| 摘要 |
A channel estimation method and apparatus for GMS/EDGE digital communications systems utilizing previously unexploited properties of GMS/EDGE training sequences to permit a more efficient initial channel estimation for equalizer operation and for joint synchronization and equalizer window sizing. In particular, any consecutive 16-symbol segment of the 26-symbol GMS/EDGE training sequences is both shift invariant and order invariant; and these properties enable channel estimation to be carried out on delayed (shifted) training sequence segments, permitting ISI corrupted leading symbols to be avoided in computations and leading taps to be estimated, using the same training sequence segments, regardless of equalizer window size; and to enable all the 1-8 tap channels to be estimated without matrix inversion, permitting a significant reduction in computational complexity. A two-dimensional recursive relation is established that leads to efficient channel tap calculation for different sync points and tappositions which enables practical joint optimization of synchronization and window sizing for the equalizer.
|
