An IoT Analytics Embodied Agent Model based on Context-Aware Machine Learning

Abstract

Agent-based Internet of Things (IoT) applications have recently emerged as applications that can involve sensors, wireless devices, machines and software that can exchange data and be accessed remotely. Such applications have been proposed in several domains including health care, smart cities and agriculture. However, despite their increased adoption, deploying these applications in specific settings has been very challenging because of the complex static and dynamic variability of the physical devices such as sensors and actuators, the software application behavior and the environment in which the application is embedded. In this paper, we propose a modeling approach for IoT analytics based on learning embodied agents (i.e. situated agents). The approach involves: (i) a variability model of IoT embodied agents; (ii) feedback evaluative machine learning; and (iii) reconfiguration of a group of agents in accordance with environmental context. The proposed approach advances the state of the art in that it facilitates the development of Agent-based IoT applications by explicitly capturing their complex and dynamic variabilities and supporting their self-configuration based on an context-aware and machine learning-based approach.

A Reference Model for IoT Embodied Agents Controlled by Neural Networks

“…Agents that can interact with other agents or the environment in which the applications are embedded are called embodied agents [3]. Examples of such agents can be found in areas such as autonomous robots and cyber-physical systems.…”

“…In particular, in our previous research work (e.g. [2], [3]), we have proposed the use of embodied agents to denote intelligent agents which are embedded in the design of IoT applications. Each agent is provided with sensors and actuators to interact with the environment, and with a 'controller' that usually contains an artificial neural network (ANN).…”

“…The identification of the main characteristics of IoT embodied agents is based on three software approaches that we previously developed for IoT embodied agents: (i) a software framework for the development of embodied agents for IoT applications [2]; (ii) an approach to configure embodied agents based on the environment [3]; and (iii) a method for testing embodied agents controlled by neural nets [8].…”

Progressive Monitoring of IoT Networks Using SDN and Cost-Effective Traffic Signatures

“…Authors of [85,88,102] used an ensemble model that includes several models trained on past data, and once the drift occurs, they train a new model purely using the fresh data and add it to the existing ensemble. Contextualized modeling is another method for handling concept drifts [90,91,98,99,103] that identifies a number of recurring contexts and generates separate models per individual context. During the testing phase, the best model is selected (using a selection strategy) from the previously trained models based on the characteristics of the testing data.…”

Context-Augmented Software Development in Traditional and Big Data Projects: Literature Review and Preliminary Framework

“…A more specific definition applied to software engineering, proposed by Murphy [8] is that it "is the information about the system under development and the environment and process in which the system is being developed". A system that can respond to these possible mutable scenarios (e.g., domain, process, technologies involved, people) beats methods that are not prepared for these contextual changes [27].…”