Resource allocation in a generalized LTE air interface virtualization framework exploiting user behavior

Bo Fan,Xiao Yan,Hui Tian

doi:10.1186/s13638-016-0608-2

Abstract

Wireless network virtualization (WNV) is a promising technique to solve the ossification of current networks. In this paper, a generalized Long Term Evolution (LTE) air interface virtualization framework is proposed where virtual operators (VOs) are enabled to share the physical resources owned by the infrastructure provider (InP). This user-centric feature provides VOs with the flexibility to manage their own virtualized networks according to different user traffic demand. Hence, we introduce the “Gini Coefficient” to quantitatively characterize the user traffic behavior among different VOs. In addition, we consider bandwidth-power allocation to optimize system energy efficiency (EE). The resource allocation problem is formulated as a mixed combinatorial and non-convex optimization problem, which is extremely difficult to solve. To reduce the computational complexity, we decouple the problem into two steps. First, for a given power allocation, we obtain the bandwidth allocation. Adopting bankruptcy game model and the well-known Shapley value, a heuristic bandwidth allocation algorithm is devised. Second, under the assumption of known bandwidth allocation, we transform the original optimization problem into an equivalent convex optimization problem and obtain the optimal solution via fractional programming. Through simulation, the results of user behavior and resource allocation are jointly analyzed. The user behavior is proved to be effective and the proposed resource allocation outperforms conventional schemes.

Highlights

Internet-dependent lifeway will greatly increase the diversity and density of communication demands in the nextgeneration mobile network (5G) [1]
To guarantee fairness, the allocation strategy should be centrally handled by the infrastructure provider (InP) based on the instantaneous traffic flow demand reported from virtual operators (VOs)
The proposed allocation scheme outperforms the legacy network setup. This is caused by the multiplexing gain from virtualization, i.e., the additional flexiblity provided to VOs of scheduling PRBs from the shared resource pool

Summary

Introduction

Internet-dependent lifeway will greatly increase the diversity and density of communication demands in the nextgeneration mobile network (5G) [1]. A low user behavior coefficient indicates the traffic demand share across all the VOs follows rather even distribution, with 0 corresponding to complete equilibrium. This establishes competition among different VOs. to guarantee fairness, the allocation strategy should be centrally handled by the InP based on the instantaneous traffic flow demand reported from VOs. The InP firstly makes an assessment on the total PRB number Y to support the whole service area. The InP firstly makes an assessment on the total PRB number Y to support the whole service area This can be realized by combining the most recent traffic demand information and average channel states in previous transmit time intervals (TTIs).

1: Initialization: 2

9: BS obtains the optimal power allocation strategy as

Findings

Conclusions