- NASA is planning to infuse AI into space communication networks.
- It uses Cognitive Radio to employ underutilized parts of the electromagnetic spectrum.
- SCaN Testbed provides tools and resources to test Cognitive Radio in space.
NASA engineers are planning to use Artificial Intelligence (AI) and Machine Learning (ML) techniques to better manage the growing communication networks between its spacecraft and Earth ground station. The spacecraft used by the agency generally rely on human-controlled radio signals to send signal back on Earth.
As overall space data increases, NASA explores cognitive radio – the integration of artificial intelligence into space communication networks, to meet demands and increase efficiency.
Modern space communication systems are based on complex software (as well as hardware) to fulfill the requirements of near-Earth orbit and deep space missions. Using AI and ML would enable satellites to control these systems seamlessly, and instead of waiting for instructions, they can make real time decisions by themselves.
The Cognitive Radio
You might be aware of the FCC (Federal Communication Commission) in the United States – it is responsible for allocating different sections of the electromagnetic spectrum to various users/organizations. For instance, the FCC provides a specific spectrum to satellite radio, cell service, WiFi or Bluetooth.
What if there is no electromagnetic spectrum is left? How could an instrument access any of the electromagnetic spectrum when all of them are already taken.
The cognitive radio (software-based radio) uses AI to employ underutilized section of the electromagnetic spectrum without human intervention. These are empty portions that are currently unused and are already licensed to be used for specific purposes. The FCC allows a cognitive radio to use the frequency that is not currently used by its primary user. When the user becomes active again, it assigns the other unused frequency.
What’s the Benefit of Using Cognitive Radio
The recent advancements in cognitive technologies have helped a lot in improving the levels of communication systems. Integrating cognitive radios and AI into NASA’s network would enhance the performance, autonomy, efficiency and reliability of space communication system.
The harsh environment of space involves some unique challenges that cognitive radio may handle. Electromagnetic radiations of the Sun and other bodies, and weather conditions disturb a specific range of frequencies.
Scientists at Glenn Research Center are currently working on cognitive radio apps capable of detecting and adapting to space environment. The machine learning techniques would be used to transmit outside the interference range or eliminate distortion.
Moreover, the cognitive radio would be able to turn off itself temporarily to deal with radiation damage during critical space weather events. This would make the system even more efficient for exploring the depths of space.
It could also present alternative data links to the ground, prioritizing and routing data via multiple paths concurrently to avoid interference. The AI of cognitive radio could allocate ground station downlinks in hours, instead of weeks, leading to more effective scheduling.
It can also make networks more efficient by reducing the need of human involvement. That means, the cognitive radio can adapt to new electromagnetic landscapes and predict basic operational setting for different conditions, which were previously handled by humans.
The SCaN (Space Communication and Navigation) Testbed provides scientists on ISS (International Space Station) with tools and resources to test cognitive radio. It consists 3 software based radio and different types of antennas and instrument, which can be remotely configured from the Earth or other spacecraft.
SCaN | Credit: NASA
The SCaN System offers strategic and programmatic supervision for the development of communication infrastructure. It is primarily built to enhance connectivity from spacecraft to Earth. While it could be simulated on the ground, there’s is an unpredictable element on space. The Testbed offers this environment a setting for technology advancements such as cognitive radio.
It will be much more difficult to implement cognitive technologies in space because of electromagnetic environment, orbital mechanics and integration with old instruments. Despite of these challenges, infusing AI and ML into the existing infrastructure will enhance the performance and reliability of space networks.
This news comes after Google claimed supercomputers capable of outsmarting humans might be just 3 years away, and they would cost as little as $1,000. The director of engineering at Google, Ray Kurzweil said that we are close to a general agreement of hardware requirement of strong AI.
Using Artificial Intelligence In Space
The objective of AI is to be more like smart assistant collaborating with human rather than just be a programmed assembly code.
In the past, NASA Jet Propulsion Laboratory has led the development of numerous examples for space AI. For instance, Opportunity rover used a program called WATCH to capture the images of dust across the Martian surface. WATCH later evolved into AEGIS that helped Curiosity rover’s ChemCam device pick new laser target without any human intervention. AEGIS can also set the pointing of the ChemCam laser.
An AI program named Autonomous Sciencecraft Experiment analyzed floods, fires and volcanoes on Earth. Also, EO-1’s Hyperion instrument used AI to detect sulfur present on the surface of glaciers.
In short, AI allows spacecraft to prioritize the data it gathers, managing other needs such as limited data storage and power supply. Autonomous management system like these is being prototyped for Mars 2020 rover.
Scientists suggest that AI enabled probes could reach as far as Alpha Centauri, closest star to the solar system (4.24 light years away from Earth). Giving a probe a mind of its own would significantly speed up the decision-making process.
While autonomy offers interesting new advantages to science teams, artificial intelligence has a long way to go.