By Dr. Fredrik Bruhn, Company Director of Unibap
Geostationary satellites are 36.000 kilometers away from space, but all of our everyday systems, such as television, navigation, communications, weather forecasts, and traffic monitoring, rely on geostationary satellites.
In traffic monitoring applications, satellite sensors capture image data and send it to the Earth Station for evaluation. In most cases, the response time is long enough and it doesn’t count every second, so this approach works smoothly. However, when evaluating data in near real time, for example in the case of a potential air traffic accident, the limits of what is possible are quickly reached.
The reason for the long transmission time is the source of the data. Color images captured in space often have a resolution of 25-130 megapixels in video mode, making it easy to generate 75 Mbytes of data in a single uncompressed image. At 30 frames / sec, this requires a transmission bandwidth of 18 Gbit / s.
The bandwidth of connections to nanosatellite orbiting the Earth at 400-600 km can reach Gbit / s, but is typically around 50 Mbit / s, so demand far outweighs performance. Larger satellites can generate terabytes of data per day. Ground stations for satellite communications are distributed throughout the globe, each with different priorities and adding delay to the data chain.
Space data latency resolution
The Swedish company Unibap is tackling the problem of data latency by installing a space cloud directly on its satellites. The solution, called SpaceCloud, uses artificial intelligence (AI) to determine relevant data locally. Instead of sending a large stream of raw data, it sends only the required data, such as analysis results and positioning commands, to the ground station. See Figure 1.
“Unibap’s Space Cloud provides a flexible and powerful infrastructure for artificial intelligence in space,” said Dr. Fredrik Bruhn, a former NASA guest researcher and now Unibap founder and CEO. “This platform integrates cloud services, intelligent data processing, sensor management, data storage, data analytics, as well as on-demand transmission of relevant information.”
There are several reference apps that show how cloud technology can be used in space, including advanced image and video compression, scientific data analysis, precision agriculture practices, and machine learning-based vehicle, ship, and aircraft recognition. I have. The development environment uses pre-analyzed satellite imagery. In principle, the development system is also suitable for any deep learning application.
Inibap’s Space Cloud made its maiden flight on D-Orbit’s IONWILD RIDE mission on June 30, 202. See Figures 2 and 3.
Despite the harsh environment, Unibap solutions use standard components including AMD’s Ryzen Embedded processor. To be eligible for orbital use, Unibap needed to prove that its solution was suitable for operation in the most harsh environments with the least possible power consumption. Acceleration and vibration, pressure, thermal fluctuations, and stresses caused by cosmic rays are so high that components specially developed for space applications were eligible only in the past.
Swedish engineers have tested it on standard processors from various manufacturers and found that AMD Embedded technology is inherently cosmic ray resistant. Unlike sunlight, cosmic rays are not electromagnetic radiation, but a flow of high-energy particles such as protons, electrons, and ionized atoms. Particles can create so-called single event effects (SEEs) on semiconductor components, causing interference, data loss, and transistor destruction via single event latchups (SELs).
For aerospace applications where data integrity is a top priority, all single calculations and all autonomous decisions must be error-free for reliability and safety. This makes it very important to protect the data stored in RAM from errors and ensure that CPU and GPU calculations strictly follow the code. Therefore, AMD SoC provides additional error correction memory. This is an important feature that helps correct in-memory data errors triggered by SEE.
Possibility of untapped
In addition to cosmic ray immunity, Unibap also chose AMD processor technology for another reason. It is a CPU and GPU integration and is very suitable for heterogeneous vision and AI computing applications.
The deep learning software ecosystem is also comprehensive and mature. Open standards such as OpenCV (Open Source Computer Vision) and OpenCL (Open Computing Language), as well as the AMD ROCm open software platform, provide a broad foundation for advanced computing with seamless CPU and GPU integration. ROCm is the first hyperscale open source platform for high speed computing and is programming language independent. It provides UNIX choice philosophy, minimalism, and modular software development for CPU computing. Application developers are free to choose tools and language runtimes, or develop their own, as Unibap did.
In addition, the ROCm ecosystem has a wide range of freely available frameworks (Tensorflow / PyTorch), libraries (MIOpen / Blas / RCCL), programming models (HIP), interconnection (OCD) and forward compatible Linux support. Includes technology. Kernel distribution – and the platform is always optimized for performance and scalability. Tools, guidance, and insights are openly shared in the ROCmGitHub community and related forums.
SpaceCloud uses several products from the AMD Embedded portfolio, including the AMD Ryzen Embedded V1000 processor family, which combines the performance of AMD Zen CPUs and Vega GPU architectures in SoCs. The rugged AMD Ryzen iTemp variant operates at temperatures as low as -40 ° C. AMD Secure Run technology encrypts data in main memory and provides encryption isolation from virtual machines, clients, and even the hypervisor itself, making it suitable for small-edge clouds in applications that integrate multi-vendor solutions. ..
Unibap is currently testing the next evolutionary step with the AMD Ryzen Embedded V2000 series. In this series, the number of cores and performance per watt are doubled compared to the previous generation. Radiation behavior is unique to each processing node and requires dedicated testing for each new product to verify its suitability for space and aerospace applications. For the V2000 series, AMD plans product availability for up to 10 years, providing space engineers with a long roadmap of support.
The intelligent vision system used in self-driving has the same system requirements as satellites, so the Unibap application in orbit is likely to work in self-driving cars, the company said.
https://www.electronicsworld.co.uk/cloud-computing-in-space/33721/ Cloud Computing in Space – Electronics World