Method of constructing and training a neural network
A method of processing a radio frequency signal includes: receiving the radio frequency signal at an antenna of a receiver device; processing, by a radio frequency front-end device, the radio frequency signal; converting, by an analog-to-digital converter, the analog signal to a digital signal; receiving, by a neural network, the digital signal; and processing, by the neural network, the digital signal to produce an output.
1. A method of constructing and training a neural network that is configured to process a signal, the method comprising:
identifying a processing function to be performed on the signal by the neural network;
determining, an optimum format for training vectors to be applied to the neural network;
determining a number of layers of the neural network, number of nodes per layer, the node connectivity structure, and initial weights of the neural network; and
inputting, into the neural network, a plurality of formatted training vectors in order to train the neural network to perform the processing function on the signal.
2. The method of claim 1 , comprising:
recording trained weights of the neural network after completion of training of the neural network;
evaluating performance of the processing function of the neural network by using the trained weights; and
iteratively adjusting the training vector format, the number of layers of the neural network, the number of nodes per layer of the neural network, and the node connectivity structure until desired performance of the processing function is achieved.
3. The method of claim 1 , wherein the processing function is classification of a modulation scheme.
4. The method of claim 1 , wherein the number of layers is two.
5. The method of claim 1 , wherein the optimum format of the training vector includes two-dimensional constellation points with resolution based on an amount or type of distortion in a transmission channel through which the radio frequency signal can travel.