Auction Based Resource Allocation For Wireless Virtualization In 5G Networks

The popularity of intelligent terminals has accelerated the development of 5G networks with higher communication efficiency.While implementing the new 5G network architecture,wireless virtualization which enables the decoupling of infrastructure from the services it provides has become an important enabling technique in 5G cellular networks for increasing resource utilization and lowering the cost per bit.However,in view of the characteristics of the wireless network environment and the fierce business competition,resource allocation has become a difficult research problem in the process of virtualization.How to design an efficient resource allocation mechanism that is suitable for the market,while satisfying the inherent requirements of inter-slice isolation and the ability of intra-slice customization is the key to solve the problem.As a typical representative of the mechanism design,auction has been widely used in the research of wireless resource allocation.Many dynamic and static efficient mechanisms have been designed and implemented.In this paper,we consider a multi-dimensional resource market among multiple mobile virtual network operators(MVNOs)and users.The main research is as follows:1.A design of 5G network virtualization resource allocation mechanism based on combinatorial double auction(CDA).Several existing works ignored the MVNOs' participation in practical business operation,and only focuses on the interests of sellers regardless of the wishes of users.What's more,most of these mechanisms involves only a single type of resource.This paper designs a multi-user multi-MVNO CDA framework,which is more suitable for the future 5G network business model.At the same time,we consider the priority of different quality of service(QoS)requirements of users,and design a heuristic algorithm and pricing scheme with low computational power consumption to ensure that the mechanism has good allocation efficiency and some economic attributes,such as truthful,individual-rational and budget-balanced.2.A design of dynamic virtualization resource allocation considering time-evolving valuation of users.Most of the existing auction mechanisms assume that users have accurate information about their bidding without regard for the impact of their own experience on the assessment of resources.Under the infinite time step,the users' valuation information is dynamically evolved over time according to the Markov process,so that the problem is translated into a special Multi-armed Bandit Problem(MABP)with multiple players under different charging places.We design a double Gittins index allocation algorithm and an average pricing scheme to take full account of the impact of the current decision on the future.Finally,we prove our mechanism has some desired economic properties including periodic ex post truthful periodic ex post individual-rational and periodic ex post budget-balanced.The simulation results show the effectiveness and superiority of the proposed mechanisms.