1、 NGMN Alliance e.V.Groer Hasenpfad 30 60598 Frankfurt Germany Phone+49 69/9 07 49 98-0 Email officengmn.org PRE-COMMERCIAL NETWORK SLICING TRIALS MAJOR CONCLUSIONS by NGMN Alliance Version:1.0 Date:22.11.2022 Document Type:Final Deliverable(approved)Confidentiality Class:P-Public Project:Network Sli

2、cing for Operating Systems of 5G Smart Phones Editor/Submitter:Danni Song(China Mobile)Contributors:Hu Hao(Anritsu),Jonathan Borrill(Anritsu),Javan Erfanian(Bell Canada),Aitong Han(China Mobile),Chenguang Jin(China Mobile),Danni Song(China Mobile),Luting Kong(China Mobile),Jakob Belschner(Deutsche T

3、elekom),Ludger Verhaag(Deutsche Telekom),Marcus Mller-Jung(Deutsche Telekom),Vida Chen(Ericsson),Raymond Cheung(HKT),Stan Wong(HKT),Junda Chen(Huawei),Gary Li(Intel),Dustin Fan(MediaTek),Xiliang Yue(MediaTek),Maoki Hikosaka(NTT DOCOMO),Masahiro Tamaoki(NTT DOCOMO),Kenny Zhang(Qualcomm),Kyungsik Han(

4、SK Telecom),Timothy Senathirajah(Smart/PLDT),Lauro Roxas(Smart/PLDT),Arvin Siena(Smart/PLDT),Giulio Colombo(TIM),Cemil Karakus(Turkcell),Qianqian Lv(UNISOC),Zhenzhen Su(UNISOC),Shiyang Zheng(ZTE)Approved by/Date:NGMN Board,21st November 2022 Pre-commercial Network Slicing Trials Major Conclusions Ve

5、rsion 1.0,22November2022 Page 2(19)Abstract While several documents have already been produced on the network slicing concept,technical details,and business cases,this White Paper tackles a more challenging,complex and practical domain by consolidating the pre-commercial network slicing test results

6、 from different chipset platforms and indicates that 5G smartphones and 5G S-modules have been able to support network slicing.Tests were performed based on the testing framework White Paper 1,where all test objectives,pre-configurations,procedures,and success criteria were defined.The network slici

7、ng tests for 5G devices use different kinds of Device Under Test(DUT),including 5G smartphones and 5G Superior Universal Modules(S-Modules).Some signalling test cases and service specific test cases were executed,including the configuration and usage of Network Slice Selection Assistance Information

8、(NSSAI),interworking with Evolved Packet Core Network(EPC),UE Route Selection Policy(URSP)configuration and application,video service and FTP/Speedtest download service.However,due to the difference of devices,the test methods vary.5G smartphones can be tested directly using the applications on smar

9、tphones for URSP related test cases.5G S-Modules can be tested by AT command with the assistance of computers.Some URSP related test cases cannot be performed in 5G S-Modules.Due to limited time and product maturity,some test cases from the previously published testing framework White Paper have not

10、 been executed yet and may be included in the future update of this White Paper.Observations introduced here give the insight of some future improvements in network slicing.Pre-commercial Network Slicing Trials Major Conclusions Version 1.0,22November2022 Page 3(19)Contents 1 Introduction.4 2 Trial

11、Results.5 2.1 General.5 2.2 Trial Results for Common Requirements.6 2.2.1 The Configuration and Usage of NSSAI.6 2.2.2 UE Route Selection Policy.8 2.2.3 Interworking with EPC.10 2.3 Trial Results for Service Specific Requirements.12 2.3.1 Video Service.12 2.3.2 FTP/Speedtest Download Service.15 List

12、 of Abbreviations.19 References.19 Pre-commercial Network Slicing Trials Major Conclusions Version 1.0,22November2022 Page 4(19)1 INTRODUCTION The Network Slicing for Operating Systems of 5G Smart Phones Project Phase 2 focuses on the network slicing trials based on the actual network elements,and h

