A learning on a weight of an adaptive array antenna which becomes necessary for a space division multiple access is carried out while a backward compatibility with a conventional standard is maintained.When a TRQ frame and a training frame for the learning are exchanged, a spoof is applied on these frames to preferably avoid collisions with a surrounding station STA4 that follows a conventional standard. Also, after the weight of the adaptive array antenna is learnt, a STA0 multiplexes and transmits the plurality of data frames addressed to a plurality of users while a wireless resource on a spatial axis is shared by the plurality of users, so that a throughput for one-to-many, that is, for the plurality of users as a whole is improved.