Note: Descriptions are shown in the official language in which they were submitted.
CA 02625710 2011-05-09
-1-
METHOD FOR RECEIVING FRAMES OF A DIGITAL STREAM
FIELD OF THE INVENTION
This invention relates to digital radiofrequency broadcasting, and in
particular digital broadcasting
by satellite.
BACKGROUND
The satellite broadcasting standard referred to as DVB-S2 was developed for
the broadcasting of
video streams as well as the broadcasting of IP application packets. This
standard enables constant,
variable or adaptive coding and modulation modes. The structure proposed by
this standard involves
a concatenation of a BCH (Bose-Chaudluri-Hocquenghem) code and an LDPC (Low
Density Parity
Check) code. This concatenation is preceded by an energy dispersion step and
followed by a QPSK,
8-PSK, 16-APSK or 32-APSK modulation step and a step of inserting a header in
n/2 BPSK.
The header contains transmission parameters. The header includes a 5-bit field
called MODCOD
defining the type of modulation and the coding rate of the frame. This field
is previously unknown
to the receiver and is protected by a Reed-Muller code (32,5) having the
following generating
matrix:
0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1
0 0 1 1 0 0 1 1 0 0 1 1 0 0 1 1 0 0 1 1 0 0 1 1 0 0 1 1 0 0 1 1
0 0 0 0 1 1 1 1 0 0 0 0 1 1 1 1 0 0 0 0 1 1 1 1 0 0 0 0 1 1 1 1
0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 l 1
The header also includes a 2-bit field called TYPE defining the length and the
type of frame. This
field is previously unknown to the receiver. The header also includes an
identification sequence at
the beginning of the frame called SOF, previously known to the receiver.
To lock the frame, at each symbol time, the receivers envisaged in the
literature compute a
correlation level by correlation computations between 26 consecutive sampled
symbols. When the
correlation level exceeds a threshold, it is considered that these 26 symbols
correspond to the SOF
identification sequence. A frequency shift error estimation is established on
the basis of these
symbols and used in a frequency-tracking loop.
CA 02625710 2011-05-09
-2-
Such a receiver has disadvantages. Indeed, the locking performed by the
identification of symbols of
the SOF sequence can be unreliable when the digital stream is transmitted in a
noisy environment.
Similarly, the frequency shift error estimation established has a reduced
reliability in such an
environment. The speed and reliability of the locking are thus insufficient,
which is particularly
detrimental since, in the absence of decoding of a header, the position of the
next headers cannot be
determined, as the frames have variable sizes.
Therefore, there is a need for simple means involving a reduced processing
power in order to
overcome one or more of these disadvantages.
SUMMARY
In accordance with one aspect, the present invention relates to a method for
receiving a digital
stream by a receiver, the stream comprising frames, each frame having a body
composed of a
variable quantity c of symbols preceded by a header having a total quantity d
of symbols, the header
comprising at least one known field of which the receiver has a priori
knowledge, and at least one
unknown field of which the receiver does not have a priori knowledge, the at
least one unknown
field composed of a fixed quantity e of symbols and encoded such that the
receiver has a priori
knowledge of a first relationship between certain groups of the e symbols and
of a second
relationship between certain other groups of the e symbols, the method
comprising:
a) storing, by the receiver, at least one series of symbols of the digital
stream including at
least a portion of the d symbols of the header;
b) performing, by the receiver, a first correlation computation on at least a
portion of the at
least one series of symbols, wherein the first correlation computation is
based on the a
priori knowledge of the first relationship between the certain groups of the e
symbols;
c) performing, by the receiver, a second correlation computation on at least a
portion of the
at least one series of symbols, wherein the second correlation computation is
based on
the a priori, knowledge of the second relationship between the certain other
groups of the
e symbols;
CA 02625710 2011-05-09
-3-
d) determining a correlation level based at least on a combination of results
of the first and
the second correlation computations; and
e) comparing the correlation level against a threshold level and, based on a
result of that
comparison, determining frame synchronization for at least one series of
symbols.
