A New White Paper On Solving the Processor Scaling Problem for Space
Avalanche Technology, the leader in next generation MRAM technology, released a white paper showing how AI computing should be implemented in Space. The white paper is titled “AI Computing in Space: A New Space-IoT Architecture to Solve the Processor Scaling Problem for Space”, and examines why we need a new approach to scaling processing and cloud architectures for Space.
Recommended:Â CIO Influence Interview with Herb Kelsey, Federal CTO at Dell Technologies
The Space Race has become our Industry 5.0. After decades of tepid focus on Space-based technologies, investment has skyrocketed to feed an insatiable need for ubiquitous broadband communications, high accuracy navigation and weather prediction, scientific study, improved reconnaissance measures, and an existential urge to create habitable spaces beyond Earth.
Dramatic increases in satellite deployments into constellations, combined with more sensors per satellite and higher resolution per sensor, have created celestial oceans of data needing storage and analysis. For satellites and other spacecraft to make split-second decisions autonomously in real time such as for defense-related applications, the highest levels of artificial intelligence—situational awareness—must be self-contained in orbit and not rely on ground-based data analysis.
Elements of the paper include learning from novel historical approaches, and how a new Space-Internet of Things architecture, available now, solves three vexing limitations of Space:
- Unreliable power
- Inability to store an infinite amount of data from IoT satellite nodes; and
- Crippling radiation
Unlike terrestrial high performance processing frameworks where the answer has been to throw more GOPS and TFLOPS at the problem, the unique conditions of space tell us that we need a new approach: a more robust but streamlined, SWaP-optimized hardware architecture to support enduring mission needs. This white paper examines:
- why legacy approaches to scaling processing and cloud architectures won’t work for space;
- what we can learn from past approaches; and
- how we can offer a viable platform solution despite certain immovable challenges in Space.
Read More :Â Kubecost Introduces Kubecost 2.0 for Advanced K8 Cost Management
“This new white paper continues Avalanche’s focus on solving Space-specific challenges,” says Danny Sabour, Avalanche VP of Marketing and Business Development. “Previously Avalanche showed how to enable Multi Mission Boot Capability, and how to design robust data centers in Space. Now we’re completing the Orbital Internet picture by providing a new architecture, available now, for how AI computing should be implemented in Space.”
The white paper details how the new Space-IoT big data processing architecture is scalable, as it can take us from Earth to the Moon and then Mars, and leverages existing technology so it can be implemented in time to support the planned 2028 NASA Moon Gateway launch.
Avalanche introduces new thinking on data duality, such as “Pseudo Static” data that is key to providing scalable high performance processing in Space. The white paper also highlights the importance of product development partnerships, such as with Blue Halo, Trusted Semiconductor, and Mercury Systems. The Avalanche Space-IoT architecture can be optimized in future iterations to reduce power and increase parallel processing through subsequent processor evolution, all while maintaining the same software framework, especially critical for parallel processing.
Recommended:Â 10 Data Center Trends to Watch in 2024
[To share your insights with us as part of editorial or sponsored content, please write to sghosh@martechseries.com]
