User-centric and context-aware resource management in wireless sensor networks for the Internet of Things

物联网(IoT)预计将通过动态的全球信息交换连接数十亿个对象和设备，从而产生将影响我们社会方方面面的各种应用。物联网系统中的感知和数据收集任务通常由无线传感器网络(WSN)来执行，无线传感器网络是任何物联网系统的基本组成部分。然而，传感器技术的异构性、动态的用户行为，以及能源、带宽和计算方面的有限资源，使得物联网系统的设计和部署非常具有挑战性。为了解决这些问题，本研究计划提出了物联网系统无线传感器网络中有效的资源管理算法和工具。特别是，我们提出了基于上下文信息和用户体验的能量、频谱和计算资源的联合优化。我们的方法是基于整个无线传感器网络的分布式学习和适应，其中节点交换关于可用的和预测的资源、环境条件和用户行为的信息。然后使用局部和全局优化算法来优化节点、集群和网络级别的资源分配。将研究新的动态层次化集群技术以及感知即服务策略。此外，还将研究异构物联网网络中的虚拟集群，以在不同的物联网垂直领域之间架起数据孤岛。在这里，我们提出了跨WSN节点、网络边缘(雾节点)和云的数据计算资源的联合优化。

The Internet of Things (IoT) is expected to connect billions of objects and devices through a dynamic global information exchange leading to diverse applications that will impact every aspect of our society. Sensing and data collection tasks in IoT systems are usually performed by wireless sensor networks (WSN) which represent a fundamental part of any IoT system. However, heterogeneous sensor technologies, dynamic user behaviour, and limited resources in terms of energy, bandwidth, and computing make IoT system design and deployment very challenging. To address these issues, this research program proposes effective resource management algorithms and tools in WSN for IoT systems. In particular, we propose joint optimization of energy, spectrum, and computing resources based on context information and user experience. Our approach is based on distributed learning and adaptation across the WSN, where nodes exchange information about the available and predicted resources, environment conditions, and user behaviour. Local and global optimization algorithms are then employed to optimize resource allocation at the node, cluster, and network levels. New dynamic hierarchical clustering techniques as well as Sensing-as-a-Service strategies will be investigated. Moreover, virtual clustering will be investigated in heterogeneous IoT networks to bridge data silos between the different IoT verticals. Here we propose joint optimization of data computing resources across WSN nodes, network edge (Fog nodes), and the Cloud.