Artificial Neural Networks (ANN) have proven capability in Forward Error Correction (FEC) encoding and decoding. This proposal strives to address the potential that ANNs may enable non-cooperating receivers to identify, configure, and recover information from a complex communication channel. Modern digital communication systems have been designed to transmit data at extremely high rates - sometimes in excess of 100 Mbits/second. To protect the information is these systems from errors, Forward Error Coding (FEC) has been implemented as a critical component in these systems. Forward error correction is a low latency method of recovering the data through a computational process that relies on redundancy in the transmitted data. The ability to rapidly identify and decode communication channels containing containing FECs is of great interest to elements within DOD. Researchers have been successful in the development of ANNs capable of performing the coding and decoding the FEC in cooperating systems. This has stimulated the concept developed in this proposal. The current analysis technique to identify a channel as having FEC is a brute force, man and machine intensive task, using algebraic techniques, and frequency analysis. It is the intent of this effort to investigate the possibility that ANNs will be able to improve the results of processing an unknown FEC. Anticipated benefits/potential commercial applications - the benefits of the application of ANNs to FECs are to be gained in the improved efficiency in analyzing new and complex communication channels. An additional benefit could be gained in the implementation of adaptive error correcting receivers, devices capable of identifying and tracking the error code selection of a transmitter, where the transmitter selects the mode of error correction in response to
