1、S P O N S O R S P R O D U C E D B YI N P A R T N E R S H I P 5G NETWORK however, all major global regions were represented. At places in this analysis, Heavy Reading compares responses from different demographic groups. In particular, at several points in this report, we compare U.S. and Rest of Wor

2、ld (RoW) responses. Where this is the case, it is noted in the text. n Gabriel Brown Principal Analyst Mobile Networks for instance, there is not yet good evidence of this in Europe and the U.S. This may be because first-generation devices tend to come with compromises (e.g., on power consumption, c

3、ost, and bugs). Looking into 2020 and 2021, newer handset models at high- and mid-tier prices will become available in volume. For example, a 5G iPhone rumored for late 2020 will be important, particularly in the U.S., where iPhone market share is high. As established earlier, over a 5-year view, op

4、erators see 5G addressing new markets and driving advanced services (Fig 7). sought insight into the differences between 5G and 4G service portfolios over a 3-year view. A fair summary would be that operators expect some differences, but not major ones. A large 43% said their company will offer a “v

5、ery similar services portfolio” for 4G and 5G users, while a comparable 45% believe their portfolio will offer “mostly common services, with some 5G-only services.” Only 8% expect to offer “many 5G-only services.” In part, this result may reflect that 5G deployed in non-standalone (NSA) mode makes e

6、xisting 4G services faster rather than fundamentally different. As discussed later, it may be that a transition to standalone (SA) is a prerequisite for service innovation. 2020.6% 2021 to 2022.43% 2023 to 2024.37% 2025 or later .13% Identical service portfolio .5% Very similar services portfolio .4

7、3% Mostly common services, but with some 5G-only .45% Many 5G-only services.8% In 12 months .12% In 12 -24 months .33% After 24 months .29% Dont know / no plans at present .27% Fig 6. By what year do you estimate more than 25% of your subscriber base will have a 5G compatible device? (N=158) Fig 7.

8、How much commonality do you expect between your 4G and 5G service portfolio over the next three years? (N=157) Fig 8. When do you expect your company to start offering URLLC (ultra-reliable, low- latency communication) services? (N=157) “ULTRA-RELIABLE LOW LATENCY COMMUNICATION (URLLC) SERVICES ARE

9、ONE OF THE DEFINING FEATURES OF 5G. URLLC REQUIREMENTS WERE INFLUENTIAL IN THE DESIGN OF THE 5G SYSTEM AND AIR INTERFACE.” 5G NETWORK it will take a year or so for the functionality to be incorporated into products and then deployed into networks. And clearly not everyone is bullish. Over half of re

10、spondents think it will be more than 2 years before URLLC will be introduced (29%) or they do not know or said their company does not have plans at present (27%). This accords with a view that URLLC is still some time from being ready for broad-based marketing to customers. Another important objecti

11、ve of the 5G technical specification process was to create a system capable of supporting the diverse needs of different industries and sectors. Accordingly, there is interest in understanding which vertical sectors may adopt 5G first and at scale. Heavy Reading asked operators (Fig 9) about the ver

12、ticals they think will be most attractive in the near term (12-18 months). Smart cities (37%), media and entertainment (36%), health (35%), automotive (30%), and manufacturing (28%) lead the pack. Of the top two, smart cities always perform well in Heavy Reading surveys. One reason for that may be t

13、hat this category incorporates many types of use cases and thus appeals to everyone. We tend to think that media and entertainment, in second, is the safest bet because 5G services often already bundle video, music, and gaming with customer plans. n Smart Cities 37% Media and Entertainment 36% Healt

14、h 35% Automotive 30% Manufacturing 28% Transportation and logistics 21% Energy therefore, the solutions will likely be found at the system level. Turning to RAN field testing, the survey asked if 5G RAN testing is a bigger challenge than LTE (Fig 13). A reasonable assumption, given 5G introduces ano

15、ther RAN layer, with significant flexibility in how it is configured and deployed, would be that it increases complexity and field test challenges. The survey results bear out this view. About a third (31%) said 5G RAN is “significantly more challenging” and over half (54%) said “somewhat more chall

