Improve Beaconless Spatial Acquisition for LCTs using Reinforcement Learning

Working for more than 20 years in the field of laser communication in space I am still surprised how many improvements remain to be discovered.

When I went for a jog and while my mind was idle all of the sudden there was this idea for a new spatial acquisition algorithm. Back home I sat down took the old compass and ruler and drew circles and lines. Tens of thousands of simulations using a spatial acquisition simulator developed by ST2C show that the new beaconless spatial acquisition algorithm:

• can tolerate certain hardware failures
• is faster
• is more reliable
• requires less complex hardware

than existing algorithms.

I was more than happy finding this new algorithm which is quite simple to implement. 

Then I started wondering how many easy solutions are out there which I haven’t discovered.

Twenty years I failed to see this little trick, what else did I miss? 

Now, this is no easy question to answer and luckily I don’t have to answer the question myself. Instead of the old compass and ruler Reinforcement Learning (RL) will do this for me during the next few weeks. In the mind map above (link for better resolution image) you can see that every Markov Decision Process (MDP) can be tackled with RL. Integrate a RL algorithm into a spatial acquisition simulator and see what the optimum policy is. Maybe I will discover a great policy like AlphaGo did for the GO game.

If you are interested in more details get in touch.

#lct #spacecom #lasercom #laserspacecom #spacedatahighway #edrs #machinelearning #reinforcementlearning