Low latency in the detection of objects in images is a fundamental aspect to provide an immediate response in different application scenarios, and to carry out the relevant actions. Deep learning-based models offer better results than traditional object detection techniques. As a result of this increased accuracy, the computational cost of these models is often very high. For this reason, several edge computing devices have emerged that perform inference quickly and efficiently due to the incorporation of hardware accelerators. In this work, different devices are evaluated using object detection algorithms based on deep learning. For this purpose, YOLOv3, YOLOv5 and YOLOX, with all their variants, have been used on an NVIDIA Jetson Nano, an NVIDIA Jetson AGX Xavier and a Google Coral Dev Board. For the evaluation of the models and devices, one of the most widespread datasets, MS COCO, is used. In addition, they have been evaluated using twenty different input sizes and three frameworks (Pytorch, TensorRT and Tensorflow Lite). From the data obtained, data can be extrapolated to other models such as YOLOv8. Additionally, the FPS/Power Consumption and FPS/Cost ratios are analyzed, as well as their feasibility in a real use scenario. As a result of this work, valuable recommendations are provided for projects where this technology is to be applied.

Abstract Manufacturing industry often uses 3D scanning technologies to inspect their products. Some of these techniques produce a point cloud that represent a section of the manufactured product. The clouds must be aligned to the model of the product in order to check its quality. Current registration methods are usually affected by dimensional problems or volumetric anomalies. This paper proposes a new method for the registration process aligning the cloud to the model in several steps. The first step is the state-of-the-art method. The second step uses the information acquired in the first one to perform a fine registration in order to not being affected by dimensional defects or little miss alignments in the previous step providing a huge improvement in the measurement of surface defects. In this paper, several techniques are proposed in order to provide a set of tools that the final users can tune to fit their needs. The whole procedure of registration can be run in real-time conditions using the sampling and caching strategies proposed. The methods proposed are tested over more than 10,000 profiles of a rail proving they successfully align the cloud and the model providing better results in the measurement of surface defects.

Abstract Condition monitoring of industrial equipment has become a critical aspect in Industry 4.0. This paper shows the design, implementation and testing of a low-cost Industrial Internet of Things (IIoT) system designed to monitor electric motors in real-time. This system can be used to detect operating anomalies and paves the way for building predictive maintenance models. The system is built using low-cost hardware components (wireless multi-sensor modules and single-board computers as gateways), open-source software and open cloud services, where all the relevant information is stored. The module collects real-time vibration data from electric motors. Vibration analyses in the temporal and frequency domains were carried out in both modules and gateways to compare their capabilities. This approach is also a springboard to using edge/fog computing to save cloud resources. A system prototype has been tested in the laboratory and in an industrial dairy plant. The results show that the proposed system can be used for continuous monitoring of any rotatory machine with similar accuracy to professional monitoring devices but at a significantly lower cost.

The edge computing model is an approach to Internet of Things (IoT) architectures based on the redistribution of services and infrastructure from centralized clouds to locations closer to IoT devices. The Web of Things (WoT) is another important IoT trend, currently led by the W3C, which aims at solving the IoT interoperability problem by adopting proven technologies and patterns from the Web. The design and validation of IoT deployments based on these paradigms is a complex task that involves multiple services, heterogeneous hardware and diverse communication technologies. Testing such projects in real world conditions usually requires a significant investment of resources. There are simulation tools that can assist in this process with much lower barriers of entry, however, they involve the designer making modelling assumptions that are not always representative of the real systems. This work presents an emulation tool for IoT projects based on the edge computing model that is able to seamlessly scale horizontally by leveraging container orchestration (Docker swarm mode). Furthermore, the W3C WoT model is included as a first-class citizen, enabling the designer to model all actors in the system as Things. The tool can run the real production code with minimal modifications and provides meaningful insights into the behaviour of the proposed architecture. This knowledge serves to rapidly iterate the optimization process, simplifying design issues and the detection of bottlenecks before committing to a real deployment in the field. A real-world scenario is also emulated in order to demonstrate its capabilities and validate its contribution.

For a long time, the common billing time slot used by cloud providers was the hour, but recently they have changed it to one second with a minimum charge of one minute. In this paper, the impact of this change on virtual machine allocation strategies is analyzed. With the minimum one-minute charge, the state-of-the-art allocation strategies are no longer optimal. This paper proposes a new analytic model to obtain the optimal allocation that minimizes the total cost considering this new billing scheme. However, the optimization problem using this model is unmanageable due to its computational complexity. Therefore, other sub-optimal allocations strategies are proposed. The performance of these strategies (in terms of total allocation cost and computational effort) is analyzed in order to assess the influence of the time slot length. Allocation time slots of one hour, one minute and one second are compared. The experimental work is carried out using synthetic workloads based on two real public traces. The study concludes that the time slot of one minute offers the best trade-off between allocation cost and computational effort. The experimentation shows that the proposed strategy can save up to 14 percent over using the time slot of one hour.

