1、Intelligent Connectivity Unleashing opportunities with the power of 5G, AI and cloud 10 December 2019 Intelligent Connectivity | Whitepaper Tom Morrod Research Executive Director Julian Watson Principal Analyst Shen Wang Senior Consultant Commissioned by OPPO IHS Markit | Title of Report Confidentia
2、l. 2019 IHS Markit. All rights reserved. 2 10 December 2019 Contents Table of Contents Preface 3 Intelligent connectivity 3 Defining intelligent connectivity 3 The building blocks of intelligent connectivity 5 The development of AI 7 AI in the cloud or at the edge 9 Global 5G development 10 5G defin
3、ed 10 Network slicing will come with 5G SA NR 11 Its a 5G world, already! 12 And China is driving it 13 The role of AI in 5G 13 The changing nature of human-machine interaction 14 The rise of the digital assistant 14 The enduring utility of smartphones 15 Unleashing opportunities with the power of 5
4、G, AI and cloud 16 Defining the opportunity 16 The opportunities and challenges of intelligent connectivity 17 Consumer opportunities: smartphones, digital assistants and entertainment 17 Video surveillance opportunities 18 Automotive opportunities 19 Factory opportunities 20 A note on 5G challenges
5、 21 Conclusion developing the ecosystem for intelligent connectivity 22 IHS Markit | Title of Report Confidential. 2019 IHS Markit. All rights reserved. 3 10 December 2019 Intelligent Connectivity Unleashing opportunities with the power of 5G, AI and cloud Julian Watson, Principal Analyst Preface In
6、telligent connectivity refers to: The ability of devices, buildings, computers and people to acquire knowledge and skills. Their capacity to take decisions and change their behavior based on the acquisition of this knowledge and skills. The role of connectivity and collaboration between devices, bui
7、ldings, computers and people in distributing this knowledge and skills. Intelligent connectivity is enabled through the collaboration of AI, connectivity (including 5G), cloud and edge and IoT. These ever-evolving technologies will work together to create new, immersive experiences for consumers, ad
8、dress challenges and create opportunities for enterprises and industry. For instance, 5G will enable much faster extraction of data from the cloud onto devices, reducing the need for storage. IoT will yield rich streams on data; AI will generate insights that can improve customer experience. The edg
9、e will deliver the rapid response required for applications that need to react to changes in the physical in real time. The diverse range of players in this space will need to think creatively and collaboratively to understand and predict the future needs of their customers. Their thinking will be g
10、uided less by the “push” of product and services, but more by the “pull” of consumer experiences and desired outcomes for enterprise and industry. This thinking will lead to changes in how products are designed and fuel entirely new business models. Intelligent connectivity Defining intelligent conn
11、ectivity The Oxford English Dictionary (OED), defines intelligence as “the ability to acquire and apply knowledge and skills”. Its three definitions of intelligence demonstrate that not only humans, but also devices, buildings and computers can possess such knowledge and skills: Having or showing in
12、telligence, especially of a high level. (Of a device or building) able to vary its state or action in response to varying situations and past experience. (Of a computer terminal) incorporating a microprocessor and having its own processing capability. Connectivity refers to: The state of being conne
13、cted or inter-connected. IHS Markit | Title of Report Confidential. 2019 IHS Markit. All rights reserved. 4 10 December 2019 (In computing) the capacity for the interconnection of platforms, systems, and applications. Therefore, the term intelligent connectivity implies: The ability of devices, buil
14、dings, computers and people to acquire knowledge and skills. Their capacity to take decisions and change their behavior based on the acquisition of this knowledge and skills. The role of connectivity and collaboration between devices, buildings, computers and people in distributing this knowledge an
15、d skills. There are varying degrees and types of intelligence. As children, humans start to develop intelligence through the experience of interacting with other children, adults and the physical world. For instance, a child touches a lightbulb and recoils in pain. He learns that the pain he feels o
16、n his hand is caused by the heat of the lightbulb. He is unlikely to touch a lightbulb again. To take another example, a smoke sensor in a building detects the presence of smoke. This automatically triggers an announcement over a public address system requesting office workers to vacate the building
17、. It also automatically triggers a phone call to the local fire department, who attend the scene. The office workers start to vacate the building. They do this because they understand from past experience that the presence of smoke could mean that there is a fire in the building that could cause the
