In Section 3, we discuss related systems. In Section 2, we describe a scenario together with requirements highlighting the need for modern IoT applications. The rest of the article is structured as follows. We provide an end-to-end real-world implementation of these components as an integrated middleware and evaluate its performance.We design a flexible API for application developers that allows them to configure dynamically an abstracted sensor toward improved accuracy or fault tolerance as needed.We propose a novel way to combine such clusters of quasi-redundant sensors into configurable, abstracted (virtual) sensors using the FTI function known from clock synchronization.This produces more meaningful clusters without a priori knowledge about the number of clusters. We show how to use graph-based community detection (known from social network analysis) to create clusters of (quasi-)redundant sensor measurements and sensors. To this end, we provide the following core contributions: We describe how the accuracy of measurements can be improved and adapted to application needs while easing data acquisition. In this article, we focus on the general concepts behind coSense. coSense combines unsupervised learning with the fault-tolerant interval (FTI) function to achieve scalable, reliable, and adaptable information fusion to empower modern IoT-driven applications with varying quality-of-service (QoS) requirements. To this end, we are introducing coSense-the collaborative, fault-tolerant, and adaptive sensing middleware for the IoT. To be of use for machine learning and data science algorithms, we need to improve the data quality of IoT devices. Additionally, they often have poor sensing quality and are prone to failure. However, using data from the IoT can be difficult, as sensors must be cheap. Modern data science algorithms and machine learning can use the IoT to get huge amounts of different kinds of input data, mostly originating from sensors. Gartner estimates up to 25 billion IoT-enabled devices by 2021. The Internet-of-Things (IoT) is becoming a reality.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |