An enhanced multi-protocol middleware solution for internet of things

Abstract

In Internet of Things (IoT), data is handled and stored by software known as middleware (located on a server). The evolution of the IoT concept led to the construction of many IoT middleware, software that plays a key role since it supports the communication among devices, users, and applications. Several aspects can impact the performance of a middleware. Based in a deep review of the related literature and in the proposal of a Reference Model for IoT middleware, this thesis proposes a new IoT middleware, called In.IoT, a scalable, secure, and innovative middleware solution based on a deep review of the state of the art and following the reference middleware architecture that was proposed along with this research work. In.IoT addresses the middleware concerns of the most popular solutions (security, usability, and performance) that were evaluated, demonstrated, and validated along this study, and it is ready and available for use. In.IoT architectural recommendations and requirements are detailed and can be replicated by new and available solutions. It supports the most popular application-layer protocols (MQTT, CoAP, and HTTP). Its performance is evaluated in comparison with the most promising solutions available in the literature and the results obtained by the proposed solution are extremely promising. Furthermore, this thesis studies the impact of the underlying programming language in the solution's overall performance through a performance evaluation study that included Java, Python, and Javascript, identifying that globally, Java demonstrates to be the most robust choice for IoT middleware. IoT devices communicate with the middleware through an application layer protocol that may differ from those supported by the middleware, especially when it is considered that households will have various devices from different brands. The thesis offers an alternative for such cases, proposing an application layer gateway, called MiddleBridge. MiddleBridge translates CoAP, MQTT, DDS, and Websockets messages into HTTP (HTTP is supported by most IoT middleware). With MiddleBridge, devices can send a smaller message to an intermediary (MiddleBridge), which restructures it and forwards it to a middleware, reducing the time that a device spends transmitting. The proposed solutions were evaluated in comparison with other similar solutions available in the literature, considering the metrics related to packet size, response times, requests per second, and error percentage, demonstrating their better results and tremendous potential. Furthermore, the study used XGBoost (a machine learning technique) to detect the occurrence of replication attacks where an attacker obtains device credentials, using it to generate false data and disturb the IoT environment. The obtained results are extremely promising. Thus, it is concluded that the proposed approach contributes towards the state of the art of IoT middleware solutions.