In this letter, blind adaptive H-infinity multiuser detection is developed
by employing a generalized sidelobe canceler (GSC) with and without subweig
ht partition scheme. It is shown that the adaptive H-infinity algorithm wit
h subweight approach has the advantages of fast convergence speed, insensit
ivity of dynamic estimate error, and suitability for arbitrary ambient nois
e over the conventional H-infinity and the RLS-based adaptive algorithms.