JP2000252969A - Method for detecting symbol timing - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce the influence caused by eliminating a non-principal wave component by selecting a numerical series that has the least numerical amplitude among plural distributed numerical series and defining the timing containing the selected numerical series as the symbol timing. SOLUTION: The N sets of numerical series R0, R1...RN-1 which are stored in a two-dimensional register 107 of (N sets × F stages) are read out by N sets of arithmetic circuits 108 which calculate the rates Ti (i=0, 1...N-1) between two main waves showing the largest value and the second largest value of received signals and the absolute value of a correlative series for those numerical series respectively. A largest value retriever 109 retrieves the largest value of the rate Ti calculated between two main waves and the absolute value of the correlative series for each of series R0 to RN-1. Then (i) having the largest Ti is defined as i0 with a sample point where the remainder is i0 when N is divided by (n) defined as the symbol timing respectively, and i0 is outputted via an output terminal 110. Thus, the timing that contains a numerical series having the least numerical amplitude is defined as the symbol timing.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for detecting a symbol timing used for extracting a transmission signal from a received signal in a receiver used in digital communication.

[0002]

2. Description of the Related Art As the use of multimedia that integrates media such as characters, voices, images, and the like progresses, moving images are also transmitted in digital communication, and digital wireless communication for transmitting moving images is performed. Then, high-speed information transmission is required. In a receiver used in this digital wireless communication, it is important to accurately detect symbol timing in order to correctly extract a transmission signal including characters, voices, images, etc. from a received signal. The mobile radio communication used as one means has a problem in reception that it is particularly susceptible to delayed waves. Radio waves transmitted from the transmitter propagate through various paths and are received by the receiver with different delay times. In order to improve the effect of the delayed wave, a waveform equalizer is used in the receiver, but the performance of the waveform equalizer depends on the timing of detecting the symbol signal included in the received signal. Timing detection is required. As a method of detecting the symbol timing,
From the transmitting side, a known symbol sequence is arranged and transmitted at the beginning of each frame of the transmission signal, and at the receiving side, a correlation sequence between the received signal and the known symbol sequence is calculated, and the calculated correlation sequence is the maximum value. There is a symbol timing detection method using a certain point as a symbol timing.

A conventional symbol timing detection method will be described using the symbol timing detection circuit shown in FIG. In the following description, N is the number of samples per symbol, M is the number of symbols in a known symbol sequence, and F is the number of symbols per frame. A received signal received by the receiver is processed, and a required received signal is input to the synchronous detector 402 via the input terminal 401. The synchronous detector 402 performs synchronous detection on the received signal input via the input terminal 401 and performs baseband signal 4
03 is extracted, and the baseband signal 403 extracted by synchronous detection is output to the low-pass filter 405. The low-pass filter 405 removes the high-frequency component of the baseband signal 403 extracted by synchronous detection input from the synchronous detector 402 and outputs the baseband signal from which the high-frequency component has been removed to the A / D converter 406. . A / D converter 4
06 is the baseband signal from which the high-frequency component input from the low-pass filter 405 has been removed, that is,
The digital signal is sampled and converted into a digital signal at every N time intervals, and the converted digital signal is output to the correlation operation circuit 407. In the following description, "1/3 of the symbol interval"
"Sampling every N time intervals" is called "N times oversampling per symbol".

[0004] The correlation operation circuit 407 includes an A / D converter 40.
6 and a separately input known symbol sequence {u} of length M symbols.
(0), u (1),..., U (M−1)} are calculated, and the calculated correlation sequence {r (n)} is calculated.
Is output to the absolute value calculation circuit 408. Absolute value calculation circuit 4
08 calculates the absolute value {| r (n) |} of the correlation sequence {r (n)} input from the correlation operation circuit 407 and sends the calculated absolute value {| r (n) |} to the register 409. Output.

Here, a detailed block diagram of the correlation operation circuit 407 will be described with reference to FIG. Correlation operation circuit 407
Are the M-1 cascaded delay elements 302 connected to the input terminal 301, the input side of the first-stage delay element 302 of the M-1 cascaded delay elements 302, and M One multiplication input is connected to each output side of the -1 stage delay element 302, and M multiplication circuits 303 each connected so that the known symbol sequence {u} is input to the other multiplication input; A configuration in which M-1 addition circuits 304 are connected so as to successively integrate the output signals of the multiplication circuits 303, and the integration output of the last addition circuit 304 is connected to an output terminal 305. It has become.

A digital signal input from an A / D converter 406 (see FIG. 4) to an input terminal 301 is shifted while being delayed for a required time by an M-1 stage delay element 302 connected in cascade. Each digital signal delayed for a required time by the delay element 302 of the (M-1) -th stage is multiplied by the second to M-th multiplication circuits 3 of the M multiplication circuits 303.
03 (a digital signal that has not been delayed from the input terminal 301 is input to the first multiplication circuit 303). This digital signal is supplied to an A / D converter 406
Are oversampled N times per symbol, and each of the delay elements 302 in the (M-1) -th stage delays by N samples, that is, by one symbol.

Here, in order to set the signal sequence input from the input terminal 301 to {y} and to set the output signal of the (M−1) th stage delay element 302 at time n to y (n), the input terminal 301 The input signal to be input is y (n + (M-1)
N). In this state, the input signal ｛y input to the input terminal 301 at a certain sample timing
(N), y (n + N), ... y (n + (M-1)
Assuming that a correlation sequence between (N)} and the known symbol sequence {u} is calculated, input signals {y (n + 1), y (n + N + 1),.
A correlation sequence between (M−1) N + 1)} and the known symbol sequence {u} is calculated. Note that the correlation operation starts from the point in time when the digital signal first input from the input terminal 301 is successively shifted through the delay element 302 of the (M-1) th stage and output from the delay element 302 of the (M-1) th stage. , M-1
The correlation sequence at the time when y (n) is output to the delay element 302 at the stage is r (n). This correlation sequence r (n) is
A known symbol sequence ｛u (0), u (1),... U (M−
1) When a series of patterns close to｝ is input, the absolute value gives the maximum value.

