隨著5G網路的普及，帶來了更多不同的工業與生活應用，在未來有更多的網路設備需求出現，為了滿足大量資料所需的高頻寬與及時資料要求的低延遲網路環境，加強5G資料層封包轉發機制為重要課題。與過去解決方案不同的是，本篇將資料層轉發功能交由ASIC(Application Specific Integrated Circuit)晶片處理，如此較不受限於CPU的能力限制，以硬體架構面對接下來大量資料流的網路應用。

本篇著重於5G UPF(User Plane Function)的轉發功能，利用P4可程式化的特性來實現5G資料層封包轉發。P4(Programming Protocol-independent Packet Processors)可自定義的特性使的網路應用更加廣泛，因此在本篇架構中，在5G核心網路運行後，藉由SMF(Session Management Function)與UPF之間的PFCP協議中的資料，直接提取需要的使用者資料並寫入P4 Pipeline當中，以此完成 P4-UPF 封包轉發功能。架構中也會利用一執行檔模擬控制層，其可利用P4 SDE(Software Development Environment)完成P4的編譯，並透過Barefoot-Runtime由控制層下達指令至 P4-UPF Pipeline當中。本篇架構並非直接加入現有的控制層方案，因整體架構建構在SONiC(Software for Open Networking in the Cloud)之中，利用SONiC將軟硬體分離的特性，在未來擁有更多的延伸研究。實驗結果顯示，P4-UPF相比5G UPF能夠有更高的頻寬，並且能夠提供低延遲、更穩定的網路環境，實驗結果中也呈現了P4-UPF架構亦能夠達到影音串流的網路應用。以上架構不僅提供了更多的開發靈活性，在未來也能融合P4透過使用者設計達到QoS(Quality of Service)、資料層遙測等研究，以應用在更多樣的網絡場域。

With the popularization of 5G network, it enables a new kind of network that is designed to connect virtually everyone and everything together including machines and devices,that is to say, there will be more demand for network equipment in the future. In order to meet high bandwidth and low-latency network environment, it is important to enhance the ability for packet forwarding in 5G user plane. Different from the past solutions, our architecture will offload packet forwarding function to the ASIC(Application Specific Integrated Circuit) chip from CPU, so that it is not limited by the capability of the CPU. Hardware is helpful to face the large amount of data flow with massive network applications in the future.

This article focuses on the forwarding data plane in 5G UPF(User Plane Function), and our design uses the programmable P4(Programming Protocol-independent Packet Processors) pipeline to realize fast packet forwarding in 5G data plane. The flexibility and customizable features of P4 make the network application more extensive. Therefore, in our architecture, the user data between 5G SMF(Session Management Function) and UPF will be eventually extracted into P4 pipeline. The execution file is for simulating the control plane, it will use the package in the P4 SDE(Software Development Environment) to issue commands from the control plane to the P4-UPF Pipeline through Barefoot-Runtime. This article uses this to simulate the control layer instead of directly adding the existing control layer solution to the architecture. Because the overall architecture is built in SONiC(Software for Open Networking in the Cloud), SONiC is used to decouple software and hardware. In the future, other related solutions can be used to combine with our architecture with control plane solutions to achieve better architecture extension characteristics.

The experimental results show that P4-UPF has higher bandwidth than 5G UPF, and provides a lower latency and more stable network environment. The experimental results also show that the P4-UPF architecture can achieve video streaming application. The above architecture not only provides higher flexibility, but can also achieve QoS(Quality of Service), data plane telemetry and other research with P4, also the architecture could be applied to more diverse network fields in the future.