In accordance with another aspect, the present invention relates to a method
for transmitting a digital
stream, comprising:
- generating frames of a digital stream, in which each frame comprises a
header and a
body, wherein:
o the header has a first portion that contains a set of predefined symbols and
a second
portion that contains a set of non-predefined symbols, the non-predefined
symbols
representing at least one parameter of the body; and
o the set of non-predefined symbols is composed of a first subset and a second
subset;
and
o the first subset is encoded with a first deterministic encoding to produce a
first
encoded portion; and
o the second subset and the first encoded portion are together encoded with a
second
deterministic encoding that defines a first set of predefined relationships
and a second
set of predefined relationships between the symbols of the set of non-
predefined
symbols;
- transmitting the frame.
The invention also relates to a method for receiving a digital stream by a
receiver, the stream
comprising frames having each a body composed of a variable number c of
symbols preceded by a
header including a number d of symbols, the header comprises at least one
field composed of a fixed
number e of symbols unknown to the receiver and defining the number c, said
method having a
locking phase comprising:
CA 02625710 2011-05-09
-4-
a) receiving at least one series of e symbols of the digital stream;
b) for a plurality of first pairs of received symbols, performing a
correlation computation between
the symbols of each pair, wherein said correlation computation is used to
verify a predefined
relationship if the e symbols form the unknown field;
c) determining a correlation level on the basis of the performed correlation
computation and
determining whether the e symbols of the digital stream form the unknown field
based on said
correlation level;
d) repeating steps a), b) and c) until it is determined that the e symbols of
the digital stream form the
unknown field.
In one embodiment of the invention, during step b), a plurality of correlation
computations between
second pairs of symbols are also performed, said correlation being assumed to
verify a predefined
relationship if the e symbols form the unknown field, and all of the symbols
of the second pairs
having symbols in common with all of the symbols of the first pairs.
In another embodiment, the symbols of the second pairs belong primarily to all
of the symbols of the
first pairs.
In yet another embodiment, the product of the symbols of each of the pairs is
assumed to be
identical if the e symbols form the unknown field.
In yet another embodiment of the invention:
the header of each frame also has a field composed of a fixed number f of
symbols known to the
receiver;
in step a), a series of d symbols of the digital stream is received;
in step b), the symbols of each first pair correspond either to the known
field or to the unknown
field;
CA 02625710 2011-05-09
-5-
in step c), the correlation level is determined on the basis of correlation
computations performed
respectively on the symbols corresponding to the known field and on the
symbols corresponding
to the unknown field;
in step d), it is determined whether the d symbols of the digital stream form
a header according
to the correlation level.
According to yet another embodiment:
a first number g summing the correlation computations performed on the first
pairs of symbols
corresponding to the known field and a second number h summing the correlation
computations
performed on the first pairs of symbols corresponding to the unknown field are
formed;
the correlation level Corr is determined by the following function:
Corr=max(lg+hl, Ig-hl).
According to another embodiment, the correlation computations are performed on
the basis of all of
the symbols corresponding to the known field and to the unknown field.
In yet another embodiment, an estimation of a frequency shift error is
performed on the basis of said
symbols received, and the demodulation frequency of a symbol receiving circuit
is corrected on the
basis of said estimation.
In another embodiment of the invention, a flexible decoding of the symbols of
the unknown field
determined is performed, a likelihood level of the decoded field is
determined, and an unlocking is
determined when the likelihood level of the unknown field is below a
threshold.
In yet another embodiment, the flexible decoding is a Reed-Muller decoding
combining the symbols
of the unknown field determined so as to establish a validity probability for
each combination of bits
of the coded word and in which the likelihood level is established on the
basis of the highest
probability.
According to an embodiment, the method further includes the following steps:
CA 02625710 2011-05-09
-6-
e) after the locking, for each frame, the e symbols forming the unknown field
of its header are
decoded so as to deduce the number c and deduce whether the receiver is
addressed by the frame;
and
f) if the receiver is not addressed by the frame, the receiver does not decode
the body and decodes
the header of the next frame.
According to another embodiment:
in the decoding of the header in step e), it is deduced whether the phase
indicators are present in the
body of the frame; and
if the presence of phase indicators is deduced in the body of the frame, the
phase of the symbols
corresponding to the body is corrected on the basis of these phase indicators.
The frames received may comply with standard DVB-S2.
The invention also relates to a method for transmitting a digital stream, the
method comprising:
generating frames of a digital stream, in which each frame includes:
a coding of a word defining at least the variable number c of bits of the body
of the frame,
wherein the field thus coded has a fixed number e of bits and redundancies of
the bits of the
word according to a predefined arrangement;
the formation of a frame including:
a header including the coded field and a identification word at the beginning
of the frame
with redundancies and including a fixed number f of bits;
the body of the frame;
transmitting the formed frame; and
receiving, according to a method defined above, the transmitted frame.
