Resilience by multi-connectivity network design in industrial IoT environments

点击翻译按钮获取中文摘要

Resilient operation of wireless sensor networks in industrial environments is an important requirement because it allows the wireless system to provide sufficient service level after a disruptive event and to automatically recover to a stable state. One useful technique to offer higher reliability and availability of the wireless links is to establish multiple simultaneous links and multiple edge-disjoint paths from the source to fusion nodes. The motivation for the project stems from the observation that dependencies between established links and paths significantly impact the resulting reliability. The project will contribute to the fundamental understanding of these dependency structures by applying copula theory. As a result, different radio access technologies should be deployed to achieve statistically independent connectivity. For all three phases during resilient operation, namely regular, survivability, and recoverability phase, the careful selection of multi-path and -connectivity is modeled, analyzed, and optimized. First, the project develops novel algorithms for node placement and configuration. The required service function is guaranteed by adapting to small changes in the network environment. A system for monitoring the network status and detecting disruptive events is installed. After detection of a simultaneous failure of several links and nodes in the network, the system enters the survivability phase. During this phase, the system will stabilize and prevent cascading failure, and at the same time, maintains a high service level function. Second, for both goals, multi-path and -connectivity will be optimized under a multi-objective programming framework. The concept of network extensibility is extended to multi-path and -connectivity environments and both deterministic as well as statistical optimization models are developed. Third, after stabilization of the network, a plan for the recovery phase and its corresponding target configuration is derived, deployed, and activated. The transition to this new configuration must be seamless, without halting the system or causing service interruptions, and according methods are provided. Finally, the project will demonstrate the developed algorithms on a small-scale demonstrator platform in a realistic industrial factory environment. The deployment of up to 10 nodes with at most two hop connections to the fusion centers is planned. The feasibility as well as the fundamental performance will be assessed. In system-level simulations the scalability of the approaches and the algorithms will be demonstrated, too. Overall, the project will show significant gains in terms of service improvement during survivability and recoverability phases. It will also demonstrate how to shorten the time to return to the recovered system state by multi-path and multi-connectivity techniques.