Distributed Radio Slice Allocation in Wireless Network Virtualization: Matching Theory Meets Auctions

Wireless network virtualization has been introduced to satisfy the ever-increasing user requirements through resource sharing, and it can reduce operating costs for the network. Virtualized resources of an infrastructure provider can be allocated as slices to mobile virtual network operators to satisfy their users' demands. Thus, an efficient resource allocation method is needed. Furthermore, existing works have mostly considered resource allocation methods using one infrastructure provider in the system model. However, in realistic and practical environments, multiple infrastructure providers should be considered so that the mobile virtual network operator can choose the appropriate infrastructure provider to maximize its revenue. Therefore, in this paper, a new approach based on matching theory and auctions is proposed for slice allocation for a system with multiple infrastructure providers. Moreover, a matching algorithm and an auction are utilized to work as the distributed methods for solving the user association problem and slice allocation problem, respectively. To connect these two problems, the user association result is used as an input of the auction model so that the mobile virtual network operator can decide on the appropriate infrastructure provider to submit the bidding value. Simulation results show that the developed solutions achieve stable matching and maximize the social welfare of all bidders.