Cloud-based and Software-defined Hyper-Cellular Networks

Software-defined Hyper Cellular Networks (SDHCN)

SDHCN aims at further improving the openness, smartness, and energy efficiency of 5G mobile systems and beyond. To realize this, base station (BS) functionalities need to be virtualized and implemented in a software-defined manner so that they can be easily reconfigured. Meanwhile, as more and more value-added services and application-specific functions are moved to the cloud, cloud-based RAN architecture with both Internet cloud and multi-layer edge clouds ought to be studied to better utilize both the communication and compute resources.

Cloud-based SDHCN and its Resource Management

1. Architecture Design of SDHCN[1][2]

A system-level design of the SDHCN based on deep convergence of communication, computing, and control has been proposed. Remote radio heads (RRHs) are universal hardware which can be dynamically configured as control BS (CBS) or traffic BS (TBS), or put into sleep mode. RRHs are connected to virtual BS (VBS) cloud via fronthaul switching network, which is software defined and therefore allowing flexible mapping between VBS and RRHs. The functions of CBS andTBS are realized as VBS applications in virtual machines (VMs) in the cloud. The VMs that run CBS and TBS functions can also be constructed or released on demand, saving energy for the computing cloud.

SDHCN

2. Radio and Compute ResourceJointAllocation[3][4]

Compared with Internet cloud, edge clouds have limited compute resources. Meanwhile, data between BSs and edge clouds is transmitted over bandwidth-limited backhauls. Through queueing analysis, we have established the theoretical relationship between radio/compute resources and users’ QoS and shown that the pooling of edge computing resources and sharing of fronthaul can provide significant multiplexing gain.

3. Baseband Functional Splitting and Task Offloading[5-7]

QoS can be further improved by the cooperation of edge and remote clouds. For this purpose, BSs’ functionalities and user equipment (UE) tasks should be properly partitioned and scheduled to be executed in appropriate places based on the availability of radio (both in fronthaul and backhaul networks) and compute (both in edge and remote clouds) resources. Under the mobility environment, the association to the edge clouds and the task offloading should also be jointly optimized so as to further efficiently use radio/compute resources and extend the UE battery life.


[1] J. Liu, T. Zhao, S. Zhou, Y. Cheng, and Z. Niu, “CONCERT: a cloud based architecture for next-generation cellular systems,” IEEE Wireless Commun., 21(6):14–22, Dec. 2014.
[2] S. Zhou, T. Zhao, Z. Niu, and S. Zhou, ‘‘Software-defined hyper-cellular architecture for green and elastic wireless access,” IEEE Commun. Mag., 54(1):12-19, Jan. 2016.
[3] J. Liu, S. Zhou, J. Gong, Z. Niu, and S. Xu, ‘‘Statistical multiplexing gain analysis of heterogeneous virtual base station pools in cloud radio access networks,” IEEE Trans. Wireless Commun., 15(8):5681-5694, Aug. 2016.
[4] L. Wang and S. Zhou, ‘‘On the fronthaul statistical multiplexing gain”, IEEE Commun. Lett., 21(5):1089-7798, May 2017.
[5] J. Liu, S. Xu, S. Zhou, and Z. Niu, ‘‘Redesigning fronthaul for next-generation networks: beyond baseband samples and point-to-point links,” IEEE Wireless Commun., 22(5):90-97, May 2015.
[6] T. Zhao, S. Zhou, X. Guo, Y. Zhao, and Z. Niu, “A Cooperative Scheduling Scheme of Local Cloud and Internet Cloud for Delay-Aware Mobile Cloud Computing,” IEEE GlobeCom'15, Dec. 2015.
[7] Y. Sun, S. Zhou, and J. Xu, ‘‘EMM: Energy-aware mobility management for mobile edge computing in ultra dense networks,” IEEE J. Sel. Areas Commun., 35(11):2637-2646, Nov. 2017.