Communication system
A communications system and method is described comprising: handshaking between a transmitter and a receiver of the communication system to initiate a training procedure, wherein said handshaking comprises a training setup request message comprising parameters for the training procedure, wherein the transmitter comprises trainable parameters and/or the receiver comprises trainable parameters; receiving identified training data from the transmitter at the receiver, wherein the training data comprises transmitter training data and/or receiver training data; sending training information from the receiver to the transmitter, wherein the training information comprises information for controlling training at the transmitter and/or the receiver; and terminating the training procedure.
1 . A receiver of a communication system, the receiver comprising:
at least one processor; and
at least one memory storing instructions that, when executed by the at least one processor, cause the receiver to:
handshake with a transmitter of the communication system to initiate a training procedure, wherein said handshaking comprises a training setup request message comprising parameters for the training procedure, wherein the transmitter comprises trainable parameters or the receiver comprises trainable parameters;
receive identified training data from the transmitter, wherein the training data comprises transmitter training data or receiver training data;
send training information from the receiver to the transmitter, wherein the training information comprises information for controlling training at the transmitter or the receiver, and performance information that enables the trainable parameters of the transmitter to be trained;
send a training interrupt message to the transmitter for indicating that training of the receiver or the transmitter is to be terminated;
terminate the training procedure; and
train, based on labelled training data, by updating the trainable parameters of the receiver,
wherein the training data received from the transmitter comprises said receiver training data,
wherein the communication system comprises an OFDM communication system having a pilot allocation that allocates pilot signals among resource elements of an OFDM transmission,
wherein the pilot allocation includes determining a ratio of power to allocate to the pilot signals associated with each resource element,
wherein the training comprises learning a mapping between received signals and log likelihood ratios (LLRs), given the pilot allocation,
wherein an allocation matrix corresponding to the pilot allocation is trainable,
wherein the pilot signals are superimposed with data symbols according to the allocation matrix, and
wherein trainable parameters are updated and the ratio of power is optimized according to a loss function.
2 . The receiver as claimed in claim 1 , wherein the receiver training data comprises a sequence of known symbols sent by the transmitter to the receiver.
3 . The receiver as claimed in claim 1 , wherein the identified training data is identified by being provided in a dedicated training data channel.
4 . The receiver as claimed in claim 1 , wherein the training information comprises feedback data indicative of a readiness of the receiver to receive further receiver training data from the transmitter.
5 . The receiver as claimed in claim 1 , wherein the at least one memory and instructions further cause the receiver to perform sending a receiver initiated training setup request to the transmitter for initiating the handshaking.
6 . The receiver as claimed in claim 1 , wherein the at least one memory and instructions further cause the receiver to perform sending a receiver training complete message to the transmitter for indicating that training of the receiver is complete.
7 . A transmitter of a communication system, the transmitter comprising:
at least one processor; and
at least one memory storing instructions that, when executed by the at least one processor, cause the transmitter to:
handshake with a receiver of the communication system to initiate a training procedure, wherein said handshaking comprises a training setup request message comprises parameters for the training procedure, wherein the transmitter comprises trainable parameters or the receiver comprises trainable parameters;
send identified training data to the receiver, wherein the training data comprises transmitter training data or receiver training data;
receive training information from the receiver at the transmitter, wherein the training information comprises information for controlling training at the transmitter or the receiver, and performance information that enables the trainable parameters of the transmitter to be trained;
send a training interrupt message to the receiver for indicating that training of the receiver or the transmitter is to be terminated;
terminate the training procedure;
train, based on the training data and the training information, by updating the trainable parameters of the transmitter, wherein the training data sent to the receiver comprises said transmitter training data, wherein the communication system comprises an OFDM system, wherein the transmitter generates a pilot allocation that allocates pilot signals among resource elements of the OFDM system, wherein the pilot allocation includes determining a ratio of power to allocate to the pilot signals associated with each resource element, and wherein the training comprises training an allocation matrix corresponding to the pilot allocation; and
superimpose the pilot signals with data symbols by scaling the pilot signals and the data symbols according to the allocation matrix, wherein trainable parameters are updated and the ratio of power is optimized according to a loss function.
8 . The transmitter as claimed in claim 7 , wherein the receiver training data comprises a sequence of known symbols sent by the transmitter to the receiver.
9 . The transmitter as claimed in claim 7 , wherein the identified training data is identified by being provided in a dedicated training data channel.
10 . The transmitter as claimed in claim 7 , wherein the training information comprises feedback data indicative of a readiness of the receiver to receive further receiver training data from the transmitter.
