New materials, quantum computing will make edge computing dominant
A pair of researchers with the Oak Ridge National Laboratory, one of the U.S. Department of Energy’s premier, big-picture research facilities, write in a new paper that using new nanomaterials will make edge computing as momentous as was cloud computing originally. These materials, along with neuromorphic computing methods could enable “nano-edge computing devices.”
Edge computing is a strategic play for companies because it represents a new way of turning data into competitive information. But aside from the increasing use of artificial intelligence, it is a new combination of existing technology — computer networking and data storage.
Indeed, edge computing typically is viewed as a sort of halfway house of corporate computing, churning away close enough to data centers to cut latency, and gradually developing the raw power and analytics of the cloud, although on smaller scales.
Nanomaterials now in development, along with advancing artificial intelligence, are the only way that edge can be a complete, energy-efficient, ultrahigh performance and secure network of billions of devices, according to Ali Passian and Neena Imam, the report’s authors.
Edge systems are already being overwhelmed by input from devices including cheap low-power sensors. Latency, a prime motivator behind the development of edge systems, is already creeping up, and 5G will not by itself solve that problem.
A good next step for edge computing would be ditching materials used in computing today that are inefficient. The pair say that almost one-third of non-man made materials can carry electricity and light without resistance and backscattering. These so called topological materials can reduce energy needs and, in the case of electricity, reduce waste heat.
| “Silicon-based transistors, developed to be increasingly smaller, experience conductivity losses, leading to energy loss by generating heat. Replacing silicon-based elements with carbon nanotubes, owing to their more efficient electron transport properties, can lead to less energy requirements.” |
Then there are carbon nanotubes, graphene and molybdenum disulfide, all nanoscale materials. They could be used to replace conventional transistors, leading to more efficient and speedier microchips and sensors.
More advanced artificial intelligence on the way will give edge systems the capacity to managed myriad devices while also controlling data flow.
Quantum networks and Quantum computing
Quantum physics, too, can play a role in a more vital edge computing environment, according to the researchers.
Logic-defying quantum effects, although exceedingly difficult to produce and control, can compute astronomical amounts of data, transmit data faster than the speed of light, store data and perhaps make network traffic impossible to be secretly hijacked.
In quantum computers, instructions can be both 0 and 1 simultaneously, calculating faster than ever thought possible for certain tasks. In fact, these devices could be used to create edge-computing networks that operate more efficiently under extreme traffic loads.
Scientists have shown that quantum computing used as part of the network could move information rapidly, indeed. Quantum effects have been employed to transmit data instantly up to about 870 miles, an effect that would have an immense impact on the ability to process data on devices at ever more remote edge nodes. Indeed, nodes would be able to operate at a peer-to-peer level for quantum applications.
Nanosystems meld with edge computing
If quantum computing isn’t mind-bending enough, consider the combination of neuromorphic computing (brain-inspired computing) and nanomaterials. The report suggests that it nanosystems and edge computing “May amalgamize to become an inseparable entity, where device and function interact dynamically.”
The report suggests that as sensing at the atomic and molecular levels becomes possible, a new era of nano-IoT could be upon us. “Molecular networks of billions of sensors already occur in biological systems, and this may be mimicked by nano-EC devices,” according to the authors.
Deep learning chips to grow 40% in Asia-Pacific by 2027
Article Topics
edge computing | edge hardware | IoT | nanosystems | quantum computing
Comments