Vol. 40 - Methods to Increase Reliability and Ensure Determinism in a White Rabbit Network

The current control and timing system at the European Organization for Nuclear Research (CERN) has been serving its accelerators for several decades and is reaching its design limits. In preparation is the next generation system, called White Rabbit. White Rabbit is intentionally based on commonly used networking technologies to ensure the flexibility, maintainability and wide commercial support that were missing in the old system. The new system is meant to coordinate the actions of thousands of individual devices constituting the CERN accelerator complex in a timely manner for several decades. At the time of White Rabbit’s conception in 2008, none of the existing networking standards could provide the unprecedented characteristics required by a future-proof accelerator control and timing system. Therefore, the most suitable solutions needed to be enhanced with new specialized services. Two non-existent enhancements are proposed and developed in the context of this thesis. The first ensures that critical information to coordinate accelerator actions is delivered to all the devices within a specified time – it ensures the network’s determinism. The second increases the probability that the system works without interruption for at least a year – it increases the network’s reliability. The methods to provide the two specialized services, along with the new network design guidelines, are proposed in this thesis. The following standards were chosen for enhancements: Ethernet-based Virtual Local Area Network, Shortest Path Bridging and Precision Time Protocol . These standard technologies are used in a network arranged into a topology that provides many alternative paths. As a result of this work, specialized hardware to switch between the faulty and the backup paths fast enough so the accelerators’ devices experience no disruption has been implemented. Similarly, the specialized hardware ensures the timely delivery of information through the network. The proposed strategy describes the network topologies in which the devised methods are allowed. This strategy and methods permit the White Rabbit network to control all CERN accelerators. The performance of the developed enhancements exceeds that of the standard solutions while being interoperable with them. When switching between alternative paths, both the synchronisation performance and the reduction in data loss are improved 1000-fold compared to the best-known implementations of the original standards. The worst-case time of information is delivery is improved fourfold. The White Rabbit devices implementing the developed enhancements can be interconnected and work with standard implementations, but in such cases the benefits of the enhancements are lost. The fact that White Rabbit is based on and interoperable with well-known technologies makes it a preferred generic solution to many control, acquisition and synchronisation problems. At CERN, with certain enhancements presented in this thesis, White Rabbit is currently used for diagnostics of the Large Hadron Collider accelerator and the distribution of the magnetic field in the Proton Synchrotron accelerator. Its integration into the CERN control and timing system will take place over the coming years. Outside CERN, the proposed network strategy is already being applied in the design of the Large High Altitude Air Shower Observatory in Tibet where White Rabbit is used for synchronisation and data acquisition. The list of other applications is long and growing. In fact, White Rabbit has the potential to become a commonly used technology not only for controlling accelerators, but also in other applications. In many cases, the specialized services presented in this thesis will play an important role.