Space may be the final frontier, but it continues to pose myriad technical challenges as commercial and government-driven space investment continues. One of those challenges is developing more effective space-based communication systems for the increasing number of satellites and spacecrafts that need to interact with one another in the void. A team of researchers has developed an algorithm to enable cognitive radio functions on satellite communications systems to adapt themselves autonomously.

Current space communication systems deploy radio-resource selection algorithms, but they are rudimentary and work with a pre-programmed look-up table. Furthermore, they have little flexibility regarding the various parameters for the performance goals the system needs to achieve. Researchers from Worcester Polytechnic Institute, Pennsylvania State University and NASA’s John H. Glenn Research Center, have designed a new algorithm that allows autonomous, multi-dimensional parameter selection for radio resource allocation using a novel artificial intelligence architecture.

Autonomous space communication is critical because space is a harsh environment. A number of things can go wrong in space, which why space communications systems should be able to operate without human intervention. The team’s algorithms could serve as the core of a new cognitive engine (CE) used as a baseline for developing communication systems for the next generation of spacecraft and satellites.

“It solves a very complex communications system problem,” said lead author Paulo Victor Rodrigues Ferreira, Ph.D, of Worcester Polytechnic Institute. “Our algorithms provide a feasible solution for larger system optimization that would be unwieldy for today’s look-up table for adaptive communications.”

The team developed a CE design that autonomously selects multiple radio transmitter settings while attempting to achieve multiple conflicting goals in a dynamically changing communications channel.  It accomplishes this by leveraging reinforcement learning (RL) and “virtual exploration” structures studied in the author’s previous research. The CE integrates these with a novel artificial neural network ensemble design and new algorithms to implement the exploitation aspect of multi-objective reinforcement learning (MORL).

Through RL, the artificial neural network can be trained to adapt to the dynamic conditions of space through multiple trials and experiments, as the algorithm is set up to learn in a manner similar to the human brain by weighing inputs to achieve a goal. In the researchers’ CE, the system can learn how to adapt to achieve multiple goals for satellite communication.

The proof-of-concept design was created through computational simulations as well as ground- and spaced-based experiments. It successfully addresses the limitations of current technology by enabling:

  • Table-free state-action mapping with fixed memory size;
  • Operation over dynamically changing channels;
  • Decoupling of states from actions; and
  • Usage of continuous action and state spaces.

“Our CE is designed to optimize selection of multiple radio parameters dynamically with the flexibility of doing so while trying to achieve multiple goals at the same time,” said Dr. Alexander Wyglinski, Ph.D., SMIEEE. “In many cases these goals can even be conflicting, imposing a complex optimization problem for the current communications channel condition. We believe our CE is among the first of its kind to be proposed and evaluated in space.”

Simulation results for the CE design achieved 80% average accuracy in the six example mission profiles. Dr. Wyglinski expects the team’s work to serve as a starting point and benchmark for other researchers working in space communications.

Satellite Communications Artificial Neural Network

Figure 1: Boxplots of normalized packet count distribution over fitness score values obtained using the ANN architecture in six different missions.

The researchers plan to expand on their architecture by experimenting on NASA’s SCaN Testbed to optimize data transfer between radio nodes, data transport and cognitive decisions in the system. The SCaN Testbed is NASA’s test facility for radio communications, so the facility provides excellent conditions to test the system. These experiments will help the researchers get more insight into how to ensure consistent satellite communication function to deploy their system in the field.

As space exploration continues to develop with trips to the moon and Mars, cognitive radio and communication systems will be essential for space flight. There will be a need for space ’inter-networking’ to manage the interaction of user spacecrafts, relay spacecrafts and ground stations. This new CE could be a needed progression for space communication to be more efficient and reliable for any situation encountered during space travel.

For more information on space communication, visit the IEEE Xplore Digital Library.