16、enging.” This indicates that 5G RAN testing is considered a stiff challenge, but not an insurmountable problem. Looking more deeply into 5G RAN field testing shows that “inter-technology handover between LTE and 5G” is considered the greatest challenge by 57% of respondents (Fig 14), ahead of “valid

17、ation of low latency URLLC applications” at 39%. This is consistent with early experiences of network operation where 5G is deployed in NSA mode, with the control plane anchored on LTE and the user plane split between the LTE and 5G bearers on the downlink (and sometimes uplink). Inter- technology o

18、peration is inherent to NSA, and there are many examples of it being challenging to implement, monitor, and optimize. In time, as operators become more familiar with inter-technology operations, and as operators move to SA mode, this challenge may moderate and other challenges will come to the fore.

19、 Massive multiple input, multiple output (MIMO)/beamforming and URLLC testing are likely to rise up the list of concerns as these technologies are introduced more widely and scaled for the mass market. n 155G NETWORK 40%) or as containerized applications in VMs (42%). This is probably because operat

20、ors now have stable telco cloud platforms based on virtualized infrastructure. A minority (18%) indicated their company will go direct to a cloud-native model with containerized applications on bare metal. The core network controls user sessions, authentication, policy, and mobility. It also connect

21、s to external networks such as the internet, to cloud providers, and into enterprises. It is therefore critical to the 5G system architecture and the 5G service offer. The major story in 5G core is the introduction of SA mode alongside NSA. NSA uses a 4G core with a combination 4G/5G access and is i

22、n all commercial networks launched to date. SA uses a 5G core and is not yet commercial. Deployment and scaling of SA is a multi-year process that affects devices, RAN, transport, and telco cloud strategies. Author: Gabriel Brown, Principal Analyst, Mobile Networks we need a 5G core to capture the f

23、ull benefits of 5G.” Fig 15. To what extent do you agree or disagree with the following statement: It will be difficult to offer the full range of 5G services in NSA mode using a 4G core; we need a 5G core to capture the full benefits of 5G. (N=151) Fig 16. When do you expect to deploy 5G core and s

24、tandalone 5G? (N=152) Strongly agreeAgreeDisagreeStrongly disagreeNot sure Within 12 months (by the end of 2020) Within 24 months (by the end of 2021) Within 36 months (by the end of 2022) In 2023 or later 16% 29% 40% 15% 6% 59% 30% 1% 4% 175G NETWORK 2021 should not be read as a worldwide industry

25、timeline. Even in Europe, 2021 may be a little too aggressive. Most operators will use “two or three vendors” in their 5G core, according to the survey (Fig 17). With a score of 47%, this is the favored option by some distance against “multiple best-of- breed vendors” (26%) and “likely to use a sing

26、le vendor” (11%). Depending on definitions, the mobile core comprises half a dozen to a dozen different network functions. Selecting two or three vendors, each providing a few closely coupled functions, is common in 4G. It makes sense the same model would prevail in 5G core. Only 11% selecting “sing

27、le vendor” is low relative to what is common today, on a global basis, in 4G core. This could reflect the fact that the survey was primarily taken by larger operators in advanced markets that have the wherewithal and incentive to manage multi-vendor cores. There is also an interesting disparity betw

28、een U.S. and RoW: just 3% of U.S. respondents selected single vendor versus 23% for RoW. There is an expectation that the 5G core will be cloud native. Although “cloud native” is not formally defined, it typically means containerized applications (workloads) composed of microservices. These are some

29、times called cloud-native network functions (CNFs), as opposed to VNFs. The challenge is that the mobile core has extremely high uptime requirements and must be very stable before an operator will risk large-scale commercial deployment. Failures in the core are the major cause of mobile network outa

30、ges. Some believe that cloud-native platforms and applications are not yet mature or stable enough for commercial 5G core, even where vendors are ready to offer them. This question (Fig 18) tested that supposition. Likely to use single-vendor .11% Likely to use two or three vendors to assemble a 5G