The invention also relates to a receiver, capable to carry out a receiving
method as defined above.
CA 02625710 2011-05-09
-7-
The invention also relates to a system comprising such receiver and a
transmitter capable to carry
out the transmitting method defined above.
BRIEF DESCRIPTION OF THE DRAWINGS
Other features and advantages of the invention will become clear from the
following description,
provided solely for indicative and non-limiting purposes, in reference to the
appended drawings, in
which:
FIG. 1 schematically shows an example of a frame structure of the binary
stream;
FIG. 2 schematically shows the structure of an example of a receiver carrying
out the invention;
FIG. 3 shows a diagram of the state of the receiver of FIG. 2.
CA 02625710 2011-05-09
-8-
DETAILED DESCRIPTION
The invention proposes a method for receiving a digital stream in which each
frame has a field of a
fixed size unknown to the receiver defining the variable size of the body of
the frame. The invention
proposes locking by performing correlation computations on the symbols assumed
to correspond to
the unknown field. As the symbols of an unknown field form pairs for which the
correlation
properties verify a predefined relationship, a correlation level is determined
on the basis of the
correlation computations performed and make it possible to determine whether
the symbols received
are the symbols of the unknown field.
As the correlation level used to perform the locking of a frame of the digital
stream thus takes into
account the symbols of the unknown field, the locking can be performed quickly
and with increased
reliability. This advantage is enhanced by the size of the unknown field,
generally greater than the
size of the known field.
FIG. 1 shows an example of frame 1 transmitted in the digital stream. Frame 1
has a header
including a field C known to the receiver and a field I unknown to the
receiver. Field C has a length
of f bits, and field I has a fixed length of e bits and the header has a
length of d bits. Frame 1 has a
body with a variable length of c bits. This body includes sections S I, S2 and
S3 separated by phase
indicators P 1 and P2.
The invention proposes locking the frame by receiving a series of e symbols of
the digital stream.
The receiver previously knows the arrangement of the symbols of the unknown
field of a frame, the
product of the symbols of the unknown field verifying a predefined
relationship. In particular, the
product of symbols of the unknown field at given positions can be constant.
The receiver performs
correlation computations between first pairs of symbols received, and the
product of the symbols of
each pair being assumed to verify a predefined relationship according to the
known arrangement. A
correlation level is determined based on the performed correlation
computations. On the basis of this
correlation level, it is determined whether the received symbols form the
unknown field I, by
comparing, for example, the correlation level with a locking threshold.
Indeed, only the symbols of
the unknown field I will have a sufficient correlation level in practice,
since they are statistically the
only ones to verify the property of pairs formed according to a certain
arrangement, and having a
product verifying the predefined relationship.
CA 02625710 2011-05-09
-9-
If it is determined that the symbols of the digital stream do not correspond
to the unknown field I, a
new level of correlation is calculated for new symbols received, until the
unknown field I is
identified in order to determine a locking. Insofar as the locking is not
performed, a correlation level
can be calculated at each symbol time on the basis of the symbols contained in
a delay line storing
the received symbols.
Defined positions of symbols forming pairs in which the product verifies a
predefined relationship
can be generated by using a symmetrical coding of the unknown field I at the
transmission, for
example by using a Reed-Muller code. A Reed-Muller code results in pairs
having a constant
product.
The symbols used for the computations form other pairs for which a product has
a predefined
relationship. Correlation computations are performed on these other pairs and
are also taken into
account in order to determine the correlation level. The reliability of the
locking is thus reinforced.
The known field C is also used to perform the locking, and thus increase its
reliability and speed.
The number d of symbols, for example, is received. Pairs of symbols assumed to
belong to the
known field C, and for which a product verifying a predefined relationship is
pending, and, as
described above, pairs of symbols assumed to belonging to the unknown field,
are formed.
Correlation computations between the symbols of each pair are performed in
order to determine the
correlation level. The known field C is, for example, an identification field
consisting of a series of
identical symbols.
The following is an example of a computation of the correlation level:
Let g be the sum of the correlation computations performed on the symbols
assumed to belong to the
known field C and let h be the sum of the correlation computations performed
on the symbols
assumed to belong to the unknown field I. A correlation level can then be
obtained by the following
formula: Corr=max(g+hl, g-hI).
