Search

Due to the increasing interest in employing smartphones as first-class citizens in high-performance Edge computing environments, the necessity of software to facilitate the evaluation of load-balancing strategies for smartphone-based clusters has emerged. Regarding this, to select the best strategy for a cluster with m smartphones, usually a number of g candidate strategies are evaluated based on a number of r scenarios that contain these smartphones, which differ in terms of the start battery levels required for these smartphones. Thus, each of the r scenarios must be prepared before evaluating each of the g strategies on each $r_{i}$ , so that the smartphones have the required start battery levels pre-configured for $r_{i}$ , which requires discharging or charging smartphones. This leads to a number of $e= r^{\ast } g$ scenario preparation events that must be sequentially developed, considering that the time required to develop each event depends on the previous event. Thus, the single-objective problem addressed here implies finding out the sequential order in which the events should be developed, so that the total time required to develop them is minimized. This problem is modeled as the ATSP (Asymmetric Traveling Salesman Problem), since defining the sequential order to develop the events is equivalent to defining the sequential order to visit the cities, and therefore, is an NP-Hard problem. Given the complexity of this problem, the novel software module BAGESS (Battery Aware Green Edge Scenario Sequencer) is proposed, which uses a genetic algorithm for defining the sequential order to develop the events. BAGESS’s performance outperforms those of the methods currently used for the problem, reaching significant savings regarding the time required to develop the events in the range [12, 85]%.

Dew computing, an evolution of Fog computing, aims at fulfilling computing needs, such as deep learning applied to object classification, close to where data is originated and using computing resources that include consumer electronic devices such as smartphones. Simulation tools like DewSim aid the study of resource allocation mechanisms for exploiting clusters of smartphones, however, there is a gap w.r.t software tools that allow to perform similar studies over real Dew computing testbeds. We have developed LiveDewStream, an open source project to model executable tasks derived from data streams to be run on real smartphone clusters. The project offers a key functionality missing in other tools: reproducibility of battery-driven Dew experiments. Our major contribution is to provide the community a common in vivo platform to study best-performing allocation mechanisms under different stream processing scenarios and/or deep learning inference models.

Motrol is a simple device that satisfies functional requirements necessary for the automatic execution of battery-driven tests and profiling of connected mobile devices. It is specifically a hardware/software platform that allows Dew Computing researchers and developers to automate performance tests on Android-based smartphones. The hardware is based on a NodeMCU Esp8266 microcontroller that runs a firmware for managing the outputs. This software allows enabling/disabling the relays that connect the sockets that power the chargers of up to 4 mobile devices minimizing the need for human intervention. The firmware runs a web server that serves Rest requests from a Rest client with the commands to drive the digital outputs. These digital outputs activate or deactivate the relays to allow current to pass or not to the sockets. Such capability is essential to automate the study of battery behavior on battery-driven devices such as smartphones. Motrol is easy to assemble, knowledge in electronics or programming languages is not necessary, it is constructed with open hardware, and it is cheap, being its total cost ∼USD 30.

Dew computing aims to minimize the dependency on remote clouds by exploiting nearby nodes for solving non-trivial computational tasks, e.g., AI inferences. Nowadays, smartphones are good candidates for computing nodes; hence, smartphone clusters have been proposed to accomplish this task and load balancing is frequently a subject of research. Using the same real—i.e., in vivo—testbeds to evaluate different load balancing strategies based on energy utilization is challenging and time consuming. In principle, test repetition requires a platform to control battery charging periods between repetitions. Our Motrol hard-soft device has such a capability; however, it lacks a mechanism to assure and reduce the time in which all smartphone batteries reach the level required by the next test. We propose an evolutionary algorithm to execute smartphone battery (dis)charging plans to minimize test preparation time. Charging plans proposed by the algorithm include charging at different speeds, which is achieved by charging at maximum speed while exercising energy hungry components (the CPU and screen). To evaluate the algorithm, we use various charging/discharging battery traces of real smartphones and we compare the time-taken for our method to collectively prepare a set of smartphones versus that of individually (dis)charging all smartphones at maximum speed.