18、m physical harm. In this example, the intelligence of a device (a smoke sensor) is distributing knowledge to office workers who respond to it accordingly. The fire officers who attended the scene locate the fire and put it out. They inform the building facilities director that the fire was caused by
19、 a boiler exploding due to overheating. A new boiler is bought and installed. This replacement boiler includes embedded sensors that monitor gas and water supply, temperature and pressure, and connectivity. This sensor data is communicated via wired or wireless technology from the boiler (now an IoT
20、 device) to a cloud-based portal, which the building director can access on her laptop or smartphone. Through the portal, she can remotely manage the operation of the boiler and how it transmits information. For instance, she could set the boiler to automatically close down when it reaches a certain
21、 temperature. She may also choose to receive an automated text message if the boiler exceeds a temperature threshold. The above scenario may reduce the likelihood of another fire in the building caused by the boiler overheating. But in this scenario the cause of the boiler overheating has not been e
22、stablished. Neither has likelihood of the boiler overheating been reduced. In an alternative scenario, the data from the embedded sensors is transmitted to a cloud-based analytical engine. This applies AI (artificial intelligence) to the data and learns what combination of factors are likely to caus
23、e the boiler to overheat and explode. This insight can be applied to automate the operation of the boiler, including the adjustment of settings and automatic shutdown. In this scenario, data flows from the device, the boiler, to the cloud-based analytical engine, and intelligence flows back from the
24、 cloud to the device. Across many industries, applications and use cases, there is a growing requirement intelligence to be applied as close as possible to source of data. In the future, an autonomous car will need to respond in real-time to changes in the physical environment. A physical obstacle m
25、ay appear in its path. The autonomous car will need to “see” the IHS Markit | Title of Report Confidential. 2019 IHS Markit. All rights reserved. 5 10 December 2019 physical obstacle, identify what it is (a cardboard box or a person) and react accordingly applying the breaks if it is a person. This
26、is an example of intelligence at the edge, commonly known as edge computing. As previously stated, the term intelligent connectivity implies not only the intelligence of devices, buildings, computers and people but also the role of connectivity and collaboration between devices, buildings, computers
27、 and people in distributing these knowledge and skills. Connectivity is the wired or wireless pipe through which data is moved and knowledge and intelligence exchanged between devices, buildings, computers and people. The wireless connectivity landscape is diverse, ranging from short-range and local
28、 area network technologies like 802.11 (Wi-Fi), Bluetooth Low Energy (BLE) and ZigBee to the cellular wide area network (WAN) technologies 2G, 3G and 4G LTE. The latter technology, 4G LTE, itself consists of several different interactions, serving applications that require deep range and a long batt
29、ery life, but limited bandwidth (e.g. NB-IoT) and applications that are bandwidth hungry (LTE-Advanced). The latest arrival to the crowded connectivity space is 5G. As we discuss later, 5G is distinct from previous generations of cellular technologies in that it has been designed at the outset to ad
30、dress many different technical requirements, device form factors, applications and audiences. The building blocks of intelligent connectivity In the previous section AI, the cloud and edge, connectivity (including 5G) and IoT were namechecked as enablers of intelligent connectivity. In this section,
31、 we discuss how these, and other building blocks will work together to create a new wave or consumer, enterprise and industrial applications and use cases. The diagram below (Synergies of transformative technologies) shows how different transformative technologies are collaborating to address the ev
32、olving requirements of people, enterprises and industry. The addition of embedded processing and sensors (IoT) to connectivity has yielded rich seams of data on the status, location and condition of connected nodes or their surrounding environment. The cloud is addressing the requirement to store an
33、d apply analytics to such large volumes of data. AI techniques are helping to manage this data and generate useful business insights from it. IHS Markit | Title of Report Confidential. 2019 IHS Markit. All rights reserved. 6 10 December 2019 The following schematic (AI diversification from Edge to C
