Characterizing the Effects of Base Station Variable Capacity on 5G Network Slicing Performance

Network slicing is a technique based on virtualization of base station resources that allows to divide the network into logical parts (slices). %This opens up the possibility for third parties to act as virtual operators in the provider's infrastructure. In 5G, traffic with very heterogeneous transmission requirements is expected, which will need to be transmitted efficiently over a common infrastructure on the millimeter waves, which is quite sensitive to blockages. In this work, we focus on to such an aspect of network modeling as the abstraction of the base station resource, which depends on the communication channel state. Quite often, it is presented by researchers in the form of a constant value of the data transfer rate in some units. However, in reality, the behavior of the channel is complex and this value is changing. We explore our resource allocation scheme among slices through simulation and characterize the effect of the variable capacity of the base station on the stability of the network slicing technology. We conclude that the variable capacity requires a careful choice of the re-slicing frequency. This can be achieved by invoking re-slicing in in timely manner, upon arrival of sessions, or by choosing an appropriate interval.