Pulse or digital communications – Equalizers – Automatic
Reexamination Certificate
2000-09-07
2001-10-16
Chin, Stephen (Department: 2634)
Pulse or digital communications
Equalizers
Automatic
C375S229000, C375S230000, C375S231000
Reexamination Certificate
active
06304599
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to an adaptive equalizer and an adaptive equalization scheme, and particularly to an adaptive equalizer applicable to digital radio communication equipment in digital mobile communication, digital satellite communication, digital mobile-satellite communication and the like.
2. Description of Related Art
In digital mobile communication, fading—variations in the amplitude and phase of a received signal—can occur because of reflection, diffraction or scattering of radio waves due to geography and terrestrial materials around a mobile station. In particular, when the delay time of delay waves cannot be neglected as compared with a symbol length, the spectrum of a signal is distorted, resulting in large degradation.
Such fading is called frequency selective fading because the spectral distortion has frequency dependency. An adaptive equalizer is one of conventional effective techniques to overcome such fading.
As configurations of conventional adaptive equalizers, are known a decision feedback equalizer (referred to as DFE from now on) that eliminates the effect of delay waves by feeding back decision results, and a maximum likelihood sequence estimation (referred to as MLSE from now on) that selects a maximum likelihood sequence from among all the sequences having possibilities to be transmitted.
Although the MLSE is a little larger in size than the DFE, it has better performance than the DEE because it can utilize the power of the delay waves.
As for fading resulting from fast time variations in channel characteristics, the adaptive MLSE is more advantageous which carries out tracking following variations in the channel characteristics not only during training period that obtains channel impulse responses (called CIR from now on) from a known training sequence, but also during data sections.
In particular, the MLSE that carries out channel estimation for respective states of Viterbi algorithm (referred to as per-survivor processing MLSE from now on) exhibits good performance even for fast time-varying channels by carrying out the CIR estimation for respective states of the MLSE.
A configuration of a per-survivor processing MLSE will be described here as a typical conventional adaptive equalizer.
FIG. 1
is a block diagram showing a configuration of a conventional per-survivor processing MLSE disclosed in H. Kubo, K. Murakami and T. Fujino, “An Adaptive Maximum-Likelihood Sequence Estimator for Fast Time-Varying Intersymbol Interference Channels”, IEEE Transactions on Communications, Vol. 42, Nos. 2/3/4, 1994, pp. 1872-1880 (called REF. 1 below). In this figure, the reference numeral 11 designates a maximum likelihood sequence estimating section;
12
a
-
12
n
each designate a CIR estimator;
101
designates a received baseband signal;
102
designates estimated CIRs of respective states;
103
designates tentative decisions of respective states; and
104
designates hard decision data.
Next, the operation of the conventional device will be described.
The maximum likelihood sequence estimating section
11
, receiving the received baseband signal
101
and estimated CIRs of respective states
102
, estimates a transmitted sequence by Viterbi algorithm, and outputs its results as hard decision data
104
.
FIG. 2
is a block diagram showing an internal configuration of the maximum likelihood sequence estimating section
11
. In
FIG. 2
, the reference numeral
21
designates a branch metric generator;
22
designates an ACS (add-compare-select) operation circuit;
23
designates a path metric memory;
24
designates a path memory;
201
15
designates branch metrics;
202
designates path metrics;
203
designates path metrics at previous timing; and
204
designates a survivor path.
In the maximum likelihood sequence estimating section
11
within the conventional per-survivor processing MLSE with the foregoing configuration, a state s
k
and a path connected to a branch s
k
/s
k−1
at time k of the Viterbi algorithm are defined by the following expressions (1) and (2).
s
k
=[Ĩ
k
, Ĩ
k−1
. . . , Ĩ
k−V+1
] (1)
s
k
/s
k−1
=[Ĩ
k
,Ĩ
k−1
, . . . , Ĩ
k−V
] (2)
where, Ĩ
k
is a candidates for the transmitted sequence determined by the state s
k
or by the branch s
k
/s
k−1
.
