The increasing popularity of mobile communications is leading to the saturation of the available capacity of existing mobile communication systems (e.g. GSM). The expected shift from voice to multimedia communications will accentuate this problem enormously. Hence the capacity of mobile communication systems needs to be increased dramatically. The only dimension that has not been exploited to gain capacity is the spatial dimension. Indeed, multiple users can communicate simultaneously and on the same frequency band by distinguishing them according to their direction This is spatial division multiple access (SDMA). In order to realize SDMA in the presence of inter symbol interference due to multipath propagation, the spatial channel transfer matrix has to be known. Its straightforward estimation requires protocol overhead of which the relative importance increases drastically with the number of users served. We wish to develop blind techniques that arrive at estimating the channel without overhead, to make SDMA feasible.