13、as the following scope:Develop a testing framework for 5G Network Slicing Devices,which allows the harmonisation of the testing methodologies among different parties conducting trials;Test 5G network slicing capabilities by lab tests/field trials with different kinds of devices,e.g.5G smartphones,5G

14、 S-Modules;Consolidate the results from different industry players and draw conclusions.Analyse the test results and provide observations that could lead to future improvement.This White Paper focuses on the results of the network slicing test with different kinds of devices,including 5G smartphones

15、 and 5G S-Modules,and provides some observations based on the test results.Sincere thanks to all the operators and manufactures for their great efforts in the network slicing trials,including China Mobile,Huawei,MediaTek,OPPO,Qualcomm,Quectel,Samsung,SK Telecom,Turkcell,UNISOC,vivo,Xiaomi and ZTE(in

16、 alphabetical order).Operators who have shared their 5G network slicing trial results in this White Paper are:Pre-commercial Network Slicing Trials Major Conclusions Version 1.0,22November2022 Page 5(19)2 TRIAL RESULTS 2.1 General Due to the difference between 5G smartphones and 5G S-Modules,some te

17、st cases are not suitable for 5G S-Modules,as shown in Table 1 below.Table 1:Test Cases Applicability to 5G Smartphones and 5G S-Modules Test Cases Smartphones S-Modules NSSAI Handling in Initial Registration NSSAI Handling during Mobility Registration Update Default Configured NSSAI Pre-configurati

18、on and Update Generic UE Configuration Update for NSSAI NSSAI Inclusion Mode A NSSAI Inclusion Mode B NSSAI Inclusion Mode C NSSAI Inclusion Mode D URSP Configuration in Initial Registration Update and Deletion of URSP Traffic Descriptors in URSP-APPID Traffic Descriptors in URSP-DNN Traffic Descrip

19、tors in URSP-IP Traffic Descriptors in URSP-IP 3 Tuple Traffic Descriptors in URSP-FQDN Traffic Descriptors in URSP-OSID+OSAPPID Traffic Descriptors in URSP-Match-all Multiple URSP Rules-URSP Rules Precedence Single URSP Rules-Route Selection Descriptor Precedence Single URSP Rules-Multiple Traffic

20、Descriptors Pre-commercial Network Slicing Trials Major Conclusions Version 1.0,22November2022 Page 6(19)Multiple URSP Rules-Concurrency of Multiple Services Handover from 5GC to EPC Redirection from EPC to 5GC Handover from EPC to 5GC Handover from 5GC to EPC,then Redirection back to 5GC Handover f

21、rom 5GC to EPC,then Handover back to 5GC Notes:indicates the test case is applicable to the DUT.indicates the test case is NOT applicable to the DUT.2.2 Trial Results for Common Requirements 2.2.1 The Configuration and Usage of NSSAI Based on the testing framework White Paper,some DUTs with the main

22、stream chipset platforms were tested.For the test cases of configuration and usage of NSSAI,5G smartphones and 5G S-Modules were tested as shown in Table 2 and Table 3 respectively,where one box means one 5G smartphone or S-Module.Take the test case NSSAI Handling during Mobility Registration Update

23、 in Table 2 as an example,four boxes as mean four 5G smartphones have executed this test case and two of them have successfully passed this test case.Table 2:Test Cases of Configuration and Usage of NSSAI for 5G Smartphones Test Cases Four 5G Smartphones NSSAI Handling in Initial Registration NSSAI

24、Handling during Mobility Registration Update Default Configured NSSAI Pre-configuration and Update Pre-commercial Network Slicing Trials Major Conclusions Version 1.0,22November2022 Page 7(19)Generic UE Configuration Update for NSSAI NSSAI Inclusion Mode A NSSAI Inclusion Mode B NSSAI Inclusion Mode

25、 C NSSAI Inclusion Mode D Notes:indicates the test case has been executed.indicates the test case has been passed.indicates the test case has NOT been executed.Table 3:Test Cases of Configuration and Usage of NSSAI for 5G S-Modules Test Cases Three 5G S-Modules NSSAI Handling in Initial Registration

