Data transmission network configuration
A method and devices for configuring a data transmission network are disclosed. The method is for configuring a data transmission network, executed by a configuration device, wherein the data transmission network comprises at least one transmitter, at least one receiver with a communication channel between the transmitter and the receiver, the method comprising: training a machine learning model of the data transmission network, wherein the machine learning model comprises at least a transmitter model including a transmitter neural network, a channel model, and a receiver model including a receiver neural network by providing a message within a sequence of messages; generating a group of transmission symbols for each message in the sequence of messages using the transmitter neural network; concatenating the groups of transmission symbols together as a sequence of transmission symbols; simulating transmission of the sequence of transmission symbols over the communication channel using the channel model to the receiver; analysing a sequence of received symbols using the reception neural network to generate a decoded message; and updating the machine learning model based on an output of said reception neural network. In this way, the machine learning model can be trained using representative sequences of message, which improves performance when deployed in a real network.
1. A method for configuring a data transmission network, executed by a configuration device, wherein said data transmission network comprises a transmitter, a receiver, and a communication channel between said transmitter and said receiver, said method comprising:
training a machine learning model of said data transmission network, wherein said machine learning model comprises at least a transmitter model including a transmitter neural network, a channel model, and a receiver model including a receiver neural network, wherein said training comprises:
providing a message within a sequence of messages;
generating a group of transmission symbols for each message in said sequence of messages using said transmitter neural network;
concatenating said groups of transmission symbols together as a sequence of transmission symbols;
simulating transmission of said sequence of transmission symbols over said communication channel using said channel model to said receiver;
analysing a sequence of received symbols using said receiver neural network to generate a decoded message; and
updating said machine learning model based on an output of said receiver neural network, wherein said training further comprises applying an inter-symbol interference function of said channel model to said sequence of transmission symbols to generate inter-message interference symbols as said sequence of received symbols.
2. The method of claim 1 , wherein said applying comprises applying a first inter-symbol interference function to said sequence of transmission symbols to generate first inter-message interference symbols as said sequence of received symbols.
3. The method of claim 2 , wherein said first inter-symbol interference function comprises a pulse-shaping function.
4. The method of claim 2 , wherein said applying comprises applying a second inter-symbol interference function to said sequence of transmission symbols to generate second inter-message interference symbols as said sequence of received symbols.
5. The method of claim 4 , wherein said second inter-symbol interference function comprises a convolution with a channel impulse response.
6. The method of claim 1 , wherein said applying comprises applying a first inter-symbol interference function to said sequence of transmission symbols to generate first inter-message interference symbols and applying a second inter-symbol interference function to said first inter-message interference symbols to generate second inter-message interference symbols as said sequence of received symbols.
7. The method of claim 1 , wherein said analysing comprises setting an observation window within said sequence of received symbols and classifying that portion of said sequence of received symbols falling within said observation window.
8. The method of claim 7 , wherein said training further comprises adjusting at least one of a location and a size of said observation window to maximise a classification accuracy of said decoded message.
9. The method of claim 1 , wherein said method is performed by a transmitter of a source device, a receiver of a destination device, or a configuration device for said data transmission network.
10. The method of claim 1 , wherein said training further comprises estimating a characteristic of said sequence of received symbols using an characteristic estimating neural network, applying a compensation function to said sequence of received symbols based on said characteristic to generate compensated received symbols and analysing said compensated received symbols using said receiver neural network to generate said decoded message.
11. A method for configuring a data transmission network, executed by a configuration device, wherein said data transmission network comprises a transmitter, a receiver, and a communication channel between said transmitter and said receiver, said method comprising:
training a machine learning model of said data transmission network, wherein said machine learning model comprises at least a transmitter model including a transmitter neural network, a channel model, and a receiver model including a receiver neural network, wherein said training comprises:
providing a message within a sequence of messages;
generating a group of transmission symbols for each message in said sequence of messages using said transmitter neural network;
concatenating said groups of transmission symbols together as a sequence of transmission symbols;
simulating transmission of said sequence of transmission symbols over said communication channel using said channel model to said receiver;
analysing a sequence of received symbols using said receiver neural network to generate a decoded message, wherein said analysing comprises setting an observation window within said sequence of received symbols and classifying that portion of said sequence of received symbols falling within said observation window; and
updating said machine learning model based on an output of said receiver neural network, wherein said training further comprises adjusting at least one of a location and a size of said observation window to maximise a classification accuracy of said decoded message.
12. An apparatus, comprising:
at least one processor; and
at least one memory storing instructions that, when executed by said at least one processor, cause said apparatus to perform a set of operations including training a machine learning model of a data transmission network comprising a transmitter, a receiver, and a communication channel between said transmitter and said receiver, wherein said machine learning model comprises at least a transmitter model including a transmitter neural network, a channel model, and a receiver model including a receiver neural network, wherein said training comprises:
providing a message within a sequence of messages;
generating a group of transmission symbols for each message in said sequence of messages using said transmitter neural network;
concatenating said groups of transmission symbols together as a sequence of transmission symbols;
simulating transmission of said sequence of transmission symbols over said communication channel using said channel model to said receiver;
analysing a sequence of received symbols using said receiver neural network to generate a decoded message; and
updating said machine learning model based on an output of said receiver neural network, wherein the training further comprises applying an inter-symbol interference function of said channel model to said sequence of transmission symbols to generate inter-message interference symbols as said sequence of received symbols.
13. The apparatus of claim 12 , wherein said applying comprises applying a first inter-symbol interference function to said sequence of transmission symbols to generate first inter-message interference symbols as said sequence of received symbols.
14. The apparatus of claim 13 , wherein said first inter-symbol interference function comprises a pulse-shaping function.
15. The apparatus of claim 13 , wherein said applying comprises applying a second inter-symbol interference function to said sequence of transmission symbols to generate second inter-message interference symbols as said sequence of received symbols.
16. The apparatus of claim 15 , wherein said second inter-symbol interference function comprises a convolution with a channel impulse response.
17. The apparatus of claim 12 , wherein said applying comprises applying a first inter-symbol interference function to said sequence of transmission symbols to generate first inter-message interference symbols and applying a second inter-symbol interference function to said first inter-message interference symbols to generate second inter-message interference symbols as said sequence of received symbols.
18. The apparatus of claim 12 , wherein said analysing comprises setting an observation window within said sequence of received symbols and classifying that portion of said sequence of received symbols falling within said observation window.
19. The apparatus of claim 18 , wherein said training further comprises adjusting at least one of a location and a size of said observation window to maximise a classification accuracy of said decoded message.
20. The apparatus of claim 12 , wherein said training further comprises estimating a characteristic of said sequence of received symbols using an characteristic estimating neural network, applying a compensation function to said sequence of received symbols based on said characteristic to generate compensated received symbols and analysing said compensated received symbols using said receiver neural network to generate said decoded message.