Deep Reinforcement Learning for Edge Computing Resource Allocation in Blockchain Network Slicing Broker Framework

Abstract

With the constant development of 5G technology, such as softwareization and virtualization, the novel concept of network slicing has been appeared. Blockchain is a decentralized technology for managing transactions and data which can ensure the security of transaction. Recently, the classical mobile network introduces a new role such as network slice agent to provide slices for one or across multiple operators, providing services for users or vertical industries over a larger time and space range. In this paper, we propose the blockchain network slicing broker (BNSB), an intermediate broker, which receive resources request and response then allocation resources between Network Slice Tenants (NST) and then schedule physical resources from Infrastructure Provider (InP) through smart contracts. The topology information is obtained according to the Complex Network theory and the value of nodes is defined according to their importance. In addition, Deep Reinforcement Learning algorithms is used to explore the optimal policy under the condition of meeting the service Level Agreement (SLA).