The branch metric generator
21
compares replicas of the received signal obtained from the estimated CIRs of respective states
102
with the received baseband signal
101
, generates branch metrics
201
for all the branch candidates s
k
/s
k−1
, and supplies them to the ACS operation circuit
23
.
Assuming that a metric criteria is a squared Euclidean distance, the branch metrics
201
can be expressed by the following expressions (3) and (4).
&Ggr;
k
[s
k
/s
k−1
]=|r
k
{circumflex over (r)}
k
[s
k
/s
k−1
]|
2
(3)
r
^
k
[
s
k
/
s
k
-
1
]
=
∑
i
=
0
L
c
i
,
k
[
s
k
-
1
]
I
~
k
-
i
(
4
)
where, &Ggr;
k
[s
k
/s
k−1
] is a branch metric
201
of the branch s
k
/s
k−1
, r
k
is the received baseband signal
101
, {circumflex over (r)}
k
[s
k
/s
k−1
] is a replica of the received signal determined by the branch s
k
/s
k−1
, c
i,k
[s
k−1
] is an estimated CIR
102
at the state s
k−1
, and L is a channel memory length. The branch
15
metric generator
21
also outputs candidates (Ĩ
k
, Ĩ
k−1
, . . . ,Ĩ
k−V+1
) of the transmitted sequence determined by the state s
k
as the tentative decisions of respective states
103
to be supplied to the CIR estimators
12
a
-
12
n.
The ACS operation circuit
22
adds the branch metrics
201
to the path metrics at previous timing
203
stored in the path metric memory
23
as the following expression (5) to obtain path metric candidates for all the branch candidates s
k
/s
k−1
.
H
k
[s
k
/s
k−1
]=H
k−1
[s
k−1
]+&Ggr;
k
[s
k
/s
k−1
] (5)
where H
k
[s
k
/s
k−1
] is the path metric candidate determined by the branch s
k
/s
k−1
, and H
k−1
[s
k−1
] is a path metric at previous timing
203
determined by the state s
k−1
. In addition, the ACS operation circuit
22
compares the path metric candidates H
k
[s
k
/s
k −1
] for each state s
k
as the following expression (6) to select a minimum path metric and supplies the minimum path metrics thus obtained to the path metric memory
23
as the path metrics
202
.
H
k
[
s
k
]
=
min
{
s
k
-
1
}
→
s
k
H
k
[
s
k
/
s
k
-
1
]
(
6
)
where, H
k
[s
k
] is the path metric
202
determined by the state s
k
. The ACS operation circuit
22
also supplies the path memory
24
with the information on the selected path as the survivor path
204
.
The path memory
25
stores the survivor paths
204
for a predetermined time period, traces the paths whose path metrics at previous timing
203
are minimum, and outputs the transmitted sequence determined by the paths as the hard decision data
104
.
Each of the CIR estimators
12
a
-
12
n
which are prepared by the number of the states of the maximum likelihood sequence estimating section
11
, receives the received baseband signal
101
and the tentative decision of respective states
103
, estimates the CIR for respective states using the LMS (least mean square) algorithm, and outputs the estimated CIR of respective states
102
. Specifically, as the following expression (7), the CIR estimators
12
a
-
12
n
update all the estimated CIRs for all the states s
k
and channels i(i=0, . . . ,L) to be output as the estimated CIRs of respective states
102
.
c
i,k+
[s
k
]=c
i,k
[s
k−1
:s
k
sv
]&dgr;(
r
k
c
i,k
[s
k−1
:s
k
sv
]Ĩ
k−i
&
Chin Stephen
Ghayour Mohammad
Mitsubishi Denki & Kabushiki Kaisha
LandOfFree
Adaptive equalizer and adaptive equalization scheme does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Adaptive equalizer and adaptive equalization scheme, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Adaptive equalizer and adaptive equalization scheme will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2600682