A Flexible Split Based 5G C-RAN to Minimize Energy Consumption & Handovers

Gupta, Himank and Franklin, Antony (2019) A Flexible Split Based 5G C-RAN to Minimize Energy Consumption & Handovers. Masters thesis, Indian institute of technology Hyderabad.

[img] Text
Restricted to Repository staff only until 26 June 2020.

Download (4MB) | Request a copy


The recent adoption of virtualized technologies in Next Generation Radio Access Network (NGRAN) has driven a significant impact on energy consumption. In C-RAN, the protocol stack of Next Generation NodeB (gNB) has been split into two components: Central Unit (CU) and Distributed Unit (DU). CUs are moved to central-offices/data-centers (commonly referred as BBU pool in CRAN literature) where cloud computing and virtualization mechanisms can be used as the key enabling technologies whereas the DUs remain at cell sites. The upper layers of the gNB protocol stack (e.g., PDCP and RRC) commonly reside in the CU whereas the lower layers (e.g., PHY, MAC, and RLC) are kept at the DU. Both CU and DU communicate through an interface also known as a fronthaul interface adhered to strict latency and bandwidth requirements. Several functional split options have been suggested by 3GPP for gNB. Each functional split option has got specific requirements such as fronthaul bandwidth, strict latency demand, and capability to support advanced services. Nevertheless, the functional split in C-RAN between CU and DU is flexible and influences the energy consumption in CU pool. In this work, a realtime testbed has been setup using an open-source implementation of 3GPP compliant cellular stack, OpenAirInterface (OAI) along with USRP-B210 SDR and commercial general purpose processor (GPP) servers. From the above prototype, we measure energy consumption for different split using running average power limit (RAPL) tool and model specific register MSR register. Further, we evaluates the variation of energy consumption in CU pool sure to various load at DU side. Key observations are studied from the above prototype to design and develop a mathematical model to study and evaluate the cloud-native implementation of CUs in data center environments. Further, novel and efficient optimization MILP model Apt-RAN that adopts flexible functional splits based on available fronthaul bandwidth and link delay, considering the spatio-temporal traffic fluctuation at DU to minimize the total energy consumption by minimizing the active numbers of CUs in CU pool. Moreover, Apt-RAN also reduces number of handovers and service disruption by consolidating neighbouring DUs to the same CU. The MILP model is solved with GAMS tool using standard solver and the run time performance is analyzed. To alleviate the computational heaviness of the MILP model for larger input size, we propose a lightweight heuristic algorithm that is scalable and competent for realistic deployment. This study also shows that, in Virtualized RAN, the virtualisation technologies, the functional split option, and the number of elements deployed in the same computational resource affect the latency budget available for the fronthaul network. Finally, it shows that, if virtualised DUs (vDUs) with lower layer split option are deployed in a computational resource, the number of vDUs deployed in the same resource is limited even if the other vDUs feature higher layer functional split. Moreover, results showed that lighter virtualisation methods (e.g., Docker) are impacting the fronthaul latency budget for Option 7 (i.e., intra-PHY) split less than heavier virtualisation methods (e.g., VirtualBox).

[error in script]
IITH Creators:
IITH CreatorsORCiD
Franklin, AntonyUNSPECIFIED
Item Type: Thesis (Masters)
Uncontrolled Keywords: 5G, C-RANs, CU, DU, Energy
Subjects: Computer science
Divisions: Department of Computer Science & Engineering
Depositing User: Team Library
Date Deposited: 26 Jun 2019 09:26
Last Modified: 26 Jun 2019 09:26
URI: http://raiith.iith.ac.in/id/eprint/5558
Publisher URL:
Related URLs:

    Actions (login required)

    View Item View Item
    Statistics for RAIITH ePrint 5558 Statistics for this ePrint Item