Simulating Quantum Scheduler Performance for an Entanglement Generation Switch

Abstract

The quantum internet improves upon the classical internet with several new possibilities. However, to create such a quantum internet, metropolitan hubs (for this research we will use an Entanglement Generation Switch - EGS - as hub) are needed to avoid a scaling problem when connecting all end nodes individually with one another. These EGSs share resources for multiple end nodes. Hence, resource sharing protocols are needed to manage these resources. Unfortunately due to the probabilistic nature of end-to-end entanglement generation, classical resource sharing protocols will not suffice for implementation in a quantum network. Therefore the need to design quantum resource sharing protocols arises. These protocols, where we specifically investigate quantum schedulers, should be simple and predictable to provide a comprehensible addition to current research by incorporating more parameters into the current EGS model while still attempting optimal resource efficiency. In this research we have extended quantum network simulation software to be used in this research and following projects. We have implemented a classical Max Weight scheduler with a constant time window size to simulate the inefficiency of such a classical protocol in a quantum environment and improve upon this protocol by implementing a scheduler with a dynamic time window size which scales with the link length and the probability of a photon arriving at the EGS. We investigate the entanglement successes, idle time and entanglement rates when using these schedulers and compared these with each other and with a third scheduler without a time window size but using a cutoff timer instead, the Cutoff scheduler. These results show an optimality of entanglement successes, idle time and entanglement rates for the Cutoff scheduler over the Dynamic Time Window scheduler and for the Dynamic Time Window scheduler over the Max Weight scheduler with a constant time window size. Also a decrease in entanglement rate for all schedulers is shown when increasing the link length.