Recent Web technologies such as HTML5, JavaScript, and WebGL have enabled powerful and highly dynamic Web mapping applications executing on standard Web browsers. Despite the complexity for developing such applications has been greatly reduced by Web mapping libraries, developers face many choices to achieve optimal performance and network usage. This scenario is even more complex when considering different representations of geographical data (raster, raw data or vector) and variety of devices (tablets, smartphones, and personal computers). This paper compares the performance and network usage of three popular JavaScript Web mapping libraries for implementing a Web map using different representations for geodata, and executing on different devices. In the experiments, Mapbox GL JS achieved the best overall performance on mid and high end devices for displaying raster or vector maps, while OpenLayers was the best for raster maps on all devices. Vector-based maps are a safe bet for new Web maps, since performance is on par with raster maps on mid-end smartphones, with significant less network bandwidth requirements.

Mobile devices have evolved from single purpose devices, such as mobile phone, into general purpose multi-core computers with considerable unused capabilities. Therefore, several researchers have considered harnessing the power of these battery-powered devices for distributed computing. Despite their ever-growing capabilities, using battery as power source for mobile devices represents a major challenge for applying traditional distributed computing techniques. Particularly, researchers aimed at using mobile devices as resources for executing computationally intensive task. Different job scheduling algorithms were proposed with this aim, but many of them require information that is unavailable or difficult to obtain in real-life environments, such as how much energy would require a job to be finished. In this context, Simple Energy Aware Scheduler (SEAS) is a scheduling technique for computational intensive Mobile Grids that only require easily accessible information. It was proposed in 2010 and it has been the base for a range of research work. Despite being described as easily implementable in real-life scenarios, SEAS and other SEAS-improvements works have always been evaluated using simulations. In this work, we present a distributed computing platform for mobile devices that support SEAS and empirical evaluation of the SEAS scheduler. This evaluation followed the methodology of the original SEAS evaluation, in which Random and Round Robin schedulers were used as baselines. Although the original evaluation was performed by simulation using notebooks profile instead of smartphones and tablets, results confirms that SEAS outperforms the baseline schedulers.

Web Services have been steadily gaining maturity as their adoption in the software industry grew. Accordingly, metric suites for assessing different quality attributes of Web Service artifacts have been proposed recently - e.g., for services interfaces in WSDL (Web Service Description Language). Like any other software artifact, WSDL documents have several inherent attributes (e.g., size or complexity) that can be measured. We present an approach to prevent a high complexity on services interfaces (WSDLs), to ease consumers to reason about services' offered functionality. Mostly, WSDLs are automatically derived from object-oriented (OO) source code, with a likely impact on complexity. Thereby, we study the statistical relationships between a recent metric suite of service interface complexity (proposed by Basci & Misra) and the well-known Chidamber & Kemerer's OO metric suite (applied to service implementations), on a data-set of 154 real-world services. First, a theoretical validation of Baski & Misra's suite (using Weyuker's properties) is presented, to prove the ability to measure complexity in WSDL documents. Then, after finding high correlation between both metric suites, we have conducted a series of experiments to analyze how certain refactorings on services' source codes prior to derive WSDLs might reduce complexity. In this way, our approach exploits OO metrics as development-time indicators, to guide software developers towards obtaining less complex service interfaces.

Building Service-oriented Applications implies the selection of adequate services to fulfill required functionality. Even a reduced set of candidate services involves an overwhelming assessment effort. In a previous work we have presented an approach to assist developers in the selection of Web Services. In this paper we detail its behavioral assessment procedure, which is based on testing and adaptation. This is done by using black-box testing criteria to explore services behavior. In addition, helpful information takes shape to build the needed adaptation logic to safely integrate the selected candidate into a Service-oriented Application. A concise case study shows the potential of this approach for both selection and integration of a candidate Web Service.

