A discrete Bayesian  filter (DBF) was  developed to work  as a pattern  matching
technique  with  database  correlation  for  mobile  location.  The  database is
constructed using a  3D deterministic radio  wave propagation prediction  model.
Three  methods to  yield final  location estimates  are presented  and compared
according to their accuracy. We carried out field measurements in a GSM  network
deployed  in a  suburban environment,  which is  very common  and widespread  in
Europe. We  show that  more work  has to  be done  in order  to make positioning
accuracy as good as in urban and dense urban areas.