26、 NSSAI Handling during Mobility Registration Update Default Configured NSSAI Pre-configuration and Update Generic UE Configuration Update for NSSAI NSSAI Inclusion Mode A NSSAI Inclusion Mode B NSSAI Inclusion Mode C NSSAI Inclusion Mode D Notes:indicates the test case has been executed.indicates th

27、e test case has been passed.indicates the test case has NOT been executed.Pre-commercial Network Slicing Trials Major Conclusions Version 1.0,22November2022 Page 8(19)Key Observations:The signalling test cases in this section are applicable for both 5G smartphones and 5G S-Modules.The DUTs,including

28、 5G smartphones and 5G S-Modules,can successfully perform the initial registration to the 5G network with the requested NSSAI or not.Some DUTs can perform mobility registration update procedure with the requested NSSAI,while some test smartphones need to be further optimised.For the inclusion mode A

29、/B/C/D test cases,four scenarios were tested for each mode,including the initial registration,mobility registration update,periodic registration update and service request.Some DUTs need to be further optimised in mobility registration update and periodic registration update scenarios.The network sl

30、icing parameter configuration of the devices has significant impact on the behaviour of the 5G devices in the network slicing test.It is necessary for the network and the devices to jointly decide on the network slicing parameter configuration.2.2.2 UE Route Selection Policy The test cases of URSP a

31、re different for 5G smartphones and 5G S-Modules.5G smartphones and 5G S-Modules were tested as shown in Table 4 and Table 5 respectively,where one box means one 5G smartphone or 5G S-Module.Take the test case Update and Deletion of URSP in Table 4 as an example,four boxes as mean four 5G smartphone

32、s have executed this test case and three of them have successfully passed this test case.Table 4:Test Cases of UE Route Selection Policy for 5G Smartphones Test Cases Four 5G Smartphones URSP Configuration in Initial Registration Update and Deletion of URSP Traffic Descriptors in URSP-APPID Traffic

33、Descriptors in URSP-DNN Pre-commercial Network Slicing Trials Major Conclusions Version 1.0,22November2022 Page 9(19)Traffic Descriptors in URSP-IP Traffic Descriptors in URSP-IP 3 Tuple Traffic Descriptors in URSP-FQDN Traffic Descriptors in URSP-OSID+OSAPPID Traffic Descriptors in URSP-Match-all M

34、ultiple URSP Rules-URSP Rules Precedence Single URSP Rules-Route Selection Descriptor Precedence Single URSP Rules-Multiple Traffic Descriptors Multiple URSP Rules-Concurrency of Multiple Services Notes:indicates the test case has been executed.indicates the test case has been passed.indicates the t

35、est case has NOT been executed.Table 5:Test Cases of UE Route Selection Policy for 5G S-Modules Test Cases Three 5G S-Modules URSP Configuration in Initial Registration Update and Deletion of URSP Notes:indicates the test case has been executed.indicates the test case has been passed.indicates the t

36、est case has NOT been executed.Pre-commercial Network Slicing Trials Major Conclusions Version 1.0,22November2022 Page 10(19)Key Observations:The test case applicability for 5G smartphones and 5G S-Modules in this section is different.Since the tested 5G S-Modules do not have applications,some test

37、cases that need to use applications are NOT applicable to 5G S-Modules but only applicable to smartphones.The DNNs tested in this section are all standardised DNN,while customised DNNs need to be further considered.For 5G smartphones:o The URSP configuration and provision test was successfully compl

38、eted.o For different types of traffic descriptors,all tested smartphones support using APPID,FQDN and IP 3 tuples to match URSP rules,and succeeded in establishing PDU session with the corresponding S-NSSAI in URSP rules.o All DUTs support the concurrency of multiple services with multiple network s

39、lices.o Not all the Traffic Descriptors provided in 3GPP specifications have been well supported,which need to be further promoted.For 5G S-Modules:o The URSP configuration,update and deletion tests were successfully completed.2.2.3 Interworking with EPC For the test cases of interworking with EPC,5

40、G smartphones and 5G S-Modules were tested as shown in Table 6 and Table 7 respectively,where one box means one 5G smartphone or 5G S-Module.Take the test case Handover from 5GC to EPC,then Redirection back to 5GC in Table 6 as an example,four boxes as mean three of the four 5G smartphones have exec

