Global Positioning System Reference
In-Depth Information
1
2
∂
φ
∂
k
=
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
Bv
+
Ax
+
w
=
o
(4.47)
1
2
∂
φ
∂
x
=−
A
T
k
=
o
(4.48)
Th
e solution of (4.46) to (4.48) starts with the recognition that
P
is a square matrix
an
d can be inverted. Thus, the expression for the residuals follows from (4.46):
P
−
1
B
T
k
v
=
(4.49)
Su
bstituting (4.49) into (4.47), we obtain the solution for the Lagrange multiplier:
k
M
−
1
(
Ax
=−
+
w
)
(4.50)
[10
with
Lin
—
*
1
——
Lon
PgE
r
M
r
=
r
B
nn
P
−
nn
B
r
(4.51)
Finally, the estimate
x
follows from (4.48) and (4.50)
=−
A
T
M
−
1
A
−
1
A
T
M
−
1
w
x
(4.52)
The estimates
x
and
v
are independent of the a priori variance of unit weight. The
first step is to compute the parameters
x
from (4.52), then the Lagrange multipliers
k
from (4.50), followed by the residuals
v
(4.49). The adjusted parameters and adjusted
observations follow from (4.38) and (4.39).
The caret symbol in
v
,
k
, and
x
indicates that all three estimated values follow from
minimizing of
v
T
Pv
. However, as stated earlier, the caret is only used consistently for
the estimated parameters
x
in order to simplify the notation.
[10
4.4.3 Cofactor Matrices
Eq
uation (4.44) shows that
w
is a random variable because it is a function of the
ob
servation
b
. With (4.37), the law of variance-covariance propagation (4.34), and
th
e use of
B
in (4.42), the cofactor matrix
Q
w
becomes
BP
−
1
B
T
Q
w
=
=
M
(4.53)
Fr
om (4.53) and (4.52) it follows that
Q
x
=
A
T
M
−
1
A
−
1
(4.54)
Combining (4.49) through (4.52) the expression for the residuals becomes
=
P
−
1
B
T
M
−
1
A
A
T
M
−
1
A
−
1
A
T
M
−
1
P
−
1
B
T
M
−
1
w
v
−
(4.55)