-
StatusOngoing
-
Status date2023-10-15
-
Activity Code6B.090
Optical Communication Systems (OCS) have successfully been used for high communication data rates, either between satellites or between satellites and optical ground stations. However, such activities are not well optimised, owing to the complexity of the system. To this end, this activity explores where Machine Learning (ML) can be applied to provide or enable such optimisation, through upgrading, improving, or simplifying OCS.
In ML4OC the focus is drawn onto improving the onboard laser communication terminals and the traffic routing in an optical network. These topics are studied in the view of ESA’s HydRON (High Throughput Optical Network) project to ensure network connectivity in orbit. To align with this, the project integrates demonstrations of ML models seeking to improve the link acquisition control loop and the acquisition time with a stretch goal targeting a ten-fold increase in traffic routing time.
The main activities and objectives of ML4OC project are:
-
Perform use case and requirement capture for ML-enabled OCS highlighting ML optimisation opportunities.
-
Develop and describe ML models along with test systems to be integrated into.
-
Demonstrate the benefits for the system by applying the ML solution instead of conventional methods.
The key technical challenge for this activity lies in the limited availability and suitability of datasets required for training ML algorithms targeting OCS use cases. This in turn poses challenges to the ability of the ML algorithms to show performance improvement over the baseline and to meet performance requirements. Autonomous components also must not compromise function in the case of anomalous events, so robust assurance methods are needed. Furthermore, there is difficulty in defining fully representative requirements to support the ML development due to limited detail on information as the HydRON is in initial phases.
The outputs of this activity are machine learning models which improve spatial acquisition and network management in space to ground optical communication systems. This benefits future HydRON and OCS SatCom missions by improving overall availability and efficiency of optical links.
For spatial acquisition optimisation, ML can be used to monitor weather conditions ahead of the satellite’s path. Network link performance and data rates can be negatively affected or even completely blocked by the weather conditions at the link position. The ML classification can feed into onboard planning such as changing the laser mode or switching to another ground station, meaning communications links can be more reliably established in the presence of outside effects. Additionally, ML techniques can be used to correct for camera sensor issues such as bad/hot pixels and optimise correct recognition of laser beacon signal.
Optimisation techniques can be applied to the network management by identifying the status of neighbouring nodes and using the output to drive local control logic to improve link acquisition times. ML can quickly determine the best path for data to be routed at higher network layers if the physical layers of the established route are suffering from degradation.
Supporting the spatial acquisition optimisation is image classification and cloud detection algorithms. These algorithms can be trained to report and classify upcoming atmospheric features, and then feed this into the onboard planning. Cloud cover can be classified in the range from clear skies wherein spatial acquisition proceeds as normal, to complete cover, wherein acquisition at the initial OGS is stopped, and alternate routes are found if possible. In between haze or intermittent cover, parameters such as laser modes can be altered to best achieve the link. ML algorithms can distinguish between different bright spots on the camera sensor during beam tracking meaning tracking can be maintained in the presence of noise. These algorithms also characterise camera performance and monitor conditions throughout camera lifetime.
For the network management, optimisation is implemented through the use of localised healing functions, wherein degradations or faults can be detected and recovered from by rerouting the traffic through nearby nodes. Fault detection is done by employing classification at each node to infer the status of neighbouring nodes. The neighbouring nodes then update their traffic flow tables to reroute the data until the performance of the original node is no longer degraded.
The system begins with rationale and detailed descriptions of the optimisation opportunities and methods chosen for use. Then the ML model is developed and described for the use cases given (spatial acquisition and network management). For each use case, this involves the creation of the algorithms and the training datasets, and training of the algorithms. Following that, the algorithms are tested, validated and verified using selected test scenarios. The results and performance of the algorithms are then be analysed and compared against the baseline and the traditional methods.
Following the kick-off meeting in February 2023, a requirements definition phase begins. The first milestone is then the Requirements Review, which sets the requirements for the remainder of the activity along with justification for proposed optimisation problems. The Mid-Term Review follows, covering description of the test system, breakdown of the ML implementation and documentation of verification and validation. The Design Review, covers analysis and results comparison of the solutions. Finally, the Final Review closes the activity, covering the final report and demonstrations which clearly outline the advantages of the ML solution.
Currently, the project has officially kicked off, with the initial managerial deliverables such as the risk register, webpage contents and executive summary completed. The project has moved into the beginning stages of the first Work Package to clearly develop and specify requirements. This stage includes a workshop with ESA to ensure accurate capture of HydRON requirements.