41、uted this test case and two of them have successfully passed this test case.Table 6:Test Cases of Interworking with EPC for 5G Smartphones Test Cases Four 5G Smartphones Handover from 5GC to EPC Redirection from EPC to 5GC Handover from EPC to 5GC Handover from 5GC to EPC,then Redirection back to 5G

42、C Pre-commercial Network Slicing Trials Major Conclusions Version 1.0,22November2022 Page 11(19)Handover from 5GC to EPC,then Handover back to 5GC Notes:indicates the test case has been executed.indicates the test case has been passed.indicates the test case has NOT been executed.Table 7:Test Cases

43、of Interworking with EPC for 5G S-Modules Test Cases Three 5G S-Modules Handover from 5GC to EPC Redirection from EPC to 5GC Handover from EPC to 5GC Handover from 5GC to EPC,then Redirection back to 5GC Handover from 5GC to EPC,then Handover back to 5GC Notes:indicates the test case has been execut

44、ed.indicates the test case has been passed.indicates the test case has NOT been executed.Key Observations:All tested 5G S-Modules support handover from 5GC to EPC.Most tested smartphones support handover from 5GC to EPC,and only one tested smartphone needs further improvement for the handover from 5

45、GC to EPC.Some tested 5G smartphones and 5G S-Modules support redirection from EPC to 5GC.Some tested 5G smartphones and 5G S-Modules can support handover from 5GC to EPC,then redirection back to 5GC.However,other tested devices need further optimisation.The support of the test cases in the interwor

46、king with EPC section needs to be further promoted.Pre-commercial Network Slicing Trials Major Conclusions Version 1.0,22November2022 Page 12(19)2.3 Trial Results for Service Specific Requirements 2.3.1 Video Service Figure 1:Network Topology for 5G Network Slicing Resource Reservation Test Trial Se

47、tup:Attach UE1 and UE2 to Slice#1(SST=1&SD=66052)Attach UE3 and UE4 to Slice#2(SST=1&SD=66051)Make sure that Slice#1 and Slice#2 have the same priority,which means no dedicated resources are reserved for any of them either in Radio or in Core network Pre-commercial Network Slicing Trials Major Concl

48、usions Version 1.0,22November2022 Page 13(19)Key Observations:Initial State(without resource reservation):o UE1 and UE2 started a video call over Video Server application o After cell was loaded with high UL/DL traffic via UE3 and UE4,packet retransmissions were observed on video call service as in

49、Figure 2.Packet retransmissions occurred due to lack of air interface resources.Figure 2:Retransmissions because of Packet Losses Observed between 5:30 PM and 5:31 PM Final State(with resource reservation):o Some air interface resources,e.g.40%of Resource Blocks(RBs),were assigned to Slice#1 on gNod

50、eB.o Then cell was loaded with high UL/DL traffic via UE3 and UE4 again.Because of dedicated resource per slice,no packet retransmissions were observed as in Figure 3.Pre-commercial Network Slicing Trials Major Conclusions Version 1.0,22November2022 Page 14(19)Figure 3:No Retransmissions between 5:3

51、3 PM and 5:34 PM after Second Traffic Load.After First Load,Retransmissions because of packet losses observed between 5:30 PM and 5:31 PM as in Figure 2 Conclusion:It is shown that when services are assigned to different network slices with dedicated resources assigned,QoS can be assured.Otherwise s

52、ervices are affected because of other traffic e.g.under high traffic load.Pre-commercial Network Slicing Trials Major Conclusions Version 1.0,22November2022 Page 15(19)2.3.2 FTP/Speedtest Download Service 2.3.2.1 Radio Resource Reservation for Slice Trial Setup:Similar test setup with video service

53、slice reservation as in Figure 1 Attach UE1 and UE2 to Slice#1(SST=1&SD=66052)Attach UE3 and UE4 to Slice#2(SST=1&SD=66051)Air interface resources,e.g.40%of downlink RBs,were assigned to Slice#1 and 60%of downlink RBs were assigned to Slice#2 on gNodeB.NR cell capacity was intentionally reduced(e.g.

