This preview shows page 1. Sign up to view the full content.
Unformatted text preview: = KM AP = ab 2 ln(bπ0a2 1 ) .
b2 −
Finally, we ﬁnd the error probabilities for the magnitude threshold test (3.12) with an arbitrary
threshold K > 0. Substituting in KM L or KM AP for K gives the error probabilities for the ML and 3.10. BINARY HYPOTHESIS TESTING WITH CONTINUOUSTYPE OBSERVATIONS 117 MAP tests:
∞ −K pf alse alarm = P {X  > K  H0 } = f0 (u)du = Φ(−K/a) + 1 − Φ(K/a) = 2Q(K/a). f0 (u)du +
−∞
K pmiss = P {X  < K  H1 } = K f1 (u)du = Φ(K/b) − Φ(−K/b) = 1 − 2Q(K/b).
−K Example 3.10.3 Based on a sensor measurement X , it has to be decided which hypothesis about
a remotely monitored room is true: H0 : the room is empty vs. H1 : a person is present in the
1
room. Suppose if H0 is true then X has pdf f0 (x) = 2 e−x+1 and if H1 is true then X has pdf
f1 (x) = 1 e−x−1 . Both densities are deﬁned on the whole real line. Find the ML decision rule, the
2
MAP decision rule for prior probability distribution (π0 , π1 ) = (2/3, 1/3), and the associated error
probabilities, including the average error probability based on the given prior.
Solution: To help with the compu...
View
Full
Document
 Spring '08
 Zahrn
 The Land

Click to edit the document details