Such a correlation level is thus insensitive to the coding differences capable
of occurring between
the known field C and the unknown field I. To increase the reliability of the
locking, all of the
symbols of fields C and I are taken into account in order to perform the
correlation computations.
CA 02625710 2011-05-09
-10-
The symbols received are used to generate a frequency shift error estimation.
This estimation is then
used to correct the demodulation frequency of a symbol receiving circuit.
A locking confirmation test is performed on the first locked header. A
flexible decoding of the
symbols of the unknown field I is performed, then a likelihood level of this
field is deduced. When
the likelihood level of the unknown field I is below a probability threshold,
an unlocking is
identified and the locking phase is reactivated.
Similarly, a synchronization test can be performed on the headers received
after the locking. After
determining a likelihood level, this level is compared with a likelihood
threshold, preferably below
the likelihood threshold used for the locking confirmation test. If the
likelihood level is below this
threshold, a synchronization loss is identified and the locking phase is
reactivated.
When the symbols of the unknown field I correspond to a word coded by a Reed-
Muller code,
combinations of each symbol received from the unknown field are established,
with each
combination corresponding to a possible code word. A probability of validity
is calculated for each
combination, with the likelihood level corresponding to the combination having
the highest validity
probability.
After the locking, the unknown field received for each frame is decoded so as
to deduce the length c
of the body. The decoding also makes it possible to determine whether the
receiver is addressed by
the frame. If this is not the case, the receiver does not process the body of
the frame and decodes the
header of the next frame.
The receiver decodes the unknown field C in order to determine whether phase
indicators are
present in the body of the frame. If this is the case, the indicators are
decoded and the phase of the
symbols corresponding to the body is corrected on the basis of these
indicators.
The example described below is a particular application of the invention in a
communication
implementing the standard DVB-S2, defined in the recommendation ETSI EN 302-
307 vl.1.1.
The transmitter forms frames as follows. In a manner known per se, the
transmitter forms the body
of the frame by performing an energy dispersion on the data bits to be
transmitted, by forming a
BCH code and an LDPC code, then by concatenating the BCH code and the LDPC
code. A QPSK,
CA 02625710 2011-05-09
-11-
8-PSK, 16-APSK or 32-APSK modulation is then applied, optionally followed by
an insertion of
pilots and a jamming step. A header in it/2 BPSK is then inserted.
The header is formed as follows. A MODCOD field, initially unknown to the
receiver, defines the
type of modulation and the coding rate of the frame. This field is initially
coded on 5 bits. A TYPE
field, initially unknown to the receiver, defines the type of frame. This
field is initially coded on 2
bits: a first bit t0 identifying the length of the frame (0 for a frame of
64800 bits and 1 for a frame of
16200 bits) and a second bit tl indicating the presence of pilots in the frame
(t1=0 in the absence of
pilots and tl=1 in the presence of pilots).
The MODCOD field is coded by multiplication with the Reed-Muller matrix (32,5)
described in the
Background section. The 32-bit vector thus formed includes bits bo to b31. The
vector b and the field
TYPE are coded in a 64-bit vector denoted c and of which the bits will be
named Eo to 863, defined as
follows:
Elk=bk+to[2] and Elk+]= Elk+tl [2].
Due to the use of the Reed-Muller coding, it can be noted that the vector 8
has the following
properties: regardless of the values of the 5 bits of the MODCOD and of the
two bits of TYPE, the
product Ek=Ek+2" is constant for a value of p between 0 and n, with n being
the number of lines of the
Reed-Muller matrix minus one. Such a coding introduces deterministic
redundancies.
The vector E is subjected to scrambling by a fixed sequence of 64 bits in
order to obtain vector sE.
An identification sequence at the beginning of the frame SOF known to the
receiver and having a
length of 26 bits is inserted in front of the vector sp in order to form a
vector named y with a length
of 90 bits. This identification sequence is intended to facilitate the locking
of the frame at the
reception. The vector y thus obtained is subjected to a modulation in ,r/2
BPSK, in order to obtain a
vector of 90 symbols denoted yx. The symbols of yx are defined as follows:
yx2k=(1-2y2k)*d 4 and yx2k+1=(1-2y2k+1)*e137ri4
FIG. 2 schematically shows an example of a receiver structure capable to carry
out the invention.
