Resource Allocation in Virtualized Sensor Networks for Highly-Deployed IoT Services
Ismael Al-Shiab · 2024
As IoT technology evolves, applications become more complex, incorporating multiple sensing services and varied qualities. Users expect fast and dependable deployment without worrying about the inherent complexities and limitations in sensing and routing resources. IoT sensor nodes face challenges such as limited resources, high heterogeneity, interoperability issues, intermittent communication, and installation time and cost. Current IoT resource management techniques do not fully meet the requirements of current and future IoT applications. Virtual Sensor Networks (VSN) present a novel approach to IoT resource management, slicing the Physical Sensor Networks (PSN) and forming virtual groups that link the needed IoT sensing and routing resources. These VSN can be initiated and released on demand, allowing for the multi-use of existing sensor node resources, thereby addressing the unique requirements of IoT applications and users. The current state of IoT-enabled VSN reveals a gap in the availability of systems that can successfully integrate optimized sensing resource allocation with dynamic resource-aware routing. Specifically, systems that can adapt the allocation and routing paths in tandem, considering the sensor node’s resource utilization, are needed. This thesis addresses the practical challenge of optimizing sensor node selection for IoT applications and forming more efficient routing paths. It proposes multiple resource-aware joint techniques with both single-objective and multi-objective weighted metrics. These techniques are not just theoretical but have been refined through a weight search and empirical analysis, making them applicable in real-world scenarios. The proposed techniques end with a generalized multi-objective routing (static and dynamic) and dynamic multi-objective sensor node selection. The routing and sensor node selection algorithms are based on the resources in the sensor nodes (such as energy, memory, and processing) and the available network communication bandwidth. The proposed methods have been rigorously tested in various scenarios, including homogeneous and heterogeneous sensor networks, grid and random node placement topologies, with and without interference consideration, and heterogeneous IoT applications. The simulation results demonstrate the efficiency of the proposed methods in slicing the sensor network resources, increasing the application deployment rate, and improving the minimum node energy. These findings promise a sustainable and more efficient use of IoT sensor networks.