Chapter V: Selected Solutions

H. V. Poor
Princeton University
October 14, 2002

Exercise 3:
Since sk depends completely on Y k0 , it behaves like a constant when conditional quantities
given Y k0 are computed. The only point at which this affects the derivation of the Kalman-
Bucy filter is in the time update equations, which now become:
ˆ t+1|t = Ft X
X ˆ t|t + E{Γt st |Y t0 } = Ft X
ˆ t|t + Γt st ,

Σt+1|t = Ft Σt|t FTt + Gt Qt GTt + Cov Γt st |Y t0 = Ft Σt|t FTt + Gt Qt GTt .
Note that the second of these two equations is the same as when there is no measurement
The measurement feedback has no effect on the measurement update equations. Al-
though the presence of this feedback may cause the state to be nonGaussian, the joint
conditional statistics of X t and Y t given Y t−1
0 are still Gaussian. Thus, the measurement
update is unchanged from the case of no measurement feedback since it depends only on
this joint Gaussian property and the linearity of measurement equation.

Exercise 4:
There are several ways of approaching this problem. An interesting one is to note that,
although U t and V t are dependent, the Gaussian vector U t ≡ U t −Ct R−1
t V t is independent
of V t . To take advantage of this, we may add the zero quantity Ct Rt [Y t − Ht X t − V t ]
to the tth state input, which yields the equivalent state equation
X t+1 = Ft X t + Gt U t + Gt Ct R−1
t (Y t − Ht X t ) .

So, we have an equivalent problem with independent state and measurement noises,
with the measurement
 feedback term Gt Ct R−1t Y t , and with the new state matrix
Ft − Gt Ct Rt Ht . We also have a different correlation matrix for the state input, since

Cov(U t ) = Qt − Ct R−1 T
t Ct .


(V. where the last equality follows by definiton of Kk . we have    E Xj − X ˆ j|k X Tk+1 = Σak+1|k . Thus. The desired recursions thus follow by eliminating either set of updates from (V.          E Xj − X ˆ j|k X Tk+1 = E X j − X ˆ j|k (Fk X k + Gk U k )T = E X j − X ˆ j|k X Tk FTk . we then have      −1    ˆ j|k X T Xj − X ˆ k|k−1 X Tk k+1 = Σk|k−1 I − Hk Hk Σk|k−1 Hk + Rk Y k − Hk X a T T E E FTk   −1    = Σak|k−1 I − HTk Hk Σk|k−1 HTk + Rk Hk E ˆ k|k−1 X T Xk − X FTk k   −1   = Σak|k−1 I− HTk Hk Σk|k−1 HTk + Rk Hk Σk|k−1 FTk = Σak|k−1 I − HTk KTk FTk . ˆ j|k . since Y t = Ht X t + V t . Fk ≡ 1. we note here that X j and Y t are jointly Gaussian conditioned on Y 0 . for k > j.14) .B.B. As with the measurement-update derivation in the Kalman-Bucy  filter.2. on applying the recursion for X          E ˆ j|k X Tk+1 = E Xj − X ˆ j|k−1 X Tk FTk − Kak E Xj − X ˆ k|k−1 X Tk FTk . This result follows by induction on t. E X j |Y t0 will be t−1 as given by the recursion if we have     −1 Cov X j . Applying the recursion for Σak+1|k . X  2 E θˆn − Θ . Rk ≡ σ 2 .Applying the result of Exercise 3 and eliminating the measurement update equations yields the given result. Σ0 ≡ v 2 . But. and Σn|n ≡ Σn|n ≡ ˆ n+1|n ≡ X Qk ≡ 0. which shows that the given equation for t = k + 1. We first note that. the result from Part a. where we use the fact that U k has zero mean and is independent of both X j and X ˆ j|k to this equation we have Now. Y k − Hk X We now assume that the given equality is true for t = k. 2 .16). Exercise 9: This is the Kalman-Bucy problem with all dimensions equal to unity. for t = j. Hk = sk . implies this equality. the given equality follows by definition. b. From the state equation we have that.B. Exercise 6: a. The induction principle thus gives the desired result. Y t |Y t−1 0 Cov Y t |Y t−1 0 = Kat . From this and the definition of Kak . The resulting estimate is the same as that found in Example IV. m0 ≡ µ. Gk ≡ ˆ n|n ≡ θˆn .