The receiver 1 comprises modules 21 to 23.
CA 02625710 2011-05-09
-12-
Module 21 is intended to perform an adapted filtering of the received signal,
a timing recovery,
perform a decoding of the headers, a locking of the frames, and a frequency
tracking and
synchronization. Module 21 includes a digital control oscillator 211, an
adapted filter 212, a timing
recovery component 213, a header decoder 14, a frame synchronizer 215 and a
frequency-tracking
component 216. The decoding of the header in module 21 makes it possible to
deduce the length of
the frame, the presence of pilots in the frame, the constellation I,Q used and
the coding rate used.
Module 22 is intended to constitute the frame. Module 22 can determine the
position of the header
on the basis of parameters decoded in header 21 by module 21. Module 22 also
determines whether
the frame is addressed to receiver 1.
Thus, if the frame is not addressed to receiver 1 (State 304 in the figure),
module 22 cannot take into
account the symbols received until the next header.
If the frame is addressed to receiver 1, module 22 performs steps 221 and 222
in the presence of
pilots in the frame (State 303 in the figure) and steps 223 and 224 in the
absence of pilots (State 305
in the figure). Module 22 initializes a physical descrambling in step 221 or
223, then applies this
physical descrambling on each sample until the next header is reached. Module
22 then performs a
phase shift of the next header in step 222 or 224. For ease of processing, the
samples of the header
and the pilots are preferably quantified in I,Q, and the remainder of the
frame can be quantified in
I, Q or in polar coordinates. Preferably, module 22 forms the frame as a
vector containing phase data
from the header, followed by a section of the body of the frame. The vector
can contain various
other phase data from any respective pilots, with each phase data item being
followed by a
respective section of the body of the frame.
Module 23 is intended to process the frames addressed to receiver I (States
303 and 305 in FIG. 2).
In the presence of pilots (State 303), module 23 performs a precise estimation
231 and correction
232 of the frequency. In step 233, the pilots are removed from the vector. In
step 234, a tracking and
phase correction are performed on the basis of the phase data surrounding each
section. For this, it is
possible to implement a progressive phase correction and a regressive phase
correction on each
section, by initializing each correction by the phase data placed in front of
or after each section.
After correction, in step 235, a quantification of the likelihoods of the bits
carried by the symbols is
CA 02625710 2011-05-09
-13-
performed. The new frame thus obtained is successively decoded in step 236 by
an LDPC decoder
and a BCH decoder, according to standard DVB-S2.
In the absence of pilots (State 305), steps 237 to 239 are implemented. In
step 237, a tracking and
phase correction are performed on the basis of the phase data surrounding the
body, i.e. the phase
data contained in the header of the frame and the header of the next frame.
For this, it is possible to
implement a progressive phase correction and a regressive phase correction on
each frame body, by
initializing each correction respectively by the phase data placed in front of
and after the frame.
Steps 238 and 239 are substantially identical to steps 235 and 236,
respectively.
FIG. 3 shows an example of a state diagram of the operation of receiver 1 of
FIG. 2.
In state 301, receiver 1 attempts to lock the frame. Receiver 1 thus searches
for the first header using
a method of which an example is provided below. After having locked the frame
and identified the
presence of a header, receiver 1 provides a first estimation of the frequency
shift 4110. This first
estimation MD is provided to frequency-tracking component 216. The header
identified is then
corrected with this first estimation AM. The header is then decoded with an
algorithm insensitive to
any phase shift. The position of the next header is then deduced from this
decoding. In state 302,
receiver 1 has locked a frame and performed a corrected estimation Afl on its
header as well as on
the next headers. This corrected estimation Afl can in particular be
determined by the Fitz
algorithm. This corrected estimation Afl is provided to frequency tracking
component 216. Each
header is decoded. On the basis of the result of the decoding, we move on to
states 303, 304 or 305.
If an anomaly is detected in this state, receiver 1 returns to state 301.
In states 303 to 305, a residual frequency phase-shift estimation Afh is
computed on the basis of 90
symbols of the header and provided to frequency tracking component 216, for
example by means of
the Fitz algorithm. If a synchronization anomaly is detected in each of these
states, we return to
frame locking state 301.
