Decentralized trust assessment

Abstract

A decentralized trust assessment system, comprising a neural network, a trust module, and a local subsystem, wherein the trust module controls whether a plurality of inputs to the local subsystem are trustworthy. The decentralized trust assessment system provides rotorcraft and tiltrotor aircraft with airborne systems able to detect bad and spoofed data from a wide variety of data streams.

What is claimed is: 
     
         1 . A decentralized trust assessment system, comprising:
 a neural network;   a trust module; and   a local subsystem;   wherein the trust module controls whether a plurality of inputs to the local subsystem are trustworthy.   
     
     
         2 . The decentralized trust assessment system of  claim 1 , wherein the neural network is located between the plurality of inputs and the trust module. 
     
     
         3 . The decentralized trust assessment system of  claim 1 , wherein the neural network is located inside the trust module. 
     
     
         4 . The decentralized trust assessment system of  claim 1 , further comprising:
 a plurality of outputs from the local subsystem and the trust module;   wherein the neural network is provided the plurality of outputs.   
     
     
         5 . The decentralized trust assessment system of  claim 4 , wherein the neural network provides feedback to the trust module through the plurality of inputs. 
     
     
         6 . The decentralized trust assessment system of  claim 1 , wherein the neural network is based upon a training set. 
     
     
         7 . The decentralized trust assessment system of  claim 1 , wherein the neural network is based upon a trusted training set. 
     
     
         8 . A method of decentralizing trust assessments, comprising:
 training a neural network to create a trained neural network;   programming a trust module to review a data stream for a condition;   reviewing the data stream with the trust module for the condition; and   analyzing the data stream for a pattern with the trained neural network.   
     
     
         9 . The method of  claim 8 , further comprising:
 flagging the data stream if the condition is met.   
     
     
         10 . The method of  claim 8 , further comprising:
 flagging the data stream if the pattern is detected by the trained neural network.   
     
     
         11 . The method of  claim 8 , the step of training comprising:
 summing the data stream before and after a local subsystem.   
     
     
         12 . The method of  claim 8 , the step of training comprising:
 summing the data stream before and after a local subsystem in combination with the trust module.   
     
     
         13 . The method of  claim 8 , wherein the step of analyzing the data stream for a pattern with the trained neural network occurs before the step of reviewing the data stream with the trust module for the condition. 
     
     
         14 . The method of  claim 8 , wherein the step of analyzing the data stream for a pattern with the trained neural network occurs after the step of reviewing the data stream with the trust module for the condition. 
     
     
         15 . A decentralized trust assessment system of an aircraft, comprising:
 at least one input data stream from the aircraft;   a local subsystem in the aircraft, the local subsystem configured to act upon the at least one input data stream;   a trained neural network; and   a trust module configured to analyze the at least one input data stream;   wherein the trust module controls whether the at least one input data stream to the local subsystem is acted upon by the local subsystem.   
     
     
         16 . The decentralized trust assessment system of  claim 15 , wherein the trained neural network is located between the at least one input data stream and the trust module. 
     
     
         17 . The decentralized trust assessment system of  claim 15 , wherein the trained neural network is located inside the trust module. 
     
     
         18 . The decentralized trust assessment system of  claim 15 , further comprising:
 at least a first output of the local subsystem and of the trust module;   wherein the trained neural network reviews the at least a first output.   
     
     
         19 . The decentralized trust assessment system of  claim 18 , wherein the trained neural network provides feedback to the trust module. 
     
     
         20 . The decentralized trust assessment system of  claim 19 , wherein the trust module replaces the first output based on the trained neural network.