34、loud) show that intelligence in the form of AI can be applied across the cloud (datacenters), the edge network of industrial PCs, gateways and edge servers and edge end-points - the diverse universe of IoT nodes. In this schematic, edge end-points, such as sensors that measure attributes like temper
35、ature and humidity, is where data is first generated and elaborated. Network edge/fog infrastructure gathers the data from edge end-points and elaborates it further before transmitting data onwards to the cloud for storage and further analytics. The parameters shown on the left (privacy and security
36、, latency, energy consumption, processor power and cost of data communication) will shape the development requirements for AI solutions: namely the application (inference) and training infrastructure required. Energy consumption is a particular issue for battery-powered edge end-point devices. Perfo
37、rming data-intensive analytical workloads such as training could reduce device battery life and ultimately negate the original investment case behind the IoT/AI application. IHS Markit | Title of Report Confidential. 2019 IHS Markit. All rights reserved. 7 10 December 2019 The development of AI AI g
38、enerically refers to the body of science that studies how to enable machines to perform independent problem solving, inference, learning, knowledge representation, and decision making. An intelligent machine does not need necessarily to show all these skills, but it must comply with one or any combi
39、nation of them. More specifically, AI can fulfill four major skills: perception, learning, abstraction, and reasoning There are several forms of AI. When machines are required to show specific self-learning skills, IHS Markit implicitly refers to a subset of AI called “machine learning” (ML). In thi
40、s extent, ML is a set of algorithms that gives machines the ability to automatically find and learn patterns by feeding them with data, withoutly explicit programming them. In the basket of ML fall several technologies and techniques to which neural network (NN) and deep learning (DL) also belong to
41、. NN and DL refer to computational models that try to emulate the structure and workings of a human brain, including process phases like training and inference. IHS Markit | Title of Report Confidential. 2019 IHS Markit. All rights reserved. 8 10 December 2019 The “training” phase see large volumes
42、of data fed into a computers NN brain. The data may relate to sounds, images or some other type of information. The computer learns to recognize specific objects and see patterns and discover how to identify objects in a variety of situations. The resulting trained model is then applied at the inter
43、ference phase to provide output (e.g. recognizing a specific image) on-real time input. As cited before, in human analogy, a child learns that touching a lightbulb will cause pain (training), so he avoids touching the next lightbulb he sees (inference). IHS Markit notes many industry verticals are a
44、ctively investigating the potential application of AI across a plurality of use cases, such as prescriptive 1 millisecond end-to-end round-trip delay (latency); preception of 99.999% availability or 10-5 packet loss rate and up to 10-year battery life for low-power, machine-type devices. 3GPP Releas
45、e 15 introduced the first set of 5G standards. 5G non-standalone (NSA) NR (New radio) specifications were completed in December 2017, six months ahead of schedule, due to a strong push from various stakeholders that wished to deploy 5G as soon as possible. 5G standalone (SA) NR specifications, also
46、part of 3GPP Release 15, were completed in June 2018. Todays commercial standards-based 5G networks are based on 5G NSA NR, which leverages existing LTE networks, and primarily serve eMBB use cases. Work items addressing Massive IoT and MCS requirements and a host of other features, such as multicas
47、t/broadcast, positioning and C-V2X (Cellular V2X) will come in Releases 16 and 17, which are currently set for completion in 2020 and 2021 respectively. This standardization work will gradually facilitate the expansion of the 5G ecosystem of chipsets, modules and devices and pave the way for industr
48、y verticals to develop proof of concepts (POCs). Network slicing will come with 5G SA NR IHS Markits Evolution from 4G to 5G Service Provider Survey 2019 suggests that in 2020 around a third of 18 leading service providers (accounting for half of the worlds telecom capex and revenue) will move to NG
49、 SA NR. This move will pave the way for 5G network slicing. IHS Markit | Title of Report Confidential. 2019 IHS Markit. All rights reserved. 12 10 December 2019 IHS Markit defines network slicing as “and end-to-end network slice is a logical partition or a virtual piece of a physical network, including fixed and mobile, physical and virtual, that has connections, capacities (bandwidth), and characteristics (e.g. Service Level Agreements or SLAs) that make the slice look and act
sgpjbg002
工作日 8:30 - 17:30
会员动态：
报告分类
新质生产力
ChatGPT
新能源汽车
大模型
Sora
机器人
微短剧
芯片
薪酬
白皮书
低空经济
数据要素
创新药
人工智能
AI产业
信息科技
互联网
消费经济
汽车交通
电商零售
传媒娱乐
医药健康
投资金融
能源环境
地产开发
传统产业
全球化
其它
扫码登录
手机快捷登录
账号登录
