Recently, Qualcomm announced the launch of the world's first fully integrated 5G new air interface millimeter wave and below 6GHz RF module for smartphones, providing support for large-scale commercial use.
This is an important milestone in the mobile communications industry. But for users, the most concerned about when can use 5G mobile phones? Therefore, Sina Technology recently interviewed many Qualcomm executives.
Push 5G mobile phone in the first half of next year
"5G is regarded as a unified connection architecture, which will become a grid-like presence anytime, anywhere." In the view of Ma Dejia, senior vice president of engineering and technology at Qualcomm, 5G will bring about drastic changes in human life, such as ultra-low latency communication. , more smart devices and higher reliability of the terminal.
Ma Dejia, Senior Vice President of Engineering, Qualcomm
Ma Dejia gave two points in time: In the second half of this year, 5G will be pre-commercially large-scale; next year, 5G will have more milestones. He said that the current 5G first-stage standard has been completed, including There are two types of non-independent networking (NSA) and independent networking (SA). Among them, independent networking is a more important standard system in China.
Ma Dejia further explained that NSA means that a device can be connected to both 4G and 5G networks, relying on 4G network and core network work. This deployment method is easier and faster, because 4G network already exists, only need 5G is added to it for joint networking, so this is a way to quickly implement 5G. Many countries tend to choose this networking method. At the same time, SA is also a good networking method, which is characterized by not relying on 4G network, its core network is a brand new 5G network, under the SA architecture, 5G's advantages in speed and other aspects will be better played.
"We believe that the NSA will be deployed on a large scale in 2019. But China's interest in SA is higher, and we don't know the final situation," Ma Dejia said.
At the MWC Shanghai Exhibition in June this year, Huawei’s rotating chairman Xu Zhijun said in a public speech that Huawei will launch a 5G mobile phone based on the Kirin 5G wafer in June 2019.
"We expect 5G mobile phones to be launched in the first half of next year," said Mike Roberts, senior director of global product marketing at Qualcomm. The time-to-market for 5G mobile phones depends on the situation in different regions. Currently, countries and markets are catching up with each other. First, but the United States and South Korea will be the first to launch. "I want to bring 5G equipment to the market as soon as possible, which requires infrastructure, operators, OEMs and other aspects to be ready."
According to public data, as early as September 2016, Qualcomm released the world's first 5G-enabled modem, named Xiaolong X50. In March 2017, Qualcomm announced that the Xiaolong X50 can achieve full-frequency coverage. Achieving multi-mode functionality for 2G/3G/4G /5G; In October 2017, Qualcomm announced in Hong Kong that the single-chip-based Opteron X50 5G modem has achieved the world's first 5G network data connection; Based on the Xiaolong X50's 5G smartphone reference design, this means that the 5G network is supported technically. It is reported that there are currently 20 equipment manufacturers and 18 operators around the world planning to use Qualcomm X50 baseband wafers.
Four factors drive AI development
Compared with 5G, Qualcomm's artificial intelligence strategy also attracted attention.
It is predicted that by 2022, the cumulative shipments of smartphones will exceed 8.6 billion. Mobile phones will become the most popular artificial intelligence platform due to the wide range of other types of terminals.
In the eyes of Qualcomm product management director Gary Brotman, AI is accelerating, and 5G guarantees efficient network connectivity, which means that future AI processing will not only be done on a single device in the cloud, but in a common collaboration. Processing and computing to achieve sharing intelligence across devices.
He believes that terminal-side AI is very important. The first is the privacy of data. For example, some users like to share personal information on social media, but at the same time should be better protected, and the security on the terminal side will be better; It is low latency, such as face recognition. The response of these services should be real-time. The third is reliability. If all operations and processing are done through the cloud, there must be multiple links, and reliability will be Dropped.
Gary Brotman said that Qualcomm launched its first personal intelligence project in 2007, studying deep learning, neural network algorithms, especially hardware processing, terminal-based artificial intelligence algorithms, etc. In 2015, Snapdragon 820 was introduced to the market. The mobile platform, the first generation of mobile artificial intelligence platform, is also the first SoC that is optimized to run the terminal side AI algorithm.
Qualcomm believes that in the AI ecosystem, hardware is the starting point and the core. From the hardware, it supports the support of a large number of machine learning frameworks in the market, and also supports different operating systems. In addition, it also works with many cloud service providers and independent software developers. The business has established a partnership to drive the market's demand for hardware.
"We believe that multiple factors will drive the future of AI," said Gary Brotman. The first is that 5G, 5G and AI will bring many new opportunities to extend intelligence to the edge of wireless networks. Distributed AI can There are many new applications. The second is design experience and expertise, especially in the development of highly integrated and powerful computing platforms. Third, the terminal-side AI that was initially invested in R&D has certain competitive advantages. Fourth, it has a broad base of cooperation. .
"Qualcomm's strategy has always been to perform AI operations through a multi-core heterogeneous computing architecture," says Gary Brotman, but that doesn't mean that Qualcomm will not consider developing dedicated AI hardware in the future, and will deploy it as it evolves.