< Terug naar vorige pagina


Software Technologies for Multi-Tenant Internet of Things Platforms

Boek - Dissertatie

The Internet of Things (IoT) is an evolution in the information technology landscape where intelligence is being embedded in everyday Things through computation, communication, actuation, and sensing capabilities.The rapid growth and unique characteristics of the IoT introduce new challenges due to the heterogeneity in hardware and software and a lack of lightweight mature software technologies.IoT infrastructures cannot sustain their growth without flexible, efficient, and secure software for both platforms and networks.To achieve the vision of a ubiquitous and powerful IoT, industry and research must provide cost-effective solutions with a clear return-on-investment that enable the IoT as a secure, open, and reusable infrastructure on which various stakeholders can safely provide their services.Example applications of this vision are smart homes, smart cities, and Industry 4.0, which bridge the physical and virtual worlds and provide new opportunities to improve society and industry.State-of-the-art IoT software technologies do not provide complete solutions for the secure and open use of all platforms in the IoT.Partial solutions given by related work originate from both mainstream computing and prior wireless sensor network research, but do not tackle the characteristics of the IoT in particular.The cloud model is a proven mainstream computing approach to securely and transparently scale and share infrastructure.The adaptation of the cloud to the IoT; often called edge computing, is pushing intelligence to the edges of the IoT network and is gaining considerable research attention.However, current efforts are not suitable for the constrained edges of the Internet-enabled network, class-1 IoT devices, and thus do not provide holistic secure and shareable solutions for all IoT platforms.Likewise, prior research for embedded devices from wireless sensor networks supplies lightweight security methods and multi-tenancy approaches, but fails to consider the broader Internet-enabled spectrum.For the further development of the IoT, software technologies are needed that advance, integrate, and adapt approaches from both fields while explicitly considering the IoT and its characteristics.This dissertation presents four contributions that support edge computing by developing software technologies for multi-tenant IoT platforms.The first contribution, called CerberOS, provides the first IoT operating system that enables resource security and multi-tenancy for even constrained class-1 IoT devices.CerberOS supports third-party services, programmed in a high-level language, to safely share constrained platforms without requiring any special hardware or software modules.The second contribution, called Polyglot CerberOS, advances the previous contribution by providing isolation, resource security, and secure interactions between services in multiple languages on class-1 IoT platforms.Polyglot CerberOS allows developers to use different programming languages to realise resource-secure services while supplying equal expressiveness, interoperability, and high-level language benefits.The third contribution is a zero configuration group communication framework that exploits the IPv6 specification standard and group management technologies to enable lightweight service group configuration and communications.The fourth and final contribution, called Niflheim, is a unifying middleware that provides end-to-end modular microservice-based application orchestration across all tiers of the IoT.Niflheim makes efficient use of heterogeneous IoT platforms, such as CerberOS, and provides a user-friendly graphical interface for application specification and deployment.The results of these contributions are a suite of software technologies that are a powerful enabler for bringing cloud computing towards the extreme edges of the IoT, promoting the use of shareable, secure, and flexible platforms by the various stakeholders.Each contribution has been experimentally validated on real hardware in representative IoT deployments through real-life use cases.Evaluation shows that these technologies are suitable for constrained class-1 IoT devices while maintaining multi-year battery lifetimes and providing security, multi-tenancy, and interoperability.Further, experiments with up to 150 devices demonstrated the feasibility and potential of both Niflheim and CerberOS as technologies to improve IoT infrastructure utilisation and scalability through secure interoperable microservice-hosting platforms.
Jaar van publicatie:2019