7B-5-2 
conventional photogrammetric flight. 
Introduction of OTF technique will enhance the 
practicability -of GPS photogrammetry, all 
limitations caused by GPS will be released The 
aircraft can fly in the same manner as in 
conventional photogrammetric flight. The 
airborne GPS receiver can be turned on while 
the aircraft is entering the photogrammetric area 
to cut down unnecessary records. 
In general, the various OTF techniques offer a 
similar strategy: firstly, an ambiguity search 
space is constructed, then searching for correct 
integer ambiguities begins by applying certain 
validation and rejection criteria. The search 
process is stopped and the ambiguities are fixed 
whenever certain assurance criteria are fulfilled. 
This paper proposes a new OTF method based 
on double difference Pseudorange/carrier phase 
Kalman filter Firstly, search space is constructed 
based on real value ambiguities and the associate 
covariance matrix. Then, optimized Cholesky 
decomposition searching process is executed. If 
a certain ambiguity combination passes Ratio 
test and OVT test, it will be the right ones. This 
paper also presents the practice of this method in 
processing the airborne GPS kinematic data of 
the Beijing GPS photogrammetric project. 
2 OTF METHOD BASED ON DOUBLE 
DIFFERENCE PSEUDORANGE, CARRIER 
PHASE KALMAN FILTER 
2.1 Double difference pseudorange, carrier 
phase Kalman fdter 
The first step of this OTF method is to determine 
real value ambiguities and associate covariance 
matrix. This can be done by processing 
pseudorange and carrier phase observations 
using least square adjustment or kalman filter. 
Kalman filter was adopted in this paper 
considering pseudorange and carrier phase are 
independent, the mathematic model of double 
difference pseudorange/carrier phase Kalman 
filter can be written as: 
X t =<1>*-A- 1 +r t . l w tA w t ~N{0,Q„) (l) 
^k, P ~H k ^ p X k +v k ' P v k 'p 
(2) 
■^k,<t> = + v k,t v k ,<i> " 
- N(0,R k j) 
(3) 
\yhere: 
X k is the system state vector, 
X k = (dx,dy,dz,x,y,z,dn 0 ,dn l ,--- ,dn n )‘ 
O* is the state transition matrix; 
H kp ,H k 0 are observation matrix for double 
difference pseudorange and carrier phase 
respectively; 
T k is system driven noise matrix; 
R k p.R-k <i> are covariance matrix for double 
difference pseudorange and carrier phase 
measurements respectively; 
Q k is the covariance matrix for system noise. 
Equation (1) is the dynamic model for aircraft, 
which is a polynomial model of second or third 
order; Equation (2), (3) express the linearized 
observation models for double difference 
pseudorange and carrier phase measurements 
respectively. The sequential estimation formula 
of Kalman filter is: 
~ ^ k,k~\X k-\,k-1 
(4) 
Vi = 
^ k^k-i^k-hk-fò k,k-\ 
■r^Qk-i r;_, 
t (5) 
^k,k,p ' 
= ('\Li + H T kp R- k l 
Hk P r l 
(6) 
K k, p - 
-- P k ,k, P H T kp R kp - x 
(7) 
x k , k ,p 
— ¿*,-1 x k P [Z kp 
-H kp x k ' k _ 
,1(8) 
II 
(Pkl P +H T t*R;}H 
k *r l 
(9) 
--i\^H T k ,R k ; x 
(10) 
Xu* = 
X k,k,p + x [Z kl/l 
~ h k( p x kkp 
](H) 
Equation (4)~(11) process pseudorange, carrier