Adaptive demodulating method for generating replica and demodulator thereof

The adaptive demodulator as set forth in claim wherein the matrix R.sub.m (i) and the vector V.sub.m (i) are defined as follows

• R.sub.m (i)=A.sub.m.sup.H (i).LAMBDA. A.sub.m (i)
• V.sub.m (i)=A.sub.m.sup.H (i).LAMBDA. X(i)
• .LAMBDA.=diag (1.lambda.. . . .lambda..sup.N-1)
• where N represents an integer equal to or greater than 1 and .lambda. represents a constant satisfying 0<.lambda.<1.

An adaptive demodulator, comprising:

• a transversal filter controlled by a coefficient vector to simulate a delayed wave component of an input signal, for producing a simulation delayed wave component;
• differential means for subtracting an output of said transversal filter from the input signal and outputting a desired signal component;
• deciding means for deciding the desired signal component supplied from said differential means and outputting a desired signal;
• and
• updating means supplied with a sequence of M desired signals from a time point i to a past time point (i-M+1) and the input signal, for calculating the coefficient vector at the time point i by using a generalized inverse matrix produced from the desired signal sequence and the coefficient vector at a time point (i-1), and supplying the coefficient vector to said transversal filter.

An adaptive demodulator, comprising:

• maximum likelihood sequence estimating means for generating desired signal sequence candidates and interference signal sequence candidates corresponding to respective state transitions at each state for an input signal at each of discrete time points at predetermined intervals, selecting a desired signal sequence candidate with a high likelihood state transition and an interference signal sequence candidate with a high likelihood state transition from all state transitions by using an error vector, and outputting the selected desired signal sequence candidate and interference signal sequence candidate;
• replica generating means for generating a desired modulated wave candidate vector and an interference modulated wave candidate vector from each said desired signal sequence candidate and said interfered signal sequence candidate supplied from said maximum likelihood sequence estimating means, and calculating an inner product of the desired wave candidate vector and the interference wave coefficient vector corresponding to each state of said maximum likelihood sequence estimating means to generate a desired wave replica and an interference wave replica;
• error calculating means for subtracting a desired wave replica vector with elements of a sequence of the generated desired wave replicas and an interference wave replica vector with elements of a sequence of the generated interference wave replicas from an input signal vector with elements of a sequence of the input signals to produce the error vector;
• and
• updating means for updating the desired wave coefficient vector at the present time point i by using a generalized inverse matrix produced from a desired modulated wave candidate corresponding to the selected desired signal sequence candidate and a desired wave coefficient vector at a past time point (i-1) and updating the interference wave coefficient vector at the present time point i by using a generalized inverse matrix produced from an interference modulated wave candidate corresponding to the selected interference signal sequence candidate and an interference wave coefficient vector at the past time point (i-1).

An adaptive demodulator, comprising;

• maximum likelihood sequence estimating means for generating signal sequence candidates corresponding to respective state transitions at each state for first and second input signals supplied from first and second branches at each of discrete time points at predetermined intervals, selecting signal sequence candidates with high likelihood state transitions from all state transitions, and outputting the selected signal sequence candidates;
• first and second replica generating means for generating modulated wave candidate vectors from each signal sequence candidate received from said maximum likelihood sequence estimating means, and calculating an inner product of the modulated wave candidate vectors and first and second coefficient vectors corresponding to each state of said maximum likelihood sequence estimating means to generate first and second replicas;
• first error calculating means for calculating the difference between a first replica vector with elements of a sequence of the first replicas and a first input signal vector with elements of a sequence of the first input signals to generate a first error vector;
• second error calculating means for calculating the difference between a second replica vector with elements of a sequence of the second replicas and a second input signal vector with elements of a sequence of the second input signals to generate a second error vector;
• likelihood generating means for calculating the sum of norms of the first and second error vectors as a signal corresponding to the likelihood and supplying the signal to said maximum likelihood sequence estimating means;
• and
• first and second updating means for updating the first and second coefficient vectors at the present time point i by using a generalized inverse matrix produced from a modulated wave candidate corresponding to the selected signal sequence candidate and the first and second coefficient vectors at the past time point (i-1).

An adaptive demodulating method for generating a replica, comprising the steps of:

• (1) providing a signal sequence candidate corresponding to a state transition for each state from maximum likelihood sequence estimating means to replica generating means;
• (2) generating by the replica generating means, a modulated wave candidate vector from a signal sequence candidate corresponding to the state transition and calculating an inner product of the modulated wave candidate vector and a coefficient vector corresponding to each state of the maximum likelihood sequence estimating means to produce a replica;
• (3) generating by error calculating means, a replica vector with elements of a sequence of generated replicas and an input signal vector with elements of a sequence of input signals, and producing a difference between these vectors as an error vector;
• (4) causing the maximum likelihood sequence estimating means to select a high likelihood state transition candidate from the error signal sequences corresponding to the state transitions for each state and output a decided signal corresponding to the most likelihood state;
• and
• (5) causing updating means to calculate the inner product of a generalized inverse matrix produced from a modulated wave candidate corresponding to the selected state transition and the error vector and update the coefficient vector with the inner product for each state.