NEURAL NETWORK AUGMENTED GPS/LORAN-C NAVIGATION SYSTEM

NETROLOGIC PROPOSES TO BUILD A NAVIGATION SYSTEM FOR SHIPS WHICH INTEGRATES GPS AND LORAN-C. THE GPS SYSTEM PROVIDES ASF ERROR CORRECTION TO LORAN-C SO THAT THE CORRECTED LORAN-C IS ABLE TO FURNISH NEAR GPS QUALITY GEOGRAPHICAL LOCATION INFORMATION. IN AREAS WHERE DGPS IS AVAILABLE THE DGPS SIGNAL CAN BE FURNISHED AS A REFERENE TO REALIZE CORRESPONDING GAINS IN ACCURACY FOR CORRECTED LORAN0-C. A NEURAL NETWORK USES LORAN-C SIGNALS AS INPUT AND GPS SIGNALS AS A REFERENCE SIGNAL TO LEARN THE DRIFT OF ERROR OVER TIME AND PROVIDES AN ESTIMATE OF TOTAL ERROR IN THE CORRECTED SYSEMS. WHEN GPS BECOMES AVAILABLE THE SYSTEM WILL AUTOMATICALLY REVERT TO THE GPS AIDED MODE AND UPDATE ITS CORRECTION TABLE. THE SYSTEM CAN BE IMPLEMENTED ON A PC IN PHASE I AND CAN BE GREATLY REDUCED IN SIZE TO FIT IN A SMALL PACKAGE FOR COMMERCIAL APPLICATIONS. TWO APPROACHES WILL BE USED IN PHASE I. ONE APPROACH IS A PURE NEURAL NETWORK APPROACH WHICH PROVIDES CORRECTED A GEOGRAPHICAL REFERENCE SIGNAL. THE SECOND APPROACH PROVIDES THE ASF CORRECTIONS IN A SEPARATE TABLE.