In state 303, receiver 1 has determined that the frame is addressed to it
contains pilots. The decoding
of the header of the frame makes it possible to deduce its phase, the position
of the header of the
next frame and its phase. As described above, receiver 1 uses the phases from
the headers and the
pilots to perform a tracking and phase correction on each section of the
frame. After correction, a
CA 02625710 2011-05-09
-14-
quantification of the likelihoods of the bits carried by the symbols is
performed. Then, these bits are
decoded in an LDPC decoder with the parameters corresponding to the type of
frame, then the bits
obtained in a BCH code are decoded. An LDPC decoder with flexible inputs and a
BCH decoder
with hard inputs are preferably used.
In state 304, receiver 1 has determined that the frame is not addressed to it.
The decoding of the
header of the frame makes it possible to deduce its phase, the position of the
header of the next
frame and its phase. On the basis of the decoding of the next header, receiver
1 goes to one of sates
303 to 305.
In state 305, receiver 1 has determined that the frame is addressed to it and
does not contain a pilot.
The decoding of the header of the frame makes it possible to deduce its phase,
the position of the
header of the next frame and its phase. As described above, receiver 1 uses
the phases deduced in
order to perform a tracking and a phase correction on the body of the frame.
After correction, the
LDPC decoding and the BCH decoding are performed as described for state 303.
In state 301, receiver 1 does not initially know the position of the headers
in the digital stream.
Knowing that a header has a fixed number of 90 symbols, receiver 1 samples, in
a manner known
per se, 90 complex symbols of the digital stream received with 2 samples per
symbol. The sampled
symbols are placed in a FIFO. Then, tests are performed on all of these
symbols in order to
determine whether they correspond to a header. It is thus attempted to
determine whether the 26 first
symbols correspond to the sequence SOF and whether the 64 next symbols
correspond to the
remainder of the header.
To compensate for the modulation in it/2 BPSK of the transmission, the uneven
index symbols are
subjected to a 90 rotation. A logic descrambling is then performed: the 26
first symbols are
multiplied by the known sequence SOF and the 64 next symbols are subjected to
a descrambling
corresponding to the scrambling at the transmission. The symbols thus modified
are stored in a
buffer storage. These symbols will hereinafter be denoted zk=sk+nk,
With: nk being a transmission noise term,
+k4j)
sk=Eke(
CA 02625710 2011-05-09
-15-
t being the phase shift at the origin,
Af being the frequency shift,
Ek being the complex symbol at time k, corresponding to an authorized
constellation
(if Ek belongs to the field SOF of the header, 6k=1 and if Ek belongs to the
remainder
of the header, Ek= 1).
As the properties of the field SOF of a header and of the remainder of the
header are different, the
distinct correlation calculations are advantageously performed on the 26 first
bits and on the 64 next
bits. A correlation term related to the 26 first symbols can in particular be
determined by the
following formula:
26-k-1k4f
Rsof(k) I z +k z, = (26 - k)e' + n,(`soF)
i=0
with term wk being a noise term independent of the phase.
A correlation term related to the 64 next symbols can in particular be
determined by the following
formula:
(32) k-I
RMODCOD-TYPE(k)= I Zkq+r+kZ2kq+r
q=0 r=0
Due to the properties of the Reed-Muller code, the product E2kq+r+k - E2kq+r
is independent of r when k
is a power of 2, with 0<r<k. We thus easily arrive at the following relation:
RMODCOD-TYPE(k)=3 2 Eeikdf+ W~MODCOD-TYPE. )
with E= 1 and wk being a noise term independent of the phase.
A frequency shift Af estimation is performed on the basis of the correlation
terms above. This
estimation is performed on a large number of symbols, and has a very high
precision. The Reed-
Muller coding provides a code in which the symbols of which the product is
constant are determined
by a simple mathematical formula, which simplifies the determination of the
correlation level.
CA 02625710 2011-05-09
-16-
With a reduced number of computations, a low correlation level is thus
obtained.
Various correlation levels can thus be computed for various values of k.
The following is an example of a computation of the correlation level:
R(k)- I max Rsof. (k) + RAttDCOD-7'YPL (k)
58-k Rs01 (k) - RH)DcoD-77I1,(k)
Such a level of correlation makes it possible to overcome the uncertainty on
the respective signs of
RsOF(k) and RMODcoD-TYPE(k), with E being capable of having the value 1 or -1.
In addition, a
correlation level dependent both on known symbols and unknown symbols makes it
possible to
increase the reliability of the locking.
