. Intelligent agents could represent the participating entities in such places, i.e, assume the role of buyers and sellers of information products. In this paper, we introduce a finite horizon bargaining model between buyers and sellers. We examine the seller's side and define a method for the 'bargaining' deadline calculation based on fuzzy-logic (FL). Such deadline indicates the time for which it is profitable for a seller to participate in the bargaining procedure, i.e., the time threshold for his offers. We represent the seller's knowledge/policy adopting the fuzzy set theory and provide a fuzzy inference engine for reasoning about the bargaining deadline. The result of the reasoning process defines the degree of patience of the seller agent, thus, affecting the time for which that seller participates in the bargaining game.