Anticipatory Resource Allocation and Trading in a Sliced Network

Dynamically sharing the network resources in a sliced multi-tenant network can provide cost efficient solutions that are able to guarantee specific service requirements for 5G and beyond networks. By automatizing the negotiations between tenants and infrastructure providers over the shared resources, it is possible to maximize the flexibility of the network and stakeholders can negotiate on a very short time scale, thus maximizing also the efficiency of the sharing. However, negotiating resources in a reactive manner can bring high risks to the tenants and also limit the gain in terms of spectral efficiency for the infrastructure provider. Therefore, in this paper we focus on how to exploit the anticipatory information of the users' future channel states to improve the efficiency of the proposed dynamic network slicing and trading framework. In particular, we analyze how to integrate a well-known prediction algorithm into our system and analyze the techno-economic impacts of the negotiations, when the tenants exploit such anticipatory information. Finally, we develop a novel filter in order to limit the impacts of prediction errors, while exploiting the predicted information. Our results prove that using anticipatory information in dynamic negotiations and resource allocation increases the tenants' utilities, while allowing the infrastructure provider to accommodate more tenants.