We are presenting a massively parallel heterogeneous cloud-based architecture oriented towards anomalous activity detection in smart homes. The architecture has very high resilience to both hardware and software faults, it is capable of collecting activity from various data sources and performing anomaly detection in real-time. We corroborate the approach with an efficient checkpointing mechanism for data processing which allows the implementation of hybrid (CPU/GPU) fault-resilience and anomaly detection through pattern mining techniques, at the same time offering high throughput.

The optimization of the cloud resources used to power a multi-agent Internet of Things architecture is an important issue which has an important impact on the overall operation cost of the architecture. The resources tenancy is a costly operation, thus their allocation and management should be optimized based on the usage patterns. The infrastructure for the multi-agent system should not be affected by the deployment or maintenance life cycle, operations require parts of the system, or even the entire system to be offline during the execution of scheduled procedures. This paper outlines of the importance of the infrastructure audit, which offers a good insight of how the resources are used, the geographical areas which are heavily used and where the allocation or release of used resources is mandatory. Also, the security audit, in a distributed multi-agent architecture that handles a large number of heterogeneous devices, represents a good mechanism for performance improvement.

The astonishing expansion of Internet of Things has opened a lot of opportunities for related domains to employ strategies that were successfully used for the “things” governance. Furthermore, because of the technology blending in the most common household devices and wearable items, it becomes very easy for the computers to sense the surrounding environment and to collect information about the inhabitants, therefore transforming the intelligent house in a Home Care System (HCS). For medical conditions like dementia and its associated diseases, it is very convenient to monitor the patients in their living space because the patient will benefit from their home comfort. In addition, the costs for in hospital monitoring will decrease. This chapter proposes an Internet of Things Governance Architecture that can be used to sustain and monitor a complex e-health system, with application especially for patients with dementia and its associated diseases.

This paper presents a Multi-agent system that facilitates the remote monitoring of the elderly patients which are susceptible to mental disorder diseases. In order to find early signs of health condition depreciation we have assessed four of the most common mental disorder diseases to find which kind of sensors can detect specific symptoms with the main purpose of creating an early warning system. The diagnosis component is based on an ontology that defines the relations between sensors, symptoms and diseases. Based on these relationships a specialized agent can inform the medical personnel about the detected symptoms.

The Semantic data is regarded as one of the best methods to describe relations between different data sources. The Semantic annotation of data improves the data aggregation methods because the relations are already defined. We have implemented a software prototype using Scala, Apache Jena and the Fuseki database server for handling a context modeled using a dynamic created ontology. For use cases we have used a hospital setting because it is the one of the most dynamic environments.

The rapid expansion of the Internet of Things (IoT) will generate a diverse range of data types that needs to be handled, processed and stored. This paper aims to create a multi-agent system that suits the needs introduced by the IoT expansion, thus being able to oversee the Big Data collection and processing and also to maintain the semantic links between the data sources and data consumers. In order to build a complex agent oriented architecture, we have assessed the existing agent oriented methodologies searching for the best solution that is not bound to a specific programming language of framework, and it is flexible enough to be applied in such a divers domain like IoT. As complex scenario, the proposed approach has been applied to medical diagnosis and motoring of mental disorders.

This paper examines the cloud resources management for a multi-agent IoT architecture. The resources tenancy is a costly operation, thus their allocation and management should be approached in an effective manner. On the other hand, the infrastructure should not be affected by the deployment or maintenance life cycle, operations that could put parts of the system offline, or even the entire system. We emphasize the need for infrastructure audit, which offers a good insight of how the resources are used, the geographical areas with an increased number of failures and where the allocation of supplementary resources is mandatory. Also, the security audit and its impact over a distributed multi-agent architecture that handles a large number of heterogeneous devices is discussed.

The Internet of Things (IoT) is the enabler of major societal changes. As an integrated part of the Future Internet, it is based on a series of enabling technologies. In the context of the IoT, "smart things/objects" are becoming more important, active players with capabilities to communicate and interact one with the other and with IoT-enabled ecosystems. Semantic annotations, things/objects search and discovery, semantic interoperability, ontology-based semantic standards, are some of the important issues to be addressed in the context of the IoT. The focus of this paper is on analyzing existing semantic approaches in the context of the Internet of Things in order to identify potential shortcomings for the issues they address.

With the current prevalence of Ubiquitous Computing, more specifically Internet of Things, almost every other appliance aims to exhibit smart behavior and expose it through internet connectivity, resulting in a plethora of sensors that expose environmental information online. As a direct result, large quantities of data are pouring online waiting to be processed. Our paper addresses this problem through an architecture which handles storage, analysis and processing of large amounts of data and can scale accordingly, providing a corner stone for Internet of Things.