Effects of Contention and Delay in a Switched Ethernet Evolved Fronthaul for Future Cloud-RAN Applications

Communications Research Group

Funded by

1. Introduction 4.Testbed Results

1.N. J. Gomes, P. Chanclou, P. Turnbull, A. Magee, and V. Jungnickel, “Fronthaul evolution: From CPRI to Ethernet,” Optical Fiber Tech., vol. 26, par. A, pp. 50-58, Dec 2015

2.3GPP, “3GPP TR 38.801 v14.0.0 Study on New Radio Access Technologies: Radio Access Architecture and Interfaces”, 2017

3.P. Assimakopoulos, M.K. Al-Hares, and N.J. Gomes, “Switched Ethernet fronthaularchitecture for cloud-radio access network”, OSA/IEEE J. Optical Commun. and Netw., vol: 8, no. 12, pp. B135-B146, Dec. 2016

4.K. Miyamoto, S. Kuwano, T. Shimizu, J. Terada, A. Otaka, "Performance Evaluation of Ethernet-based Mobile Fronthaul and Wireless CoMP in Split-PHY Processing,“ OSA/IEEE J. Optical Commun. and Netw., vol: 9, no. 1, pp. A46-A54, Jan. 2017

5.“OpenAirInterface (OAI)”, OpenAirInterface Software Alliance [Online]. Available: http://www.openairinterface.org/

6. Bibliography

5. Conclusions

2. Functional Split for the Evolved Fronthaul

3. Evolved Fronthaul Testbed

Project partners: Viavi, University of Essex, ADVA Optical Networking, Orange Labs, Fraunhofer-HHI, Telekom Slovenije, Primetel, Wellness Telecom, IAF GmbH, Interdigital, BT/EE, Qualcomm, NEC, Techgate

Split traffic data rates and overheads

Philppos Assimakopoulos, Gurtej S. Birring, Nathan J. Gomes

Fronthaul is evolving. New fronthaulInterfaces mean that in addition tosampled IQ (time-domain) waveforms,higher RAN protocol layerinformation may be transported.

At higher RAN layers, theinformation may consistof user data, control,primitives and timing/synchronization.

Classes of service may be identifiedfor different users and informationtypes.

3GPP definitions for RAN functional split points are shown. The CPRI corporation has recently defined eCPRI defining, transport for higher layer split points. Around the MAC/PHY split there is a compromise, allowing DL coordinated/joint and reduced data rates, but with strict timing/synchronization requirements.

The fronthaul testbed is based on a 4G network and uses traffic generated from a MAC/PHY split implemented in an OpenAirInterface software base station. The data at the interface is mapped, using a Fronthaul Interface Library, into Ethernet frames and these share trunk link(s) between Ethernet switches with background traffic (which may emulate other split or IQ traffic). VLANs are used to identify the different flows, and packet types the type of information within the MAC/PHY split being transported.

Packet generation, buffering and sampling points for measurement

Ethernet packet (frame) delay is less than the delay for the whole LTE subframe, but the delay variation is smoothed out over the subframe. Figures above show for 200 Mb/s bursty background traffic with different burst sizes, left figure shows for 400 Mb/s constant bit rate background traffic.

Packet and subframe delays have been measured when using Strict Priority (SP) queuing in the Ethernet fronthaul switch, to favour the LTE traffic over the background traffic. SP has been suggested for fronthaul in bodies such as IEEE 802.1CM. Only small improvements to packet delay, and generally lesser improvement to subframe delay were observed, even at high utilisation levels.

Evolved Ethernet fronthaul carrying MAC-PHY RAN functional split interface data has been demonstrated.

Delay of Ethernet packets, including individual packet types carrying different split interface data have been measured, as has total LTE sub-frame delay over the fronthaul. The latter is crucial for maintaining over-the-air transmission timing and radio network performance.

Strict priority queuing has been shown to provide only limited improvement (delay reduction), even at high background traffic loads, when the functional split traffic is relatively small.