Source code for modest.signals.signalsource

## @file SignalSource.py holds the SignalSource base class
#
from abc import ABCMeta, abstractmethod
from scipy.stats import multivariate_normal as mvn
from scipy.linalg import block_diag
import numpy as np


class SignalSource():
    __metaclass__ = ABCMeta
    nextSignalID = 0
    
    def __init__(
            self,
            stateObjectID=None
    ):
        self.__signalID__ = SignalSource.nextSignalID
        SignalSource.nextSignalID += 1
        self.stateObjectID = stateObjectID
        return

    def signalID(self):
        return self.__signalID__

    def computeAssociationProbability(
            self,
            measurement,
            stateDict,
            validationThreshold=0
    ):
        
        dY = None
        R = None
        H = None
        myMeasMat = {}

        if not self.stateObjectID:
            return 0
        
        substate = stateDict[self.stateObjectID]['stateObject']
        myMeasMat = substate.getMeasurementMatrices(measurement, source=self)
            
        for key in myMeasMat['dY']:
            if H is None:
                H = myMeasMat['H'][key]
                R = myMeasMat['R'][key]
                dY = myMeasMat['dY'][key]
            else:
                H = np.vstack([H, myMeasMat['H'][key]])
                R = block_diag(R, myMeasMat['R'][key])
                dY = np.append(dY, myMeasMat['dY'][key])

        if dY is not None:
            P = substate.covariance()
            Pval = P.convertCovariance('covariance').value
            S = H.dot(Pval).dot(H.transpose()) + R

            myProbability = mvn.pdf(dY, cov=S)
        else:
            myProbability = 0
        return myProbability