Use of neural network based matched filter for fast response time in high-speed communications channels
A neural network is used within a receiver to discriminate a large set of input waveforms without using a very large set of conventional matched filters. The neural network is trained under actual line conditions as opposed to the requirement for ideal signals when using matched filters. The finite waveforms are based on digital modulation principles. A best match is made between a received waveform from the noisy channel and that of previously trained waveforms in order to extract data. Neural network based matched filter allows data be discriminated separately for each sub-carrier channel in the receiver. The neural network system allows fast processing and is suitable for high-speed data communications systems.
1. A communication channel transmitting a number of subcarrier frequencies comprising:
super resonance filters; and
digital neural network based matched filters, wherein each of the super resonance filters and each of the neural network based matched filters are equal in number to the number of said subcarrier frequencies and wherein said digital neural network is trained in the presence of channel noise.
2. The channel of claim 1 , wherein at a 1-bit output at each said neural network matched filter said subcarrier frequencies equals in number to a number of bits for the channel.
3. The channel of claim 1 , wherein at a 2-bit output at each said neural network matched filter said subcarrier frequencies is half in number to a number of bits for the channel.
4. The channel of claim 1 , wherein at a 3-bit output at each said neural network matched filter said subcarrier frequencies is one-third in number to a number of bits for the channel.
5. The channel of claim 1 , wherein said neural network matched filter includes a cubic polynomial transfer function.
6. A method of reducing hardware complexity and speed of processing data recovery in a communication system comprising:
training digital neural network matched filters, in the presence of channel noise;
receiving a number of subcarrier frequencies in said system;
providing a number of super resonance filters, wherein said number of super resonance filters is equal in number to the number of subcarrier frequencies and said digital neural network matched filters.
7. The method of claim 6 , wherein after said training of said neural network, said neural network is adapted while in use.