Aug 30, 2018
Qualcomm products mentioned within this post are offered by Qualcomm Technologies, Inc. and/or its subsidiaries.
As many mobile technology enthusiasts have come to understand, 5G will serve as a unifying connectivity fabric — across spectrum, services and deployment models — with the ability to connect virtually everything. With its flexible design, 5G will be the innovation platform for tomorrow’s yet-to-be-created services while meeting our expanded connectivity needs for the next decade and beyond. As 5G proliferates globally, future innovations could happen virtually anywhere in the world — not just at today’s technology hubs. However, connectivity alone is not sufficient to realize 5G’s full potential — we need to bring new capabilities to the devices themselves.
There are already billions of connected devices and it is just a matter time until we reach a world with connections in the trillions. As these devices communicate and sense their surroundings, from a high-definition camera recognizing an object to a simple sensor measuring the temperature, the massive amount of generated data cannot be easily transmitted back to a central entity (i.e., the cloud) to be processed and managed.
To solve this massive scale challenge, as well as address privacy and security concerns, we need to shift to a decentralized model. In this new paradigm, intelligence is not just associated with a central cloud, it is distributed to the devices that form the Wireless Edge. This will require on-device capabilities for smartphones, cars, sensors, and other connected things, so they can perceive, reason, and act on their own, processing low entropy data and transmitting only the relevant content back to the cloud. The shift to on-device intelligence can bring in broader societal benefits ranging from improved driver safety, more personalized virtual assistants, superior photography, and enhanced security, to cameras with privacy, better connected health care, and more intuitive robotics.
Qualcomm Technologies already has essential technologies to transform the Wireless Edge — from low-power processing and sensing capabilities, to security solutions and connectivity — and realize the promise of running on-device Artificial Intelligence (AI) efficiently on a massive scale.
In fact, the proliferation of on-device AI has already begun with today’s connectivity fabric for the Wireless Edge — LTE Advanced Pro, Wi-Fi and Bluetooth — and its benefits include:
Our mobile ecosystem leadership allows us to deliver these power-efficient and highly integrated on-device solutions on a massive scale to the benefit of consumers, via devices powered by Qualcomm Snapdragon mobile platform. Today, our IoT solutions already combine our connectivity and on-device processing strengths; for example, the Qualcomm Vision Intelligence Platform leverages advanced AI for computer vision to help ensure only relevant information and notifications are sent from the camera to the cloud, instead of sending all of the unprocessed raw video stream.
With this distributed intelligence, we also need to move content, control, and augmented processing closer to the end-device, to what the tech industry collectively calls the edge cloud. This functionality could be located as close as possible to the devices and users to realize the low latency needed for some applications. Alternatively, it could be distributed deeper in the network, farther from the devices for latency tolerant services, depending on the use case, economics, and performance tradeoffs.
As mentioned earlier, the Wireless Edge transformation has already begun and society is already reaping many benefits today. However, as 5G grows and becomes a unifying connectivity fabric, new capabilities will emerge that further enhance services like ultra-high definition video streaming, transform industries like automotive and the industrial IoT, and create entirely new experiences with distributed functions over 5G, such as boundless extended reality (XR).
While a centralized cloud is important for big data training along with supporting latency-tolerant content and storage, AI execution will increasingly run in power-efficient devices at the Wireless Edge. As we move forward, we envision training to gradually emerge on devices, with applications such as learning for personalization, one of our key AI research areas.
The automobile, whether it’s autonomous or equipped with advanced driver assistance systems sensors, is a great example of why on-device intelligence is essential. In a vehicular environment, the car or driver must act in real time with data, as opposed to relying on the cloud.
Meanwhile, as 5G eMBB (Enhanced Mobile Broadband) provides more capacity for in-car infotainment and telematics, 5G C-V2X (cellular-vehicle to everything) is a beyond line-of-sight sensor that complements line-of-sight sensors such as radar and computer vision, allowing cars to communicate directly with each other and support latency-sensitive safety use cases, such as warning the driver of an approaching car or bicycle around a blind corner, or providing the driver with traffic light status. 5G C-V2X is also evolving for autonomous vehicle use cases — one example is the sharing of intentions with other vehicles, which results in more efficient and safer traffic flows. In my blog post earlier this year, I explained this use case in greater detail.
With 5G based Industrial IoT, it is possible to control critical equipment over an ultra-reliable, ~1 millisecond latency link. This requires on-premise control with a customized local network at the Wireless Edge — allowing greater equipment reconfigurability and flexibility to meet changing manufacturing demands. I recently explained how we can achieve this with 5G by leveraging multiple antennas that coordinate their transmission. This is just one example as the devices at the factory will require different forms of 5G connectivity that scale from low-complexity sensors to extreme mobile broadband for Industrial XR with information overlay for maintenance and repair on machinery.
Qualcomm Technologies is committed to continue delivering the best possible on-device capabilities in the 5G era, while exploring new possibilities to distribute functions over 5G to create enhanced or new services. Boundless XR is an example of this new class of services, where on-device capabilities are augmented by the edge cloud (compute, storage, rendering, content, and more) and connected with 5G.
Mobile XR provides the best immersive experience at any location — and we are delivering this today with on-device, low-power processing. If we can harness the power of the edge cloud with rendering over 5G’s high speed, low-latency, ultra-reliable link — with a completely different power envelope — we could augment the on-device processing to deliver a photo-realistic experience. It’s important to note that in order to avoid user discomfort, latency-bounded adjustments must be done on the device to meet the less-than-20 ms-latency requirements (e.g., the image projected in XR glasses must be in sync with the user’s actual movements).
2019 is going to be an exciting year as we enable commercial 5G NR smartphone launches around the world. In parallel, we are working on a broader 5G vision to evolve and expand the ecosystem and transform industries. To learn more, check out my webinar on the Wireless Edge transformation and download the presentation.