[1] “Cisco Annual Internet Report (2018–2023) White Paper,” Cisco, Tech. Rep., 2020.
[Online]. Available: https://www.cisco.com/
[2] “IMT Vision–Framework and overall objectives of the future development of IMT
for 2020 and beyond,” ITU-R, Tech. Rep., 09 2015. [Online]. Available: https:
//www.itu.int/dms_pubrec/itu-r/rec/m/R-REC-M.2083-0-201509-I!!PDF-E.pdf
[3] R. MacDavid, C. Cascone, P. Lin, B. Padmanabhan, A. ThakuR, L. Peterson,
J. Rexford, and O. Sunay, A P4-Based 5G User Plane Function. New York, NY,
USA: Association for Computing Machinery, 2021, p. 162–168. [Online]. Available:
https://doi.org/10.1145/3482898.3483358
[4] E. Kfoury, J. Crichigno, and E. Bou-Harb, “An exhaustive survey on p4 programmable data plane switches: Taxonomy, applications, challenges, and future
trends,” IEEE Access, vol. PP, pp. 1–1, 06 2021.
[5] P. Bosshart, G. Gibb, H.-S. Kim, G. Varghese, N. McKeown, M. Izzard,
F. Mujica, and M. Horowitz, “Forwarding metamorphosis: Fast programmable
match-action processing in hardware for sdn,” in Proceedings of the ACM
SIGCOMM 2013 Conference on SIGCOMM, ser. SIGCOMM ’13. New York, NY,
USA: Association for Computing Machinery, 2013, p. 99–110. [Online]. Available:
https://doi.org/10.1145/2486001.2486011
[6] P. Bosshart, D. Daly, G. Gibb, M. Izzard, N. McKeown, J. Rexford, C. Schlesinger,
D. Talayco, A. Vahdat, G. Varghese, and D. Walker, “P4: Programming protocol independent packet processors,” SIGCOMM Comput. Commun. Rev., vol. 44, no. 3,
p. 87–95, jul 2014. [Online]. Available: https://doi.org/10.1145/2656877.2656890
[7] Accton, “Intel® DPDK Performance on the SAU5081I
Server.” [Online]. Available: https://www.accton.com/Technology-Brief/
intel-dpdk-performance-on-the-sau5081i-server/
[8] D. Scholz, H. Stubbe, S. Gallenmüller, and G. Carle, “Key properties of programmable data plane targets,” in 2020 32nd International Teletraffic Congress (ITC
32), 2020, pp. 114–122.
[9] ONF, “P4 Integrated Network Stack (PINS).” [Online]. Available: https:
//opennetworking.org/pins/
[10] 3GPP, “5G; System Architecture for the 5G System (5GS) ,” 3rd Generation
Partnership Project (3GPP), Technical Specification (TS) 23.501, 01 2021,
version 16.7.0 Release 16. [Online]. Available: https://www.etsi.org/deliver/etsi_ts/
123500_123599/123501/16.07.00_60/ts_123501v160700p.pdf
[11] “gRPC.” [Online]. Available: https://grpc.io/
[12] “ P4Runtime Specification ,” ONF (Open Networking Foundation), Tech. Rep., 07
2021, version 1.3.0. [Online]. Available: https://p4.org/p4-spec/p4runtime/main/
P4Runtime-Spec.html
[13] 3GPP, “5G; 5G System; Technical Realization of Service Based Architecture;
Stage 3,” 3rd Generation Partnership Project (3GPP), Technical Specification (TS)
29.500, 10 2018, version 15.1.0. [Online]. Available: https://www.etsi.org/deliver/
etsi_ts/129500_129599/129500/15.01.00_60/ts_129500v150100p.pdf
[14] “free5GC.” [Online]. Available: https://github.com/free5gc/free5gc
[15] DPDK, “5G-UPF Flow Based QoS using VPP and DPDK.” [Online]. Available: https://www.dpdk.org/wp-content/uploads/sites/35/2020/11/
5G-UPF-QoS-DPDK-FDIO-09-Mar-2019-2.pdf
[16] C. Kim, “Evolution of networking, Networking Field Day 21, 2:01,” Tech. Rep.,
2019. [Online]. Available: https://www.youtube.com/watch?v=-TY8qdNjc2U
[17] Open Networking Foundation, “Clarifying the differences between P4 and OpenFlow,” 05 2016. [Online]. Available: https://opennetworking.org/news-and-events/
blog/clarifying-the-differences-between-p4-and-openflow/
[18] Accton, “The Benefits of Programmable Switch ASICs.” [Online]. Available:
https://www.accton.com/Technology-Brief/benefits-programmable-switch-asics/
[19] E. C. Molero, S. Vissicchio, and L. Vanbever, “Hardware-accelerated network control
planes,” in Proceedings of the 17th ACM Workshop on Hot Topics in Networks, ser.
HotNets ’18. New York, NY, USA: Association for Computing Machinery, 2018, p.
120–126. [Online]. Available: https://doi.org/10.1145/3286062.3286080
[20] J. Vestin, A. Kassler, D. Bhamare, K.-J. Grinnemo, J.-O. Andersson, and G. Pongracz, “Programmable event detection for in-band network telemetry,” in 2019 IEEE
8th International Conference on Cloud Networking (CloudNet), 2019, pp. 1–6.
[21] “P4 language tutorials.” [Online]. Available: https://github.com/p4lang/tutorials
[22] “Open Tofino.” [Online]. Available: https://github.com/barefootnetworks/
Open-Tofino
[23] A. Nötzli, J. Khan, A. Fingerhut, C. Barrett, and P. Athanas, “P4pktgen:
Automated test case generation for p4 programs,” in Proceedings of the Symposium