31、core .47% Likely to use multiple vendors to create a best-of-breed 5G core .26% Dont know / too early to say .15% As VNFs in VMs.40% As containerized applications in VMs .42% Containerized applications direct to bare-metal with cloud native orchestrator (e.g. Kubernetes) .18% Fig 17. Thinking about

32、your 5G core network, do you plan to assemble the functions that make up the service- based architecture (SBA) 5G core from multiple vendors or from a single vendor? (N=150) Fig 18. On what technology platform do you expect to deploy your initial 5G core? (N=150) “MOST OPERATORS WILL USE TWO OR THRE

33、E VENDORS IN THEIR 5G CORE” 5G NETWORK offer an end-to-end testing solution from RAN through the edge to the mobile core; and stress test the impact of RAN traffic on the core network. Service providers need to prepare their networks for the demands of a massive number of connections through the Int

34、ernet of Things (IoT) while also coping with bandwidth-hungry, delay-sensitive edge applications such as augmented reality. 195G NETWORK less than 20% have “not started.” In all but one case (network virtualization), less than 20% claim their deployment as “mature.” The clustering of response around

35、 the middle options indicates the rollout of edge cloud is well underway in advanced operators, but far from finished. U.S. respondents were slightly more likely to select “mature” than their RoW counterparts. However, only in the categories of “containerized network functions” and “containerized wo

36、rkloads” was this significant to the tune of a 10% higher score. This may indicate U.S. operators are a little more advanced than peers elsewhere. Asked about the risks of running different types of workloads at the edge (Fig 22), most respondents view edge cloud as “moderately risky” for nearly all

37、 workloads, with security at the edge considered more severe. A third (35%) believe edge security presents “extreme risk.” Quality control and cost control are considered somewhat less risky aspects of edge cloud deployment, perhaps because respondents feel these are issues operators are already fam

38、iliar with when designing and deploying new infrastructure. One might argue that the inclusion of the word “risk” in the question led respondents to express greater concern about security than they might have otherwise. However, security is highlighted elsewhere in the survey as a primary concern. F

39、or example, “security” was identified as the biggest risk of working with external cloud providers to provide enterprise 5G services. The transition to virtualized and cloud- native telecom networks must consider the infrastructure on which workloads will run. This question (Fig 23) asked specifical

40、ly about the edge and the mix between containers and VMs in 2020 and 2023. The straightforward analysis of the response is that operators expect to support both VMs and containers now and in the medium term, which leads to the conclusion that edge cloud 5G Edge Cloud In an attempt to identify respon

41、dents primary motivation for edge cloud deployment in 5G networks, the first question in this section (Fig 20) allowed only one answer. The result shows there are several reasons, each with solid support and no clear winner. “Ensuring application performance” scores highest at 32%, followed by “vert

42、ical industry services” at 28%. These two options in combination account for 60%. The percentage of respondents that are more focused on operators internal priorities (21% “reduced transport cost” and 19% “differentiated communications services”) is lower at 40%, but not by a huge amount. Reduce ban

43、dwidth use/cost .21% Offer differentiated communications services (vs competitors) .19% Offer vertical industry services (e.g. in-vehicle scanning for ambulances, advanced real-time analytics for investment banking) .28% Ensure application performance .32% Fig 20. What is your PRIMARY motivation to

44、move workloads to the edge? (N=143) 215G NETWORK operated internally by the CSP Purchasing a pre- integrated slice solution developed by a System Integrator or vendor (capex model); operated internally by the CSP Purchasing a managed service slice solution from a single vendor or a System Integrator

45、 Purchase a slice solution provided using a SaaS model (subscription- based) from a cloud provider 378 372 345 295 1 3 2 4 Ranking Score (The score is calculated by assigning a weight to each rating where the highest priority rating holds the highest weight.) Fig 26. What procurement models do you s

46、ee as most important in achieving your business objectives for offering a slice-based enterprise solution? Please rank in order of importance (1 = Most important and 4 = Least important). (N=139) Yes, we have a plan to do this / already doing this .18% Probably this is likely in the medium term.40% Probably in the longer term .23% Not likely no plans at present .11% Dont know .