A new conception of Decision Feedback-feedforward Equalizer （DFFE） is proposed. Two types of optimum receivers derived under MMSE criterion with and without ZF constraint are described

and their tap adjustment algorithms are also given. An upper bound on the error probability in DFFE with ZF constraint is derived. It is proved that either with or without ZF constraint

the error probability in DFFE is lower than in DFE. Computer simulation shows that the adaptive scheme is feasible.For simple implementation

the preequalizer may be omitted in DFFE. Convergence is then independent of channel parameters

and the number of its taps is reduced to a minimum.