In this paper, we present a Semi-Supervised Deep Learning approach for anomaly detection of Wind Turbine generators based on vibration signals. The proposed solution is integrated into an IoT Platform as a real-time Workflow. The Workflow is responsible for the whole detection process when a new sample is inserted in the IoT Platform, performing data gathering, preprocessing, feature extraction, and classification. The chosen Semi-Supervised Deep Learning model is a DAE, which builds a normality model using healthy data. The classification consists of comparing the reconstruction error for the computed entry with a normality threshold. The normality threshold is selected through an F1-Score analysis of the reconstruction errors over labeled data. Finally, the Workflow can produce notifications to the users whenever unhealthy behavior is noticed. The ability of the proposed mechanism to detect abnormal behavior in wind turbines on an IoT Platform is evaluated using a case study of real-world healthy and unhealthy data from a Wind Turbine. The solution was able to correctly classify every unhealthy sample and presented a low false-positive rate. Moreover, Workflow results can be improved by conditioning alarm triggering with a windowed-based anomaly accumulation.

In this paper, we introduce a Framework to Design and Implement Real-time Multicore Schedulers using Machine Learning techniques applied to the very own data such systems produce as they operate. The framework builds on sensors and event counters present in modern hardware platforms and on variables kept by the operating system to capture run-time data that are subsequently subjected to ML tools to produce scheduling heuristics targeting specific optimization goals. It provides non-intrusive mechanisms to collect such data while the system runs real task sets with real workloads, thus preserving the quality of the captured data. It abstracts the Performance Monitoring Unit, thermal sensing, energy monitoring, and Dynamic Voltage and Frequency Scaling available on such platforms through a lean, architecture-independent API. After describing the framework in details, we demonstrate its applicability with the implementation of an energy-efficient, load balancing, real-time, multicore heuristic for a PEDF scheduler. The measured overhead imposed by the framework on the tasks it schedule is at most 0,0003583% and the maximum added jitter is less than 40μs, corroborating the ability of the framework to support the development of effective domain-specific schedulers using machine learning techniques.

In this paper, we present an approach to assess the schedulability and scalability of Cyber-Physical Systems (CPS) Networks through an algorithm that is capable of estimating the load of the network as its utility grows. Our approach evaluates both the network load and the laxity of messages, considering its current topology and real-time constraints while abstracting environmental specificities. The proposed algorithm also accounts for the network unreliability by applying a margin-of-safety parameter. This approach enables higher utilities as it evaluates the load of the network considering a margin-of-safety that encapsulates phenomena such as collisions and interference, instead of performing a worst-case analysis. Furthermore, we present an evaluation of the proposed algorithm over three representative scenarios showing that the algorithm was able to successfully assess the network capacity as it reaches a higher use.

In this paper, we present an Anomaly Detection implementation with the usage of Artificial Neural Network (ANN) for Multicore Embedded Systems. The detector is built over a sophisticated Real-Time Multicore scheduling framework that allowed capturing high-quality run-time data for the Machine Learning (ML) process and provided the necessary infrastructure for the ANN to be embedded. To conceive the detector we first defined the system’s sane behaviour through a set of performance counters, providing the necessary information to define an anomaly. After describing the ML process and the ANN embedding details, we evaluate the results of the detection adding a different task to the execution and showing the embedded detector was able to successfully classify over 95% of the execution, never misinterpreting an anomaly as a sane task, with no interference on application execution time, once the anomaly detector runs on core 0, which is reserved for system management and control operations. Also, the maximum delay to detect that the running task is an anomaly was equal to 1 sampling of the performance monitoring counters (configured with captures spaced by 10ms, or 100 captures per second). We conclude the experiments showing the effectiveness of our runtime ANN anomaly detector by actuating on the suspension of the tasks classified as an anomaly, maintaining a sane execution by mitigating anomalies.