Characterisation of mobile radio signals: base station crosscorrelation

A three-dimensional model of mobile radio propagation has been used to study the crosscorrelation between signals received on two spatially separated antennas at a base station site. The factors that have the greatest influence have been identified. It has been shown that, although vertical separation can be used to obtain signals that are sufficiently decorrelated to produce significant diversity gain, horizontal separation is still generally preferable. A vertical separation of 20 wavelengths (20 lambda ) results in a crosscorrelation which is always less than 0.7, irrespective of mobile location, in a cell of radius 1 km, but the same value is only obtainable from 50% of locations if the cell radius is doubled. For horizontally separated antennas, the correlation coefficient is less than 0.7 for signals received from 98% of locations in a cell of radius 1 km, and this is only reduced to 87% of locations if the radius is increased to 5 km. In some circumstances, composite separation (vertical plus horizontal) may be useful.<>