3.2 Adjusting Evaluation Model Adaptively
In the performance evaluation of existing satellite communication system, the communication scenario is fixed in advance, and weights of evaluation parameters are set
for this communication scenario. For example, the transmission capability evaluation
typically assumes interference only comes from noise. However, in an actual satellite
communication scenario, the rain attenuation may have a serious impact on the
transmission capacity. As a result, the assessment results are difficult to match the
actual situation. It is necessary to modify the communication scenario and evaluation
parameters’ weights artificially. There are many kinds of satellite communication
scenarios, thus artificial modification requires a lot of manpower, material and financial
resources, and it is difficult to operate.
The neural network can adaptively adjust weights of the evaluation parameters
according to different evaluation scenarios. For example, a neural network model for
evaluating terrestrial mobile communication scenarios can be used to evaluate maritime
mobile communication scenarios. Although land mobile communication focuses on
signal attenuation caused by building occlusion, maritime mobile communication
focuses on rain attenuation, as long as there is actual communication data, neural
network can adaptively adjust the evaluation model through training data, and manual
adjustment is rare.
3.3 Comparing Different Satellite Constellations’ Performance
Different communication needs require different satellite constellations. For example,
more coverage needs more GEO satellites, less delay needs more LEO satellites. In
order to find out the optimal constellation to meet a certain communication need,
performance comparison of various constellation orbits becomes a must. Existing
methods set constellation orbits by experience, and performance comparison is done by
computer simulation. This method is greatly influenced by experts’ subjective judgment, and only covers a small number of orbits.
Neural network could classify or cluster constellation orbit parameters based on
satellite constellation simulation results when there is a communication need. When the
classifying or clustering model of constellation orbit parameters is established, performance comparison of most constellation orbits is feasible. Computer simulation can
further verify the correctness of the comparison, and later guide the improvement of
neural networks.
4 Conclusion
This work summarizes current research on performance evaluation of satellite communication system and proposes application prospect of neural network in performance
evaluation of satellite communication system. In the future work, a neural network
based on actual satellite communication scenario will be given to verify the feasibility
of performance evaluation.
Application of Neural Network in Performance …
243
In the performance evaluation of existing satellite communication system, the communication scenario is fixed in advance, and weights of evaluation parameters are set
for this communication scenario. For example, the transmission capability evaluation
typically assumes interference only comes from noise. However, in an actual satellite
communication scenario, the rain attenuation may have a serious impact on the
transmission capacity. As a result, the assessment results are difficult to match the
actual situation. It is necessary to modify the communication scenario and evaluation
parameters’ weights artificially. There are many kinds of satellite communication
scenarios, thus artificial modification requires a lot of manpower, material and financial
resources, and it is difficult to operate.
The neural network can adaptively adjust weights of the evaluation parameters
according to different evaluation scenarios. For example, a neural network model for
evaluating terrestrial mobile communication scenarios can be used to evaluate maritime
mobile communication scenarios. Although land mobile communication focuses on
signal attenuation caused by building occlusion, maritime mobile communication
focuses on rain attenuation, as long as there is actual communication data, neural
network can adaptively adjust the evaluation model through training data, and manual
adjustment is rare.
3.3 Comparing Different Satellite Constellations’ Performance
Different communication needs require different satellite constellations. For example,
more coverage needs more GEO satellites, less delay needs more LEO satellites. In
order to find out the optimal constellation to meet a certain communication need,
performance comparison of various constellation orbits becomes a must. Existing
methods set constellation orbits by experience, and performance comparison is done by
computer simulation. This method is greatly influenced by experts’ subjective judgment, and only covers a small number of orbits.
Neural network could classify or cluster constellation orbit parameters based on
satellite constellation simulation results when there is a communication need. When the
classifying or clustering model of constellation orbit parameters is established, performance comparison of most constellation orbits is feasible. Computer simulation can
further verify the correctness of the comparison, and later guide the improvement of
neural networks.
4 Conclusion
This work summarizes current research on performance evaluation of satellite communication system and proposes application prospect of neural network in performance
evaluation of satellite communication system. In the future work, a neural network
based on actual satellite communication scenario will be given to verify the feasibility
of performance evaluation.
Application of Neural Network in Performance …
243