An Architecture for Dependable Connectivity in OSGi-Enabled Dynamic Distributed Systems

Abstract

From air pollution monitoring to debatable surveillance for better security, dynamic distributed systems led to the birth of versatile smart environments. Dependability of such systems is challenged by the reliability of the communication links between various sub-systems. In this paper, we address this issue by designing a TCP/I P based Client-Server architecture using diverse channels, to ensure zero tolerance with regards to outage of service. The prototype system uses Raspberry Pi 3, as a remote client, to intelligently communicate with the server node by choosing one or a set of available communication links, e.g., Ethernet (LAN), Wireless-LAN (Wi-Fi), and Bluetooth channels. We further implement the system using Open Services Gateway Initiative (OSGi), a modular and interpolate-able code foundation, to withstand the challenges faced by the modern software industry e.g. complexity and scalability. Prototype system was successfully tested for hardware and software fault tolerance using different test scenarios to ensure uninterrupted service delivery. In the end, we also present a machine learning technique to mitigate the effects of severe channel hostilities for diverse channels system. The results show improvement in the quality of data transmission by exploiting the flexibility of alternate channels. We demonstrate this intelligent and seamless communication link switching technique using Support Vector Machine (SVM) in MATLAB.