54、lower MIMO order,lower modulation)to load cell easily during the test to make sure all the RBs were used.This is why lower maximum throughputs were observed during the test.Key Observations:Slice#1 terminals(UE1 and UE2)got approximately 1,5 times lower DL throughput than Slice#2 terminals(UE3 and U

55、E4)after resource reservation to slices as in Figure 4.It is shown that when air interface resources were assigned to specific network slices,the UEs got DL throughput proportional to the assigned RB resource ratio.Pre-commercial Network Slicing Trials Major Conclusions Version 1.0,22November2022 Pa

56、ge 16(19)Figure 4:Slice#1 Terminals(UE1 and UE2)on the Left Side Approximately Get 1,5 times Lower DL Throughput than Slice#2 Terminals(UE3 and UE4)on the Right Side after Resource Reservation to Slices.2.3.2.2 Core Throttling for Specific Slice Network Topology:Figure 5:Network Topology for 5G Netw

57、ork Slicing in SK Telecom Pre-commercial Network Slicing Trials Major Conclusions Version 1.0,22November2022 Page 17(19)The Service Orchestrator manages the slice template,SLA attribute,life cycle,and supports the fast network provisioning and network slice configuration of network functions for the

58、 B2C/B2B services which have different traffic characteristics.UDM,NRF,NSSF and PCF have the network function information,network slice configuration and customers subscription.The gNBs and core network equipments come from different vendors and jointly provide services to different network slices a

59、s defined by network slice policy and subscription of customers.Speed Test App is capable of URSP defined by App-Id and can handle traffic route toward the network slice for the specific app service.Trial Setup:Network Slice#1(fast speed,50Mbps):URSP(set by DNN)o Traffic descriptor type=DNN o o Defa

60、ult rule:downlink 50Mbps/uplink 50Mbps Network Slice#2(slow speed,2Mbps):URSP(set by App-Id)o Traffic descriptor type=OS APP ID o skt.speedtest.ursp o Default rule:downlink 2Mbps/uplink 2Mbps Speed Test App and File Server o Develop the speed test application for UE URSP test o Parsing the URSP and

61、extracting the traffic descriptor to classify the traffic Key Observations:Speed test app downloaded a file from the file server at a speed of approximately 47.9Mbps on network slice#1 which was defined by network slice policy of DNN.o Traffic descriptor type=DNN Speed test app downloaded a file fro

62、m the file server at a speed of approximately 2Mbps on network slice#2 which was defined by network slice policy of App-ID.o Traffic descriptor type=OS APP ID Pre-commercial Network Slicing Trials Major Conclusions Version 1.0,22November2022 Page 18(19)Speed URSP App and Speed internet App worked si

63、multaneously in the same UE.End-to-end service control was possible through the UEs service detections and specific network slice allocation function.Figure 6:Speed Test Results(a)Speed Test App “Speed Internet”App:Execute and test the DNN based DL/UL speed “Speed URSP”App:Execute and test the App-I

64、D based DL/UL speed(b)Network Slice#1 DNN based network slice policy Downlink 50Mbps/uplink 50Mbps(C)Network Slice#2 App-ID based network slice policy Downlink 2Mbps/uplink 2Mbps Pre-commercial Network Slicing Trials Major Conclusions Version 1.0,22November2022 Page 19(19)LIST OF ABBREVIATIONS 5GC 5

65、G Core Network 5G S-Module 5G Superior Universal Module DL Downlink DNN Data Network Name DUT Device Under Test EPC Evolved Packet Core Network FQDN Fully Qualified Domain Name MIMO Multiple-Input Multiple-Output NR New Radio NRF Network Repository Function NSSAI Network Slice Selection Assistance I

66、nformation NSSF Network Slice Selection Function PCF Policy Control Function QoS Quality of Service RB Resource Block SLA Service Level Agreement S-NSSAI Single Network Slice Selection Assistance Information UDM Unified Data Management UL Uplink URSP UE Route Selection Policy REFERENCES 1 NGMN Alliance,Definition of the Testing Framework for 5G Device Network Slicing Pre-commercial Trials,March 2022