Optimizing big data collected by sensors into real-time, clear and actionable intelligence to improve multidomain decision-making: This is the promise of AI at the edge, which, in the defense industry, is now more a requirement than a futuristic idea.
The need for AI at the edge
Traditional sensors remain crucial components to our current defense assets. They also collect vast amounts of data, known as big data, which needs processing and optimization to decipher what is useful. For example, 18 hours of data from a sensor on a field-deployed Mine-Resistant Ambush Protected vehicle, or MRAP, might yield just five minutes of actionable intelligence.
But to first identify that snippet of info as valuable, sensor data must undergo a time-consuming journey via satellite to an off-site physical server building for processing — and back again. Add up the number of sensors across today’s multidomain defense platforms and technologies — from ships to drones to satellites — and the complications quickly multiply.
AI at the edge provides a more efficient approach via the placement of processing power near or at the point of ingestion. In other words, AI at the edge integrates sensors with onboard central and graphic processing units (CPUs and GPUs), field programmable gate arrays (FPGAs) and advanced AI software, eliminating the need to transfer data via satellite for processing. This means that the MRAP just mentioned can instead now process sensor data using AI and deep learning software via an onboard, rugged server at the point of data ingestion, immediately optimizing big data into useful, real-time intelligence.
Because AI and deep learning software get better with more information, big data now also becomes an aid, not a hindrance. By integrating processing and AI learning power in the field, instead of a server warehouse, sensors can also receive new abilities via cloud-based pushing of containerized apps to each asset, without the need for on-site IT support.
Learn more about AI at the edge
AI and big data processing are necessary components in developing safe and smart defense systems and are required for the modern, real-time intelligence-gathering needs of today’s forces. Join our on-demand webinar to learn more from leaders at Mercury Systems and from global partners NVIDIA and SigmaX.ai, or read our white paper about how Mercury’s Intel-powered rugged servers are bringing AI power to everything, everywhere.