|
Motorola provide services for optimising mobile phone network performance
in terms of network parameters such as
pilot powers, antenna down tilts, antenna azimuths and neighbour lists.
This project focuses on assessing the changes in performance of CDMA
networks after an optimisation of the network is made.
Mobile phone call data is monitored from the network for some
period of time before and after the optimisation. However, seasonal
fluctuations, market changes, and long-term growth of a network make
accurate assessment of the effect of a change difficult. An additional
complicating factor is that
physical problems with the
network and data collection equipment can cause periods of data from
some
regions of the network to go unrecorded.
Part of this work is concentrates on statistical measures, particularly
Bayesian changepoint analysis for multiple timeseries, for assessing
the effect of change. A second focus of the work concerns relating
key performance indicators (KPIs) from actual networks to performance
metrics from simulations. Statistical pattern recognition methods are
being employed for this aspect of the project.
|
 |