A Web Service has an interface described in a machine-processable format (specifically WSDL). Service providers expose their services by publishing their WSDL documents. Service consumers can learn about service capabilities and how to interact with services. Service descriptions (WSDL documents) should be ideally understood easily by service stakeholders so that the process of consuming services is simplified. In this work we present a practical metric to quantify readability in WSDL documents. We adapted and extended an existing ontology-based semantic readability metric to focus on WSDL documents by using WordNet as a lightweight concept hierarchy. We have validated our approach by performing both qualitative and quantitative experiments. The first one consists of a controlled survey with a group of service consumers. The results showed that consumers (software engineers) required less time and effort to analyze WSDL documents with higher readability values. The second experiment compares our approach with two ontology-based approaches. The third experiment compares the readability values of a dataset of real-life service descriptions before and after rewriting them. The results showed the effectiveness of our approach to assess readability of Web Services interfaces.

BCS (Body Condition Score) is a method to estimate body fat reserves and accumulated energy balance of cows, placing estimations (or BCS values) in a scale of 1 to 5. Periodically rating BCS of dairy cows is very important since BCS values are associated with milk production, reproduction, and health of cows. However, in practice, obtaining BCS values is a time-consuming and subjective task performed visually by expert scorers. There have been several efforts to automate BCS of dairy cows by using image analysis and machine learning techniques. In a previous work, an automatic system to estimate BCS values was proposed, which is based on Convolutional Neural Networks (CNNs). In this paper we significantly extend the techniques exploited by that system via using transfer learning and ensemble modeling techniques to further improve BCS estimation accuracy. The improved system has achieved good estimations results in comparison with the base system. Overall accuracy of BCS estimations within 0.25 units of difference from true values has increased 4% (up to 82%), while overall accuracy within 0.50 units has increased 3% (up to 97%).

With self-provisioning of resources as premise, dew computing aims at providing computing services by minimizing the dependency over existing internetwork back-haul. Mobile devices have a huge potential to contribute to this emerging paradigm, not only due to their proximity to the end user, ever growing computing/storage features and pervasiveness, but also due to their capability to render services for several hours, even days, without being plugged to the electricity grid. Nonetheless, misusing the energy of their batteries can discourage owners to offer devices as resource providers in dew computing environments. Arguably, having accurate estimations of remaining battery would help to take better advantage of a device’s computing capabilities. In this paper, we propose a model to estimate mobile devices battery availability by inspecting traces of real mobile device owner’s activity and relevant device state variables. The model includes a feature extraction approach to obtain representative features/variables, and a prediction approach, based on regression models and machine learning classifiers. On average, the accuracy of our approach, measured with the mean squared error metric, overpasses the one obtained by a related work. Prediction experiments at five hours ahead are performed over activity logs of 23 mobile users across several months.

Historically, software engineers have conceived metric suites as valuable tools to estimate the quality of their software artifacts. Recently, a fresh computing paradigm called Service-Oriented Computing (SOC) has emerged at the crossing of massively distributed and heterogeneous software. This paper presents a statistical correlation analysis showing that classic software engineering metrics can be used to predict the most relevant quality attributes of WSDL documents, the essential software artifact when materializing this novel computing paradigm withWeb-based technologies. For the experiments, two recent WSDL-level metrics catalogs and 154 real world WSDL documents have been employed.

The Service-Oriented Computing (SOC) paradigm has gained a lot of attention in the software in- dustry since SOC represents a novel way of architecting distributed applications. SOC is mostly materialized via Web Services, which allow developers to expose their application as building blocks for other applications by defining a clear and public interface. Although conceptually and technologically mature, SOC still lacks adequate development support from a methodological point of view. We present the EasySOC project: a catalog of guidelines to build service-oriented applications and services. This catalog synthesizes best SOC development practices that arise as a result of several years of research in fundamental SOC-related topics, namely WSDL-based technical specification, Web Service discovery and Web Service outsourcing. In addition, we describe a plug-in for the Eclipse IDE that has been implemented to simplify the utilization of the guidelines. We believe that the practical nature of the guidelines, the empirical evidence that supports them, and the availability of IDE-integrated tools that enforces them will help software practitioners to rapidly exploit our ideas for building real SOC applications.