on SDN Research, ser. SOSR ’18. New York, NY, USA: Association for Computing
Machinery, 2018. [Online]. Available: https://doi.org/10.1145/3185467.3185497
[24] D. Hancock and J. van der Merwe, “Hyper4: Using p4 to virtualize the
programmable data plane,” in Proceedings of the 12th International on Conference
on Emerging Networking EXperiments and Technologies, ser. CoNEXT ’16. New
York, NY, USA: Association for Computing Machinery, 2016, p. 35–49. [Online].
Available: https://doi.org/10.1145/2999572.2999607
[25] H. M. Technologies, “Implementing Load balancer functionalities –DPDK vs
P4,” Tech. Rep. [Online]. Available: https://networkbuilders.intel.com/ecosystem/
happiest-minds
[26] K. Gökarslan, Y. S. Sandal, and T. Tugcu, “Towards a urllc-aware programmable
data path with p4 for industrial 5g networks,” in 2021 IEEE International Conference
on Communications Workshops (ICC Workshops), 2021, pp. 1–6.
[27] C. Cascone(ONF), “Using P4 and Programmable Switches to Implement a 4G/ 5G UPF in Aether,” Tech. Rep., 2021.
[Online]. Available: https://opennetworking.org/news-and-events/blog/
using-p4-and-programmable-switches-to-implement-a-4g-5g-upf-in-aether/
[28] R. Shah, V. Kumar, M. Vutukuru, and P. Kulkarni, “Turboepc: Leveraging
dataplane programmability to accelerate the mobile packet core,” in Proceedings
of the Symposium on SDN Research, ser. SOSR ’20. New York, NY, USA:
Association for Computing Machinery, 2020, p. 83–95. [Online]. Available:
https://doi.org/10.1145/3373360.3380839
[29] L. Molnár, G. Pongrácz, G. Enyedi, Z. L. Kis, L. Csikor, F. Juhász, A. Kőrösi,
and G. Rétvári, “Dataplane specialization for high-performance openflow software
switching,” in Proceedings of the 2016 ACM SIGCOMM Conference, ser. SIGCOMM
’16. New York, NY, USA: Association for Computing Machinery, 2016, p. 539–
552. [Online]. Available: https://doi.org/10.1145/2934872.2934887
[30] E. H. Campbell, W. T. Hallahan, P. Srikumar, C. Cascone, J. Liu,
V. Ramamurthy, H. Hojjat, R. Piskac, R. Soulé, and N. Foster, “Avenir:
Managing data plane diversity with control plane synthesis,” in 18th USENIX
Symposium on Networked Systems Design and Implementation (NSDI 21).
USENIX Association, Apr. 2021, pp. 133–153. [Online]. Available: https:
//www.usenix.org/conference/nsdi21/presentation/campbell
[31] F. Paolucci, F. Cugini, P. Castoldi, and T. Osiński, “Enhancing 5g sdn/nfv edge
with p4 data plane programmability,” IEEE Network, vol. 35, no. 3, pp. 154–160,
May 2021.
[32] “SONiC.” [Online]. Available: https://github.com/sonic-net/SONiC/wiki
[33] 3GPP, “Universal Mobile Telecommunications System (UMTS); LTE; Architecture
enhancements for control and user plane separation of EPC nodes ,” 3rd
Generation Partnership Project (3GPP), Technical Specification (TS) 23.214, 5
2017, version 14.2.0 Release 14. [Online]. Available: https://www.etsi.org/deliver/
etsi_ts/123200_123299/123214/14.02.00_60/ts_123214v140200p.pdf
[34] 3GPP, “LTE; 5G; Interface between the Control Plane and the User Plane nodes ,”
3rd Generation Partnership Project (3GPP), Technical Specification (TS) 29.244, 11
2020, version 16.5.0 Release 16. [Online]. Available: https://www.etsi.org/deliver/
etsi_ts/129200_129299/129244/16.05.00_60/ts_129244v160500p.pdf
[35] N. Dahad, “新 式 TCAM 記 憶 體 性 能 媲 美
SRAM.” [Online]. Available: https://www.eettaiwan.com/
20181217nt01-rram-based-tcam-shown-to-match-sram-performance/
[36] “P4 Language Tutorial.” [Online]. Available: https://bit.ly/p4d2-2018-spring
[37] B. Networks, “Barefoot Networks Announces P4 Studio™ - NextGeneration Software Development Environment.” [Online]. Available: https://www.globenewswire.com/news-release/2018/09/25/1575391/0/en/
Barefoot-Networks-Announces-P4-Studio-Next-Generation-Software-Development\
-Environment.html
[38] ONF, “P416 Programming for Intel® Tofino™ using Intel P4 Studio™.”
[Online]. Available: https://opennetworking.org/wp-content/uploads/2021/05/
2021-P4-WS-Vladimir-Gurevich-Slides.pdf
[39] ONF, “Programmable Data Plane at Terabit Speeds .” [Online]. Available:
https://www.youtube.com/watch?v=DptVDfxjoSk
[40] “UERANSIM.” [Online]. Available: https://github.com/aligungr/UERANSIM
[41] Acccton, “CSP-7551 Specifications,” Tech. Rep., 2021. [Online]. Available:
https://www.accton.com/wp-content/uploads/2021/05/CSP-7551_Datasheet.pdf
[42] “Video Latency in Live Streaming.” [Online]. Available: https://aws.amazon.com/
media/tech/video-latency-in-live-streaming/?nc1=h_ls