Decentralized management schemes for real-time and reliable communication in industrial wireless sensor and actuator networks

Abstract

Current wireless technologies for industrial applications, such asWirelessHART and ISA100.11a, use a centralized management approach in which a central network manager handles the requirements of the static network. However, such a centralized approach has several drawbacks. For example, it cannot cope with dynamicity/disturbance in large-scale networks in a real-time manner while it also incurs a high communication overhead and latency for exchanging management traffic. In this thesis, we address the drawbacks of the centralized management approach utilized in WirelessHART and ISA100.11a for real-time industrial monitoring and control applications. More specifically, we propose new decentralized network management schemes to provide an end-to-end reliable and real-time communication for battery-powered and harvested-powered devices in a distributed manner. These schemes enable the network devices to join the network, schedule their communications, establish end-to-end connections by reserving communication resources to address real-time requirements, and cope with network dynamicity (e.g., node/edge failures) in a distributed manner. To evaluate wireless protocols in the domain of industrial monitoring and control, a reference point is needed. To that end, we developed aWirelessHART simulator in NS-2 as a reference point to evaluate other protocols. We validated theWirelessHART simulator with a WirelesHART deployment at an industrial plant. To the best of our knowledge, this is the first implementation that supports the WirelessHART network manager as well as the whole stack of the WirelessHART standard. To address the requirements of battery-powered I/O devices, we propose a distributed management scheme to address real-time and reliable communication requirements. This scheme considers the full mesh topology in which I/O devices are capable of participating in routing and distributed network management tasks, such as communication resources scheduling. We then propose a second distributed management scheme for hybrid networks to be used for real-time industrial wireless automation. This scheme addresses the requirements of energy constrained I/O devices. In this scheme, the I/O devices cannot participate in routing and distributed management tasks. The routers can dynamically reserve communication resources and manage the I/O devices in the local star sub-networks. We demonstrate that the proposed scheme achieves higher network management efficiency compared to the ISA100.11a standard, without compromising the latency and reliability requirements of industrial wireless networks. To better support and address the requirements of energy harvested I/O devices, we extend ISA100.11a. The proposed extension makes management more decentralized by delegating a part of the management responsibility to the routers in the network. It also allows the I/O devices to choose the best routers according to different metrics using local statistics and advertised routers’ ranks.