Sep 21, 2021
Qualcomm products mentioned within this post are offered by Qualcomm Technologies, Inc. and/or its subsidiaries.
We envision a world where virtually all devices are much more intelligent, simplifying and enriching our daily lives. Today, artificial intelligence (AI) is delivering enhanced experiences and new capabilities to our society in more ways than ever. AI not only provides the ability for our devices to perceive, reason, and act intuitively, but also changes how we approach and solve technical challenges. For more than a decade, we have been conducting foundational AI research to make AI ubiquitous — we have taken a system-level approach that optimizes power efficiency across hardware, software, and algorithms. This is exemplified by our leadership in delivering powerful performance at ultra-low power consumption across a wide variety of devices through our Qualcomm AI Engine, now on its 6th generation.
At the same time, our foundational research in 5G is making it possible to build a cellular system that can efficiently connect virtually everything around us. Today, the global transition to 5G is now in full swing: there are over 175 mobile operators that have launched commercial 5G services globally and many more are actively investing in new 5G deployments. We are driving the continued technology evolution into the next phase — 5G Advanced, which starts with 3GPP Release 18, and it is a set of future standards that are expected to elevate 5G capabilities to new levels as well as expand 5G into new devices, deployments, and industries.
5G+AI = Connected Intelligent Edge
5G and AI are two essential ingredients that fuel future innovations, and they are inherently synergistic — AI advancements can help improve 5G system performance and efficiency, while the proliferation of 5G connected devices can drive distributed intelligence with continued enhancements in AI learning and inference. With the role of on-device intelligence becoming ever more important, the transformation of the connected intelligent edge has begun, and it is the key to realize the full potential of our 5G future.
To efficiently scale intelligence, AI processing needs to take place closer to the end users, on devices like a smartphone, a car, a laptop, or XR glasses. By processing data at the edge, we can realize better system efficiency, enhanced privacy, improved performance, and new levels of personalization. We envision the edge cloud, connected by a high-bandwidth, low-latency 5G connection, to play an increasingly important and complementary role to on-device AI, further augmenting the system’s processing capability and delivering new capabilities.
For the remainder of this blog post, I’ll focus on how AI or machine learning is solving key wireless challenges, why it is an important tool that can accelerate 5G advancements, and what role Qualcomm Technologies is undertaking to make wireless machine learning a reality.
Solving difficult wireless challenges
As we push the technology boundaries of wireless communications, we are encountering increasingly challenging problems that need to be solved before we can realize new levels of performance and efficiency. For instance, there are many hard-to-model problems that cannot be efficiently solved by a traditional model-driven design, as well as various system optimizations (e.g., modem parameters) required to achieve the best possible system performance, especially those that involve non-linearity. A data-driven design powered by machine learning can efficiently solve these problems. At Qualcomm Technologies, we are uniquely positioned to take on this new approach, as we possess deep domain knowledge in both wireless and machine learning, which is required to optimally apply AI to wireless.
In addition to applying AI in the core and RAN (radio access network) to enable intelligent network operation (e.g., enhanced QoS, better efficiency, simplified deployment, and improved security), on-device AI can also bring benefits to the overall 5G system. The underlying enabling capability is radio awareness, which provides useful knowledge through environmental and contextual sensing that can reduce overheads and latency. Through radio awareness, the 5G system can support:
- Enhanced device experiences like more intelligent beam forming and power management
- Improved system performance like reduced interference and better spectrum utilization
- Better radio security like better detection and protection against malicious attacks
Standardizing machine learning for 5G
We are working closely with the wireless ecosystem to introduce novel machine learning techniques to accelerate 5G advancements across a wide range of technology areas. Today, we are seeing wireless machine learning being studied across various standards and industry organizations, including 3GPP, ITU, O-RAN, GSMA and NGMN. In 3GPP, where 5G specifications are developed, a new framework (e.g., a new Network Data Analytics Function — NWDAF) has been introduced as part of Release 16 to facilitate wireless machine learning in core network. Release 17 includes a collection of new features that further augments the 5G system’s machine learning capability in RAN. For the upcoming 5G Advanced starting in Release 18, we expect machine learning features to be a key innovation vector that can expand to new and enhanced use cases, such as physical layer protocols.
A transition to a data-driven system design means enhancements will no longer be only possible with a new standard release, which can be a lengthy 1.5-year interval between releases in the current 3GPP work cadence. With an AI-native air interface design expected in the future, the system will be able to support continual improvements through self-learning, where both sides of the air interface — the network and device — can adapt to their surroundings dynamically and optimize operations based on what they experience. This is a fundamental shift in the way wireless systems can improve, and we envision this AI-native design methodology to become part of the future 6G system.
Showcasing 5G advancements using machine learning
At Mobile World Congress Barcelona this year, we showcased some of our latest 5G technology innovations across multiple disciplines. One common thread in many of our demonstrations is the use of machine learning techniques to improve the overall 5G system. Our demos showcase how AI can benefit performance and efficiency of 5G operating in sub-7 GHz and mmWave across different use cases. Watch our demos that use machine learning to enhance:
- Massive MIMO channel state feedback for increased user throughput and system capacity
- Mobile mmWave beam prediction for higher capacity and extended battery life
- Positioning accuracy by combining 5G measurements, GNSS, multi-path profiles, sensor inputs
- Network planning for efficiently expanding mmWave coverage with diverse node types
- AI-assisted RF sensing for indoor positioning using Wi-Fi today, also applicable in the future for 5G
To hear how 5G and AI are two synergistic, essential ingredients that are fueling future innovations, sign up for our LightReading webinar.