Fluid Service Abstraction for Large-Scale Cloud IoT Systems

The next frontier for innovation in the area of Internet of Things (IoT) is the creation of large-scale smart applications such as smart cities, buildings, energy, and health. As such applications are proliferating into a massive scale, thereby, their data (i.e., IoT data) and services are being pressed to move to the Cloud to enable powerful processing and sharing of IoT data beyond the capability of individual IoT devices. Moreover, to handle mobility of IoT devices and low-latency of time-critical tasks, cloud platforms have been recently extended to the edge of the network-Fog Computing. Using fog devices, IoT services run on resources throughout the network, including industrial controllers, routers, and switches. Therefore, the huge number of IoT services in smart applications, accessible over end IoT devices, fog nodes and the Cloud, will be used by and shared among IoT applications. The main goal of the DILUTE project is to devise novel software services engineering principles and development models that enable highly dynamic and scalable integration of smart IoT devices and their services into cloud and fog platforms. To this end, DILUTE will investigate: i) how to design, develop, and integrate IoT services when massive IoT devices are accessible in hierarchies of network nodes from IoT devices toward fogs and the Cloud, and ii) how to manage and dynamically adapt these services with respect to the mobility of IoT devices and the dynamic availability of fog nodes. DILUTE will achieve these through a new software services engineering and management model for next-generation IoT systems, called Fluid Services. A Fluid Service is distributed across network devices with different capabilities and its constituents can migrate from one node in the network to another node for performance or context awareness reasons. Results achieved so far include a new modeling concept for the development of software services over IoT-Fog-Cloud computing platforms. This model is focused on addressing data-centric nature of IoT services and processing them on intermediate Fog devices. In addition, we have worked on the mobility models, and migration of data processing tasks (proactive and reactive service migration) on fog/edge devices using machine learning techniques. This has been further extended with an intelligent computation offloading architecture with service caching, considering both vertical and horizontal computation offloading. We have also developed an efficient data-flow programming model and task processing model for IoT-fog-cloud computing continuum. In 2023, we mainly focused on predicting resource demand and resource reservation in edge computing systems when migrating the software services from one node to another node.

The creation of smart spaces (e.g., smart cities, energy, health, etc.) is the next frontier for innovation in the IoT domain. Smart spaces, such as smart cities and smart buildings, are proliferating into a massive scale, thereby, IoT data, services and applications are being pressed to move to the Cloud. IoT Cloud integration can enable ubiquitous cyber-physical services and powerful processing of IoT data beyond the capability of individual things. Moreover, this has been recently extended from the core of the network to the edge of the network (i.e., Fog Computing) to address better mobility support, location-awareness and low latency. This has resulted in running IoT application logic on resources throughout the network, including routers and dedicated computing nodes. Therefore, the huge number of IoT services in smart spaces accessible over end IoT nodes, fog nodes and the Cloud, will be used by and shared among a wide range of applications. This will also include cooperation of IoT devices and their ability to become orchestrated into on-demand services accessible in the Cloud. Thus, a crucial need in future smart spaces is scalable and dynamic integration of smart devices into cloud and fog platforms. The main contribution of this project is to conduct basic research in the area of integration of future large-scale IoT applications into fog and cloud computing platforms. It develops knowledge, software service design concepts and mechanisms for scalable and dynamic integration of huge number of things and their services into future cloud-based IoT systems and the intermediate fog computing platforms. It develops solutions for: 1.modeling, developing, and integrating IoT services when massive IoT devices expose their services to hierarchies of computing nodes from IoT devices toward fogs and the Cloud 2.managing and adapting these services with respect to dynamicity of IoT devices in the environment and the dynamic availability of fog resources.