System and method of robust sequence estimation in the presence of channel mismatch conditions
AssigneeInternational Business Machines Corporation
InventorFrancesco Gozzo
Life of this patent
Estimating the term…
Abstract
An adaptive communications receiver for data-transmission which comprises a channel estimator and sequence estimator. The channel estimator consists of a recursive least squares algorithm which uses a known training signal or past decisions to provide a channel estimate. The sequence estimator consists of a new recursive algorithm which uses the channel estimate to estimate the information sequence transmitted over the channel. The receiver exploits the commonality between the sequence and channel estimator to provide an efficient implementation. The receiver can have varying degrees of non-linearity and is lower in complexity and more robust than linear or decision feedback equalizers in the presence of channel mismatch.
Claims32
A receiving system for use with an adaptive receiver for providing an output signal x.sub.k-(L-1) 80 in response to an input signal r.sub.k 60 received from a discrete-time equivalent channel which has a linear channel filter 20 having channel filter coefficients being coupled to a transmitted information sequence x.sub.k and additive white gaussian noise 40, said receiving system comprising:
- a sequence estimator and a means for providing a channel estimate;
- said sequence estimator 70 having as inputs a received signal r.sub.k and a channel filter estimate 95 from a channel estimator 90 whose L coefficients are determined by said means for providing a channel estimate, and as output the hard decision x.sub.k-(L-1) 80 which is an estimate of the transmitted information sequence x.sub.k 10 according to the recursive relation:
- U.sub.k =.gamma..sup.-1 P.sub.k-1
- v.sub.k =U.sub.k q.sub.k ##EQU12##
- .eta..sub.k =r.sub.k -q.sub.k.sup.T c.sub.k-1
- P.sub.k =U.sub.k -k.sub.k v.sub.k.sup.T
- c.sub.k =c.sub.k-1.sup.D +k.sub.k .eta..sub.k
- x.sub.k-(L-1) =sgn(fc.sub.k)
- c.sub.k-1.sup.S =c.sub.k-1.sup.T S+x.sub.k g
- c.sub.k-1.sup.D =sgn.sub.D (c.sub.k-1.sup.S)
- P.sub.k =S.sup.T P.sub.k S+.epsilon..sup.-1 gI
- where .gamma. is the desired exponential weighting parameter for the sequence estimator, P.sub.k-1 is an L by L inverse correlation matrix initialized to .epsilon..sup.-1 I, I is the L by L identity matrix, .epsilon. is the desired initialization parameter for the sequence estimator, sgn(z) denotes a binary threshold device which returns a one (1) if its parameter z is greater than or equal to zero and minus one (-1) if z is less than zero, c.sub.k is the soft-estimate vector of the transmitted information sequence x.sub.k through x.sub.k-(L-1) initialized to an arbitrary vector, sgn.sub.D (z) denotes the D-threshold device which applies the sgn-function to the left-most D bits in the vector z, f is the L-length pick-off vector given by
- f=[1 0 . . . 0]
- g is the L-length initialization vector given by
- g=[0 . . . 0 1]
- S is the L by L shift matrix given by ##EQU13## q.sub.x is the L-length channel estimate vector assigned by the channel estimator 90.
A receiver system for estimating an information sequence from a signal received on a channel, said signal comprising said information sequence with a partially unknown channel filter effect, said system comprising:
- an estimator coupled to receive said signal;
- and
- means for selectively inputting to said estimator in addition to said signal,
- during a plurality of time intervals, parameters of said estimated information sequence to estimate said channel filter effect, and
- during another, different plurality of time intervals, parameters of said estimated channel filter effect to estimate said information sequence;
- and wherein
- same physical hardware, firmware and/or software of said estimator receives and estimates both said channel filter effect and said information sequence and said estimator uses same error criterion to estimate both said channel filter effect and said information sequence.
A method for estimating an information sequence from a signal received on a channel, said signal comprising said information sequence with a partially unknown channel filter effect, said method comprising the steps of:
- selectively inputting to an estimator, during a plurality of time intervals, said signal and parameters of said estimated information sequence to estimate said channel filter effect;
- selectively inputting to said estimator during another, different plurality of time intervals, said signal and parameters of said estimated channel filter effect to estimate said information sequence;
- and
- wherein same physical hardware, firmware and/or software of said estimator receives and estimates both said channel filter effect and said information sequence and said estimator uses same error criterion to estimate both said channel filter effect and said information sequence.
A receiver system for estimating an information sequence from a signal received on a channel, said signal comprising said information sequence with a partially unknown channel filter effect, said system comprising:
- an estimator coupled to receive said signal;
- and
- means for selectively inputting to said estimator in addition to said signal,
- during a plurality of time intervals, parameters of said estimated information sequence to estimate said channel filter effect, and
- during another, different plurality of time intervals, parameters of said estimated channel filter effect to estimate said information sequence;
- and wherein
- same physical hardware, firmware and/or software of said estimator receives and estimates both said channel filter effect and said information sequence, said estimator uses same error criterion to estimate both said channel filter effect and said information sequence, and said estimator, when estimating said information sequence, is based on the following:
- U.sub.k =.gamma..sup.-1 P.sub.k-1
- v.sub.k =U.sub.k q.sub.k ##EQU16##
- .eta..sub.k =r.sub.k -q.sub.k.sup.T c.sub.k-1
- P.sub.k =U.sub.k -k.sub.k v.sub.k.sup.T
- c.sub.k =c.sub.k-1 +k.sub.k .eta..sub.k
- where .gamma. is a desired exponential weighting parameter for the sequence, P.sub.k-1 is an L by L inverse correlation matrix initialized to .epsilon..sup.-1 I,I is an L by L identity matrix, .epsilon. is a desired initialization parameter for the estimator for the information sequence, c.sub.k is a soft-estimate vector of the information sequence initialized to an arbitrary vector, and q.sub.k.sup.T is a training vector whose coefficients are past L known training bits x.sub.k.
A receiver system for estimating an information sequence from a signal received on a channel, said signal comprising said information sequence with a partially unknown channel filter effect, said system comprising:
- an estimator coupled to receive said signal;
- and
- means for selectively inputting to said estimator in addition to said signal,
- during a plurality of time intervals, parameters of said estimated information sequence to estimate said channel filter effect, and
- during another, different plurality of time intervals, parameters of said estimated channel filter effect to estimate said information sequence;
- and wherein
- same physical hardware, firmware and/or software of said estimator receives and estimates both said channel filter effect and said information sequence, said estimator uses same error criterion to estimate both said channel filter effect and said information sequence, and said estimator, when estimating said information sequence, is based on the following:
- c.sub.k =[Q.sub.k.sup.T W.sub.k Q.sub.k ].sup.-1 Q.sub.k.sup.T W.sub.k r.sub.k,
- where a received signal vector r.sub.k represents a L-length vector with a past L received signals r.sub.k, a training matrix Q is an L by L matrix whose rows are past L training vectors q.sup.T.sub.k whose contents are given by the estimator for the channel filter effect, W.sub.k is an arbitrary weighing matrix, and r.sub.k is updated with each new bit by a relation r.sup.T.sub.k =r.sup.T.sub.k-1 S+r.sub.k g.
Dependencies are inferred from the claim text.