In this thesis, new adaptive OCXO
frequency drift correction algorithms are proposed for the timing module
on the base transceiver stations. The recursive system
identification methods are used to replace the previous Batch Least
Squares (BLS) method in the algorithm. Two different recursive
system identification methods are evaluated and compared, the Recursive
Least Squares (RLS) method and the Kalman Filter
method.
New system models which include the digital control loop are created.
Simulation results show that the new system model has better performance than
the previous model.
The Cumulative Time
Error (CTE) upperbound of the timing module is analyzed. This upperbound
determines the performance bound of the timing module system. First, a simple
model structure of the OCXO frequency stability is used
to investigate the CTE upperbound. In this simple
model, the temperature is linear related to the frequency stability. Then, a
refined model structure is used to investigate the CTE
upperbound. In this refined model, both temperature
effect and ageing effect are considered. The control loop is included in both
simple and refined model structures. The simulation results show that the CTE upperbound can be obtained
analytically.