Automatic detection of atmospheric emissions is becoming essential in current factories. There is a lack of systems to detect emissions, although there are several systems to detect smoke that could be adapted for detecting emissions. This is the approach followed in this work: using convolutional neural networks (CNN) to classify images as containing emissions or not. Then, emissions are detected by analyzing the sequences of images classified as containing emissions. This detection system, based on a CNN, can be used in any factory simply by training the CNN with images of the factory with and without emissions.

Summary The advent of hybrid cloud technologies and public Infrastructures as a Service (IaaS) allows service developers to offer services to their customers with little upfront investment and to adapt services to different workload sizes. The problem of minimizing the costs of the hired public infrastructure while providing the quality of service needed by the final customer arises when using hybrid clouds. Several scheduling strategies have been proposed to solve this problem for services dealing with deadline‐constrained bag‐of‐tasks workloads. Most of these solutions do not consider the variable performance of the clouds, the provisioning delay of virtual machine instances that affects the elasticity, and the impracticality of having good processing time estimations in real systems. We propose a scheduler algorithm that overcomes previous limitations and can minimize the cost of the infrastructure while maximizing the number of deadlines met by the service. Our solution can work autonomously by using sampled observations of the processing times and considers the heterogeneity and the provisioning time of the virtual machine instances. An evaluation was conducted by simulating different scenarios and workload types. Simulation results show that our solution obtains better or similar results than previous techniques in most scenarios in terms of deadlines met.

The use of interactive multimedia communications is widespread in business and education, providing an enriched user experience in teleconferencing, distance learning and online meetings. Overlays networks are the underlying mechanisms enabling multipoint communication in interactive multimedia platforms. They must cope with constraints on heterogeneity, scalability and availability of communications, making their deployment and management a true challenge. Thus, several issues might affect the resilience of overlay networks, threatening ongoing communications. Overlay networks usually use self-organization and self-stabilization techniques to achieve resilient communications. Techniques implementing other self-management properties such as self-regulation also help in improving the resilience of the overlay. In this paper, the resilience of a reflector-based overlay network is improved with a self-regulation technique based on media transcoding. This balances the workloads of the reflectors of the overlay avoiding excessive usage of network links. As a result, the overlay is more resilient to failures. Extensive simulations have been carried out, and various media transcoding schemes have been compared, proving that self-regulation techniques improve the resilience of the reflector-based overlay network.

Cluster management has become a multi objective task that involves many disciplines like power optimization, fault tolerance, dependability and online system operation analysis. Efficient and secure operation of these clusters is a key objective of any data center policy. In addition, the service provided by these servers must fulfill a level of quality of service (QoS) to the customers. Applying self-management techniques to these clusters would simplify and automate its operation. Current self-management techniques that take into account service level agreements (SLAs) do not cover at the same time the two most important sides of the cluster operation: self-optimization, for an efficient and profitable operation, and self-healing, for a secure operation and high level of quality of service perceived by users. This work integrates a self-optimization strategy for Internet server clusters that optimizes the power consumption, using dynamic provisioning of servers, with a self-healing strategy that improves the reaction of the cluster to a server failure, by using the spare capacity of the cluster intelligently. The self-management technique is based on empirical response time and power consumption models of the servers that simplify its operation. Additionally, the technique presented in this paper guarantees the fulfillment of the SLA.

The sophistication of malicious software (malware) used to break the computer security has increased exponentially in the last years. Frequently, malware is hidden into a computer by software components called root kits. Therefore, early detection of root kits is of primary importance to avoid the uncontrolled operation of malware. Most of current techniques for root kit detection only allow a late detection after the malware has already been hidden by a root kit. In this paper, a new technique is presented that enables the proactive detection of root kits while they are hiding malware, and therefore, allowing that hiding can be avoided. The technique has been designed for root kits that operate in kernel-mode. This root kits are particularly difficult to detect because both the detector and the root kit are executed with the same privileges. This technique can be used to improve the detection capabilities of intrusion detection and prevention systems.