With a plurality of values of k, a plurality of correlation levels can thus be
computed and be
compared with respective thresholds in order to determine whether the f
symbols received
correspond to a header. The reliability of the locking can thus be increased.
Insofar as the correlation level(s) have not exceeded their locking threshold,
the locking process and
the computations of the correlation levels are reproduced with each new symbol
received.
A likelihood test is performed when decoding the header. This test can be
performed either on the
header that has completed the locking or on a subsequent header, so as to
respectively validate the
locking or validate the frequency synchronization.
For each unknown field to be decoded, the samples are processed as described
for the locking, the
modulation in it/2 BPSK of the transmission is compensated, and the uneven
index symbols are
subjected to a 90 rotation. A logic descrambling is then performed. If the
header is not the one that
generated the locking, the frequency shift error determined before is
corrected on the samples. The
corrected samples are denoted Zk. The samples z2k and Z2k+1 and (for samples
corresponding to the
MODCOD-TYPE field, i.e. for k between 26 and 89) correspond to the same phase
codeword in the
absence of a pilot, or phase-shifted by 180 in the presence of pilots.
CA 02625710 2011-05-09
-17-
A computation independent of the metrics corresponding to Z2k and z2k+I is
performed in order to
generate 32 complex metrics pairs corresponding to the 32 possible code words.
Below, yk (with
0<k<32) will be considered as vector z2 or vector z2k+1.
Then, the metrics are generated by the following iterative method. In a step
n, 25-" metrics are
formed Lo to L25-n and with a length of 2 .
In a step n+l, the metrics are grouped in pairs: Lo and LI, L2 and L3 ... .
Each metric pair thus obtained is replaced by a list of length 2"+', as shown
below:
PO /10 + P1 0
L2i =
,u2õ,u2õ_1 +,u'2õ-]
Li
f11o do - duo
L2i +l
PP2 -1 -P"" t
Thus, the first three iterations appear as follows in the example:
Y0 Y0 +Y] (Y0 + YI) + (Y2 + Y3) ((Yo + YI) + (Y2 + Y3)) + ((Y4 + Y5) + (Y6 +
Y7))
Y, Y0 - Y1 (Yo - Y,) + (Y2 - Y3 )
Y2 Y2 + Y3 (Y0 + Yi) - (Y2 + Y3 )
Y3 Y2 - Y3 (Y0 - YO - (Y2 - Y3 )
Y4 Y4 + Y5 (Y4 + Ys) + (Y6 + Y7) ((Y0 + Y,) + (Y2 + Y3)) - ((Y4 + Y5) + (Y6 +
Y7))
Y5 Y4 -Y5 (Y4 - Ys) + (Y6 - Y7 )
Y6 Y6 +Y7 (Y4 +Y5)-(Y6 +Y7)
Y7 Y6 - Y7 (Y4 - Y5) - (Y6 - Y7 )
step I step 2 step 3
CA 02625710 2011-05-09
-18-
At the end of the fifth iteration, we have a list of complex metrics of all of
the possible code words.
Note each list obtained by mk for Z2k and m'k for Z2k+1. It can be determined
that the most probable
code word will correspond to the value of k, for which Vr=Imkl2 + jm'kI2 has
the maximum value.
The likelihood level Vr is compared with a threshold. If this level is below a
likelihood threshold, it
is deduced that the synchronization of the frame is lost and that the locking
must be performed
again. The likelihood threshold used for the header serving for the locking is
greater than the
likelihood thresholds used for the headers of the next frames. Indeed, as the
likelihood level for the
first header is intended to confirm the locking, the tolerance at this level
must be reduced. During
the tracking, the tolerance at the likelihood levels of the headers may be
greater.
The TYPE field is then decoded. The value of the low-weight bit tO of the TYPE
field is determined
as follows:
- if Re(mk=m'*k)<O, then t0=1 and a value m=mk-m'k is defined;
- if Re(mk=m'*k)>O, then tO=O and a value m=mk+m'k is defined.
The value of the high weight bit tl of the TYPE field is determined as
follows:
- if Re(m.(*soF)<O, then tl=l; otherwise tl=O.
With:
1 25
sOF`- 26 Y, Zk
k=0
corresponding to the arithmetic mean of the SOF field.
This decoding thus takes into account the fact that the phase is unknown when
it is produced.
