Method and system for free space optical communication performance prediction

Patent metadata

Field	Value
Publication number	US-11323177-B2
Application number	US-202017021289-A
Country	US
Kind code	B2
Filing date	Sep 15, 2020
Priority date	Sep 15, 2020
Publication date	May 3, 2022
Grant date	May 3, 2022

Abstract

Various embodiments provide a method for free space optical communication performance prediction method. The method includes: in a training stage, collecting a large number of data representing FSOC performance from external data sources and through simulation in five feature categories; dividing the collected data into training datasets and testing datasets to train a prediction model based on a deep neural network (DNN); evaluating a prediction error by a loss function and adjusting weights and biases of hidden layers of the DNN to minimize the prediction error; repeating training the prediction model until the prediction error is smaller than or equal to a pre-set threshold; in an application stage, receiving parameters entered by a user for an application scenario; retrieving and preparing real-time data from the external data sources for the application scenario; and generating near real-time FSOC performance prediction results based on the trained prediction model.

First claim

What is claimed is:

1. A free space optical communication (FSOC) performance prediction method, comprising:
in a training stage, collecting a large number of data representing FSOC performance from external data sources and through simulation in five feature categories;
dividing the collected data into training datasets and testing datasets to train a prediction model based on a deep neural network (DNN);
evaluating a prediction error by a loss function and adjusting weights and biases of hidden layers of the DNN to minimize the prediction error;
repeating training the prediction model until the prediction error is smaller than or equal to a pre-set threshold;
in an application stage, receiving parameters entered by a user for an application scenario;
retrieving and preparing real-time data from the external data sources for the application scenario; and
generating near real-time FSOC performance prediction results based on the trained prediction model.

2. The method according to claim 1 , wherein:
the five feature categories include propagation range, rain rate, visibility, cloud height, and cloud thickness.

3. The method according to claim 1 , wherein:
the DNN includes one input layer, a plurality of hidden layers, and one output layer;
the input layer includes five neurons or units; and
the output layer includes one neuron or unit.

4. The method according to claim 3 , wherein:
the plurality of hidden layers include five hidden layers;
a first hidden layer includes 128 neurons;
a second hidden layer includes 64 neurons;
a third hidden layer includes 64 neurons;
a fourth hidden layer includes 32 neurons; and
a fifth hidden layer includes 32 neurons.

5. The method according to claim 3 , wherein:
the plurality of hidden layers include nine hidden layers;
a first hidden layer includes 256 neurons;
a second hidden layer includes 128 neurons;
a third hidden layer includes 128 neurons;
a fourth hidden layer includes 64 neurons;
a fifth hidden layer includes 64 neurons;
a sixth hidden layer includes 64 neurons;
a seventh hidden layer includes 64 neurons;
an eighth hidden layer includes 32 neurons; and
a ninth hidden layer includes 32 neurons.

6. The method according to claim 3 , wherein:
a softplus activation function ln(1+e x ) is used in each hidden layer; and
a sigmoid function 1/(1+e −x ) is used in the output layer.

7. The method according to claim 1 , wherein:
the pre-set threshold is approximately 0.2%.

8. The method according to claim 1 , wherein:
the external data sources include terrain data from Cesium and weather data from Open Weather.

9. The method according to claim 1 , wherein:
collecting the FSOC performance data through simulation includes using a computer program called MODerate resolution atmospheric TRANsmission (MODTRAN) to simulate the FSOC to obtain the FSOC performance data.

10. The method according to claim 1 , wherein:
the loss function

Max_error
=

p
(

max
i

⁢

{



y
i
′

-

y
i



}

)

+

(

1
-
p

)

⁢

(

1
n

⁢

∑
i

⁢

(

y
i
′

-

y
i

)

)

,

where p is 0.5, and y′ and y denote the predicted and actual transmittance.

11. The method according to claim 1 , wherein:
the performance prediction result is generated and display on a user interface within approximately 2 seconds after the user launches a performance prediction task for an application scenario through the user interface.

12. A toolchain system for FSOC performance prediction, comprising:
a memory storing computer program instructions; and
a processor coupled to the memory and, when executing the computer program instructions, configured to perform:
in a training stage, collecting a large number of data representing FSOC performance from external data sources and through simulation in five feature categories;
dividing the collected data into training datasets and testing datasets to train a prediction model based on a deep neural network (DNN);
evaluating a prediction error by a loss function and adjusting weights and biases of hidden layers of the DNN to minimize the prediction error;
repeating training the prediction model until the prediction error is smaller than or equal to a pre-set threshold;
in an application stage, receiving parameters entered by a user for an application scenario;
retrieving and preparing real-time data from the external data sources for the application scenario; and
generating near real-time FSOC performance prediction results based on the trained prediction model.

13. The toolchain system according to claim 12 , wherein:
the computer program instructions include a service-oriented architecture to seamlessly integrate software tools distributed across Internet and are structured as a toolchain-based software-in-the-loop platform to capture real-time environmental data and to use data streaming implementation to continuously generate near real-time FSOC performance prediction results.

14. The toolchain system according to claim 12 , wherein:
the five feature categories include propagation range, rain rate, visibility, cloud height, and cloud thickness.

15. The toolchain system according to claim 12 , wherein:
the DNN includes one input layer, a plurality of hidden layers, and one output layer;
the input layer includes five neurons or units; and
the output

Assignees

• Intelligent Fusion Tech Inc

Inventors

• Li Lun
• Li Yi
• Wei Sixiao
• Shen Dan
• Chen Genshe

Classifications

• H04B10/112 Primary

  Line-of-sight transmission over an extended range · CPC title

• G06N3/048

  Activation functions · CPC title

• G06F18/217

  Validation; Performance evaluation; Active pattern learning techniques · CPC title

• G06F18/2431

  Multiple classes · CPC title

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title