Synchronous e-training is emerging as an alternative for developing human resources training plans in large organizations. Real-time communications are used to emulate face-to-face interaction that occurs in on-campus learning environments. However, the security concerns that a synchronous e-training platform must face may compromise the integrity, availability and confidentiality of corporate information, which may lead to serious economic and legal consequences. The disclosure of corporate information or the unauthorized participation in e-training activities must be prevented. In this paper, the security issues in synchronous e-training are identified, and the threats to a real e-training platform are analyzed. The platform is organized into four virtual networks with different security requirements and vulnerabilities. The platform assumes that multicast communications are available in the underlying corporate network. The threats affecting each element of the platform and their impact on e-training activities are discussed. Finally, a security scheme is proposed fixing the aforementioned vulnerabilities. Digital certificates and encryption algorithms solve most of the vulnerabilities, but other techniques such as access control lists and user skills on security basics are essential. Most of the proposed scheme is applicable to other real-time communication systems, since the e-training platform is built using standard technologies commonly used in voice over IP systems.

Synchronous e-learning tools always include an audio-conference feature so participants in e-learning sessions can communicate orally. Audio quality is critical to avoid misunderstandings and improve user experience. Therefore, it is important to evaluate the audio quality that speech codecs can provide. Few existing assessment works consider the resources consumed (CPU or bitrate) to provide the audio quality, although synchronous e-learning sessions usually involve various participants, making resource consumption an important issue. In this paper, both objective and subjective audio quality measurement methods are used to characterize and estimate the audio quality of twenty speech codecs as a function of the resources consumed during synchronous e-learning sessions. Although users' opinions on audio quality are often more pessimistic than the evaluation provided by objective measurements, the correlation between objective and subjective measurements is high for medium-quality codecs. Users perceive lower audio quality for low-quality codecs than indicated by objective measurements, while they are not able to identify high-quality codecs, scoring them similarly to medium-quality codecs.

The detection of periodical defects is of primary importance in the manufacturing of many long flat products. As an example, a thick steel block is rolled (passed through several pairs of rolls) to obtain a long steel strip. When a roll has a flaw, it provokes a periodical defect on the strip. If the defect is not detected promptly, a large number of manufactured strips will be marked with the periodical defect. The economic losses incurred when roll flaws are not detected are very high, because the strips can not be sold to the customers and all the resources consumed in their manufacturing are wasted. This paper presents an algorithm for detecting periodical defects by analyzing the single defects detected by an inspection system based on computer vision. Because these defects form a periodical pattern, pattern matching techniques can be used for their detection. The paper also contains an analysis of the metrics used to characterize the performance of the algorithm and the experimental methodology used to find the optimal values for its configuration parameters. The optimal configuration must maximize true detections, and also minimize false detections. False detections can shut the manufacturing line down unnecessarily to search for non-existent roll flaws. Finally, the results obtained are compared with those provided by the most widely used system in this field. In most cases, the results provided by the algorithm proposed were better.

Synchronous e-training is emerging as an alternative for developing human resources training plans in large organizations. However, synchronous e-training platforms must face many issues affecting their resilience. Self-recovering and self-protection software techniques should be implemented to enable the continuity of e-training activities faced with network failures or entities of the platform going down. In this paper, the resilience issues in synchronous e-training are identified, and those presented in a real e-training platform are analyzed. The platform includes a self-optimizing technique to make efficient use of network resources as participants join and leave. Furthermore, various self-recovering and a self-protection techniques are proposed, making possible the continuity of ongoing e-training activities with reduced functionality until the automatic full system recovery. Most of the proposed techniques are applicable to other real-time communication systems, since the e-training platform is built using standard technologies such as SIP and RTP, commonly used in voice over IP systems.

Nowadays, enterprises providing services through Internet often require online services supplied by other enterprises. This entails the cooperation of enterprise servers using Web services technology. The service exchange between enterprises must be carried out with a determined level of quality, which is usually established in a service level agreement (SLA). However, the fulfilment of SLAs is not an easy task and requires equipping the servers with special control mechanisms which control the quality of the services supplied. The first contribution of this research work is the analysis and definition of the main requirements that these control mechanisms must fulfil. The second contribution is the design of a control mechanism which fulfils these requirements and overcomes numerous deficiencies posed by previous mechanisms. The designed mechanism provides differentiation between distinct categories of service consumers as well as protection against server overloads. Furthermore, it scales in a cluster and does not require any modification to the system software of the host server, or to its application logic.

Stochastic analysis techniques for real-time systems model the execution time of tasks as random variables. These techniques constitute a very powerful tool to study the behaviour of real-time systems. However, as they can not avoid all the timing bugs in the implementation, they must be combined with measurement techniques in order to gain more confidence in the implemented system. In this paper, a set of tools to measure, analyze and visualize traces of real-time systems is presented. These tools are driven by stochastic models. In order to find bugs in the timing behaviour of the system, two metrics, called "pessimism" and "optimism", are proposed. They are based on two random variables, the optimistic and the pessimistic execution time, which are also introduced in this paper. These metrics are used in the debugging tools to compare the model and the measured system in order to find errors. The metrics are examined in three case studies.