11 . The transmitter as claimed in claim 7 , wherein the at least one memory and instructions further cause the transmitter to perform sending a transmitter-initiated training setup request for initiating the handshaking.
12 . The transmitter as claimed in claim 7 , wherein the at least one memory and instructions further cause the transmitter to perform sending a transmitter training complete message to the receiver for indicating that training of the transmitter is complete.
13 . A method, comprising:
handshaking between a transmitter and a receiver of a communication system to initiate a training procedure, wherein said handshaking comprises a training setup request message comprising parameters for the training procedure, wherein the transmitter comprises trainable parameters or the receiver comprises trainable parameters;
receiving identified training data from the transmitter at the receiver, wherein the training data comprises transmitter training data or receiver training data;
sending training information from the receiver to the transmitter, wherein the training information comprises information for controlling training at the transmitter or the receiver, and performance information that enables the trainable parameters of the transmitter to be trained;
sending a training interrupt message to the transmitter for indicating that training of the receiver or the transmitter is to be terminated;
terminating the training procedure; and
training, based on labelled training data, by updating the trainable parameters of the receiver,
wherein the training data received from the transmitter comprises said receiver training data,
wherein the communication system comprises an OFDM communication system having a pilot allocation that allocates pilot signals among resource elements of an OFDM transmission,
wherein the pilot allocation includes determining a ratio of power to allocate to the pilot signals associated with each resource element,
wherein the training comprises learning a mapping between received signals and log likelihood ratios (LLRs), given the pilot allocation,
wherein an allocation matrix corresponding to the pilot allocation is trainable,
wherein the pilot signals are superimposed with data symbols according to the allocation matrix, and
wherein trainable parameters are updated and the ratio of power is optimized according to a loss function.
14 . A method, comprising:
handshaking between a transmitter and a receiver of a communication system to initiate a training procedure, wherein said handshaking comprises a training setup request message comprising parameters for the training procedure, wherein the transmitter comprises trainable parameters or the receiver comprises trainable parameters;
sending identified training data from the transmitter to the receiver, wherein the training data comprises transmitter training data or receiver training data;
receiving training information from the receiver at the transmitter, wherein the training information comprises information for controlling training at the transmitter or the receiver, and performance information that enables the trainable parameters of the transmitter to be trained;
sending a training interrupt message to the receiver for indicating that training of the receiver or the transmitter is to be terminated;
terminating the training procedure;
training, based on the training data and the training information, by updating the trainable parameters of the transmitter, wherein the training data sent to the receiver comprises said transmitter training data, wherein the communication system comprises an OFDM system, wherein the transmitter generates a pilot allocation that allocates pilot signals among resource elements of the OFDM system, wherein the pilot allocation includes determining a ratio of power to allocate to the pilot signals associated with each resource element, wherein the training comprises training an allocation matrix corresponding to the pilot allocation; and
superimposing the pilot signals with data symbols by scaling the pilot signals and the data symbols according to the allocation matrix, wherein trainable parameters are updated and the ratio of power is optimized according to a loss function.
15 . A non-transitory computer-readable medium comprising instructions encoded thereon which, when executed on an apparatus, cause the apparatus to perform at least:
handshaking between a transmitter and a receiver of a communication system to initiate a training procedure, wherein said handshaking comprises a training setup request message comprising parameters for the training procedure, wherein the transmitter comprises trainable parameters or the receiver comprises trainable parameters;
receiving identified training data from the transmitter at the receiver, wherein the training data comprises transmitter training data or receiver training data;
sending training information from the receiver to the transmitter, wherein the training information comprises information for controlling training at the transmitter or the receiver, and performance information that enables the trainable parameters of the transmitter to be trained;
sending a training interrupt message to the transmitter for indicating that training of the receiver or the transmitter is to be terminated, or sending a training interrupt message to the receiver for indicating that training of the receiver or the transmitter is to be terminated;
terminating the training procedure; and
training, based on labelled training data, by updating the trainable parameters of the receiver,
wherein the training data received from the transmitter comprises said receiver training data,
wherein the communication system comprises an OFDM communication system having a pilot allocation that allocates pilot signals among resource elements of an OFDM transmission,
wherein the pilot allocation includes determining a ratio of power to allocate to the pilot signals associated with each resource element,
wherein the training comprises learning a mapping between received signals and log likelihood ratios (LLRs), given the pilot allocation,
wherein an allocation matrix corresponding to the pilot allocation is trainable,
wherein the pilot signals are superimposed with data symbols according to the allocation matrix, and
wherein trainable parameters are updated and the ratio of power is optimized according to a loss function.