Survivability Assessment of 5G Network Slicing During Massive Outages

Mobile networks support variety of heterogeneous services, including the emergency and mission-critical ones. The next generation of mobile networks introduces the concept of network slicing where different services can have a dedicated, logically separated virtual network running over a shared physical infrastructure. Each slice may have a specific set of functional and non-functional requirements including performance, security, resilience, and survivability. Given the importance of emergency services during massive outages caused by a natural disaster, the network operators need an efficient way to evaluate the performance of the sliced network in such adverse circumstances. In this paper, we describe how survivability quantification framework can be applied to assess and compare the performance of different slicing configurations during and after massive outages. We demonstrate our proposal in a simplified use-case scenario where the performance metric for each stage of the recovery is represented with delay and throughput of the clients at a sliced, shared bottleneck. The metrics are acquired from OMNeT++ simulations. Survivability is then obtained from an analytical model and the time until the critical services (for the first responders) are recovered is of particular interest. In the scenario we consider 8 application types, 4 priority levels, and 5 approaches to map clients to slices. The results show significant performance variations between different slicing configurations, both for the critical and non-critical applications and thus highlight the importance of having a slicing configuration optimally tailored to the use case.