This paper presents the study carried out on the data traffic collected on the network of a cable operator based on hybrid fiber-coax technology, and the subsequent simulation model developed to predict the bandwidth requirements of the network channels. The paper starts with the analysis of the traffic measurements, taken over two periods of time in one year, on all the network's channels. This analysis identifies the main characteristics of the traffic, as well as some relationships between network parameters and their persistence over time. The paper proceeds to present the development of a simulation model, which represents the cable network. This model is built from the results of the traffic analysis and network parameters. The major challenge of this model is to predict the traffic on each channel of the cable network, related to parameters of the network configuration, such as the number of assigned subscribers to the channel and the time of day. To reach this goal, the simulation model has been developed with a modular structure, which gives it flexibility to adapt to changes in the network. The process followed involves the establishment of a traffic model, a system model, and finally, the validation of the results.

The utilization bound for real-time rate monotonic (RM) scheduling on uniprocessors is extended to multiprocessors with partitioning-based scheduling. This allows fast schedulability tests to be performed on multiprocessors and quantifies the influence of key parameters, such as the number of processors and task sizes on the schedulability of the system. The multiprocessor utilization bound is a function of the allocation algorithm, so among all the allocation algorithms there exists at least one allocation algorithm providing the minimum multiprocessor utilization bound, and one allocation algorithm providing the maximum multiprocessor utilization bound. We prove that the multiprocessor utilization bound associated with the allocation heuristic worst fit (WF) coincides with that minimum if we use Liu and Layland's bound (LLB) as the uniprocessor schedulability condition. In addition, we present a class of allocation algorithms sharing the same multiprocessor utilization bound which coincides with the aforementioned maximum using LLB. The heuristics first fit decreasing (FFD) and best fit decreasing (BFD) belong to this class. Thus, not even an optimal allocation algorithm can guarantee a higher multiprocessor utilization bound than that of FFD and BFD using LLB. Finally, the pessimism of the multiprocessor utilization bounds is estimated through extensive simulations.

A key issue in any system for distribution of continuous media-on-demand is the capacity of the system with regard to quality of the service specifications, that is, the number of simultaneous streams the system can provide until degradation of quality of reproduction and interactivity perceived by the users. This work presents the evaluation of interactive video-on-demand systems, with special attention to limitations caused by the video server. The evaluation is based on a set of metrics designed to determine the number of streams a video server can support under specific requirements of quality of the service as perceived by the users. To validate the utility of these metrics, a prototype of a news-on-demand service has been built and the load for this kind of systems has been characterised. In addition, a load generator which emulates the concurrent access of several users to the system has been built. The evaluation establishes the relationship between the video server limitations and the quality of the service perceived by the users.

This paper describes a stochastic analysis method for general periodic real-time systems. The proposed method accurately computes the response time distribution of each task in the system, thus making it possible to determine the deadline miss probability of individual tasks, even for systems with maximum utilization factor greater than one. The method uniformly covers both fixed-priority scheduling (such as rate monotonic) as well as dynamic-priority scheduling (such as earliest deadline first) and can handle arbitrary relative deadlines and execution time distributions. The accuracy of the method is proven by comparing the results from the analysis with those obtained from simulations, as well as other methodologies in the literature.

This paper deals with the problem of finding utilization bounds for multiprocessor rate monotonic scheduling with partitioning. The minimum and maximum utilization bounds among all the reasonable allocation algorithms are calculated. We prove that the utilization bound associated with the reasonable allocation heuristic Worst Fit (WF) is equal to that minimum. In addition, we prove that the utilization bound associated with the heuristics First Fit Decreasing (FFD) and Best Fit Decreasing (BFD) is equal to the maximum, of value (n+1)(2/sup 1/2/-1), where n is the number of processors.

Presents the utilization bound for earliest deadline first (EDF) scheduling on homogeneous multiprocessor systems with partitioning strategies. Assuming that tasks are pre-emptively scheduled on each processor according to the EDF algorithm, and allocated according to the first-fit (FF) heuristic, we prove that the worst-case achievable utilization is 0.5(n+1), where n is the number of processors. This bound is valid for arbitrary utilization factors. Moreover, if all the tasks have utilization factors under a value /spl alpha/, the previous bound is raised, and the new utilization bound considering /spl alpha/ is calculated. In addition, we prove that no uniprocessor scheduling algorithm/allocation algorithm pair can provide a higher worst-case achievable utilization than that of EDF-FF. Finally, simulation provides the average-case achievable utilization for EDF-FF.