Supplement:

1. Scalar case (3.14)

𝐸[𝑥|𝑧] = 𝑥̂ = 𝑥̅ + 𝑃𝐻𝑇 𝑉 −1 (𝑧 − 𝐻𝑥̅ ) (3.14)

1.1 If the noise is very small, i.e., V−→ 0

𝑉 𝑀
𝐸[𝑥|𝑧] = 𝑥̅ + 𝑧 −→ 𝑧
𝑀+𝑉 𝑀+𝑉

 The 𝑥̂ MV is only dependent of the measurement

1.2 If the noise is very large, i.e., V−→ ∞

𝑉 𝑀
𝐸[𝑥|𝑧] = 𝑥̅ + 𝑧−→ 𝑥̅
𝑀+𝑉 𝑀+𝑉

 The 𝑥̂ MV is only dependent of the state regardless of the measurement

2. Batch process / recursive process

Example: a random process as

𝑥𝑘 = x1 , x2 , … , xn

Find the average of 𝑥𝑘

2.1 Batch Process

1
𝑠̅ = (𝑥 + 𝑥2 + ⋯ + 𝑥𝑛 ) (1)
𝑛 1

2.2 Recursive process

Define
𝑘
1
s𝑘 = ∑ 𝑥𝑘
𝑘
𝑖=1

Then

𝑘+1 𝑘
1 1 1
s𝑘+1 = ∑ 𝑥𝑘 = 𝑥𝑘 + ∑ 𝑥𝑘
𝑘+1 𝑘+1 𝑘+1
𝑖=1 𝑖=1

1 𝑘
= 𝑥𝑘 + 𝑠
𝑘+1 𝑘+1 𝑘
k 1
= 𝑠𝑘 + 𝑥 (2)
k+1 𝑘 + 1 𝑘+1
 2.3 Merits for recursive way

- The memory size is lower than the batch type

- The result can be acquired at every step, which is more informative.

3. (3.20) example

z = x + v , x~N(2,1), 𝑣~𝑁(0,1/2)

We have three measurements of z=(1,-1,3). Find the E[x|z].

3.1 batch Process

1) Algorithm

𝐸[𝑥|𝑧] = 𝑥̂ = 𝑥̅ + 𝑃𝐻𝑇 𝑉 −1 (𝑧 − 𝐻𝑥̅ ), 𝑃 = (𝑀−1 + 𝐻𝑇 𝑉 −1 𝐻)−1

Where

1
M = 1, H = 1, V =
2

2) With the first z1 = 1

−1
1 −1
- 𝑃 = (𝑀−1 + 𝐻𝑇 𝑉 −1 𝐻)−1 = (1−1 + ( ) ) = 1/3
2

- 𝐸[𝑥|𝑧1 = 1] = 𝑥̂ = 𝑥̅ + 𝑃𝐻 𝑇 𝑉 −1 (𝑧 − 𝐻𝑥̅ )

1 1 −1 2 4
= 2 + ( ) ( ) (1 − 2) = 2 + (−1) =
3 2 3 3

3) two measurements

(z1 , 𝑧2 ) = (1, −1) −→ 𝐸[𝑥|𝑧1 , 𝑧2 ]

𝑧1 1 1
 [𝑧 ] = [ ] 𝑥 + [ ] 𝑣
2 1 1
−1
1 −1 1 1
 P = (M −1 + 𝐻𝑇 𝑉 −1 𝐻)−1 = (1−1 + [1 1] ( ) [ ]) = (1 + 4)−1 =
2 1 5

1 1 −1 1 1 2 −1
 𝑥̂ = 𝑥̅ + 𝑃𝐻𝑇 𝑉 −1 (𝑧 − 𝐻𝑥̅ ) = 2 + ( ) [1 1] ( ) ([ ] − [ ] 2) = 2 + ( ) [1 1] ([ ])
5 2 −1 1 5 −3

2 8 2
= 2 + ( ) [−1 − 3] = 2 − =
5 5 5

4) Three measurements
(z1 , 𝑧2 , 𝑧3 ) = (1, −1,3) −→ 𝐸[𝑥|𝑧1 , 𝑧2 , 𝑧3 ]
 1 −1
1 −1 1
 P = (M −1 + 𝐻𝑇 𝑉 −1 𝐻)−1 = (1−1 + [1 1 1] ( ) [1]) = (1 + 6)−1 =
2 7
1

1 1 −1
1 1 2
−1
 𝑥̂ = 𝑥̅ + 𝑃𝐻𝑇 𝑉 −1 (𝑧 − 𝐻𝑥̅ ) = 2 + [1 1 1] ( ) ([−1] − [1] 2) = 2 + [1 1 1] ( [−3])
7 2 7
3 1 1
2 8
= 2 + (−3) =
7 7

3.2 Recursive M.V. estimator

1) Algorithm

k=1, x~N(𝑥̅ , 𝑀), 𝑣𝑘 ~𝑁(0, 𝑉𝑘 )

𝑥̂1 = 𝑥̅ + 𝐾1 (𝑧1 − 𝐻1 𝑥̅ )
K1 = (𝑀−1 + 𝐻1𝑇 𝑉1−1 𝐻1 )−1 𝐻1𝑇 𝑉1−1
P1−1 = 𝑀−1 + 𝐻1𝑇 𝑉1−1 𝐻1

k=2

𝑥̂2 = 𝑥̂1 + 𝐾2 (𝑧2 − 𝐻2 𝑥̂1 )

K 2 = (𝑃1−1 + 𝐻2𝑇 𝑉2 𝐻2 )−1 𝐻2𝑇 𝑉2−1
P2−1 = 𝑃1−1 + 𝐻2𝑇 𝑉2−1 𝐻2

2) With the first measurement z1 = 1

1 −1
P1−1 = (1−1 + (1) ( ) (1)) = 3
2

1 1 −1 2 4
𝑥̂1 = 𝑥̅ + 𝐾1 (𝑧1 − 𝐻1 𝑥̅ ) = 2 + ( ) ( ) (1 − 2) = 2 − =
3 2 3 3

3) With the second measurement (z1 , 𝑧2 ) = (1, −1)

1 −1
P2−1 = 𝑃1−1 + 𝐻2𝑇 𝑉2−1 𝐻2 = 3 + (1) ( ) (1) = 5
2

4 1 1 −1 4 4 2 7 4 14 6 2
𝑥̂2 = 𝑥̂1 + 𝐾2 (𝑧2 − 𝐻2 𝑥̂1 ) = + ( ) ( ) (−1 − ) = + (− ) = − = =
3 5 2 3 3 5 3 3 15 15 5

3) dsa

Sda