A network of devices capable of transmitting quantum information called a quantum network has promising applications with vast benefits. One of the most near term achievable application is the quantum key distribution. It could be used for sharing a secret key between two end-users through any insecure authenticated communication channel. Other than sharing a secret key, a quantum network can be used for connecting quantum computers. Quantum computers have the potential to solve computational problems that their classical counterparts would require significantly more time to do so. By connecting such devices, distributed computation problems such as leader election or distributed consensus between nodes can be solved securely.

For such applications to be put into practice on a large scale, there are open practical questions still to be answered. In a quantum network, a data qubit containing quantum information is transmitted by an operation called quantum teleportation. For it, two nodes need to share entanglement. Quantum teleportation then ensures the safe transmission of a qubit by using the shared entanglement and the exchange of two classical bits. This operation, however, consumes entanglement between the nodes, changing the network topology with each served request.

In quantum networks, certain nodes share entanglement between each other. Routing in quantum networks entails determining which of these shared entanglements are used to transmit quantum information. As this proves to be a difficult task, we study quantum networks and in particular consider the problem of routing entanglement in quantum networks.

The average latency for routing entanglement in a quantum network has been studied so far using a distributed routing approach. Thus, we present numerical simulation results of centralised routing for the average latency of demands in two existing entanglement generation models. In the first, shared entanglement between nodes is created on-demand. In the second, certain nodes pre-share entanglement before the demand comes. Our results show the intuition that using a centralised routing approach in the on-demand model results in drastically more average latency than in the model with pre-shared entanglement.

Since some nodes might not be informed about the change in topology, we study the effect of the propagation of information about the network topology to nodes in the network. Our observations based on simulation results show that the average latency is significantly higher if the information about the topology is not propagated well in the network. As propagating information to all nodes in a network would be a significant load for the network, we consider information propagation within a radius and still observe a considerable decrease in average latency.

At last, we present a technique called link prediction which can be performed by any node without the need for information propagation and still achieve a considerable decrease in average latency. It can be effectively used to predict the change of topology in a quantum network by using knowledge about the network topology for a certain future point in time and previous knowledge about the network traffic.