[0008] Further, in FIG.
The register 409, to which the absolute value {| r (n) |} calculated from Eq. 8 is input, stores the number of samples FN per frame (F is the number of symbols per frame, and N is the number of samples per symbol). ). The register 409 stores the absolute value of the correlation sequence for one frame ｛| r
(0) |, | r (1) |,... | R (FN-1) |｝ are stored. The maximum value searcher 410 reads 1
The absolute value of the correlation sequence for the frame is read, the maximum value is searched, and the point giving the maximum value is defined as n ₀ . The remainder n ₀ mod N is calculated by dividing the point n ₀ giving the maximum value by the number N of samples per symbol, and the remainder obtained by dividing n ₀ by N is n _0.
A sample point equal to modN is set as a symbol timing, and n ₀ modN is output to the output terminal 411.

The correlation sequence of the received signal shown in FIG. 5 has a length of 16 symbols (M = 16) at the beginning of 192 symbols (F = 192) as shown in FIG. A known symbol sequence as a reference is arranged, and signal points are arranged and transmitted using π / 4 shift QPSK.
This is an example of a correlation sequence (N = 4 samples / symbol) when one delay wave (D / U (ratio of required signal to unnecessary signal) = 0 dB) with a delay time of one symbol period is added. FIG. 6 is an enlarged view of the correlation sequence of the received signal shown in FIG. 5 near the maximum value. As shown in FIGS. 5 and 6, the absolute value | r (n) | of the correlation sequence becomes the maximum value when the number of samples is n = 60, so that n ₀ = 60 is selected by the maximum value searcher 410, and n ₀ mod N = 60 mod 4 = 0, and n
Points at which the remainder of dividing by 4 is 0 (n = 0, 4,
8,...) Is set as the symbol timing, and 0 is output to the output terminal 411.

In the symbol timing detection method, the received signal is generally band-limited, so that the transmitted signal sequence is considered to be a random signal sequence, and if there is no delayed wave, the received signal and the known symbol The correlation sequence r (n) with the sequence is like the correlation sequence 801 between the direct wave and the known symbol sequence shown in FIG. Assuming that a delay time is one symbol, D / U = 0 dB, and a delayed wave having the same phase as the direct wave exists, a correlation sequence r (n) between the delayed wave and the known symbol sequence is represented by a delay indicated by a broken line in FIG. It becomes like a correlation sequence 802 between a wave and a known symbol sequence. The correlation sequence r (n) between the received signal obtained by combining the direct wave and the delayed wave and the known symbol sequence is a correlation sequence 801 between the direct wave and the known symbol sequence and a correlation sequence 802 between the delayed wave and the known symbol sequence. And a correlation sequence 803 between the combined received signal and the known symbol sequence, which is the sum of

FIG. 9A shows an absolute value | r (n) | 901 obtained by calculating the absolute value of the correlation sequence 803 between the synthesized received signal and the known symbol sequence. In the symbol timing detection method according to the related art, an absolute value 901 of a correlation sequence between a combined received signal and a known symbol sequence is used.
The point n _MAX 902 where is the maximum value is detected, and FIG.
The correlation sequence 903 extracted at the symbol period based on the point n _MAX 902 having the maximum value shown in FIG.

The waveform equalizer for improving the influence of the delay wave can be roughly classified into two types, a decision feedback equalizer (DFE) and a maximum likelihood sequence estimator (Viterbi equalizer). Here, the problem of the symbol timing detection method according to the related art when the latter maximum likelihood sequence estimator is used as an equalizer will be considered. FIG. 10 is a block diagram showing a concept of a maximum likelihood sequence estimator as a virtual circuit. The propagation path estimator 1006 obtains an estimated value h ′ (t) of the propagation path characteristic h (t) of the transmission path 1002 from the received signal y (t) 1005 transmitted through the transmission path 1002. There suitable correlation sequence {a _n}, and the transmission path characteristic h (t) of the estimated value h 'replicas through estimated CIR 1007 (t) y'
Calculating _{a; ({a n} t)} 1008. Received signal y
(T) 1005 and the replica _{y '(t; {a n} }) 100
8 is compared by a comparison circuit 1009, and the received signal y (t) 1
005 and replica _{y '(t; {a n} }) between the signals of 1008 distance d ({a _n}) correlation sequence which is minimum {a _n}
Is searched by the detection circuit 1010. Signal distance d (｛a
_In order to search for a correlation sequence that minimizes _n {), it is necessary to search for all of the correlation sequences {a _n }, but the search can be performed efficiently by using the Viterbi algorithm.

[0013]

In the symbol timing detection method according to the prior art, when the symbol timing is determined, the non-main component is removed infinitely, as shown as a correlation sequence 903 in FIG. 9B. When you do
There is a problem that the effect of removing the non-mainstream wave component is great. The present invention solves the above problem, and determines the symbol timing so that the ratio of two main waves in the correlation sequence is maximized, as shown as a correlation sequence 904 in FIG. An object of the present invention is to provide a symbol timing detection method in which the influence of the removal is reduced.

[0014]

In order to achieve the above object, a symbol timing detecting method according to the present invention is a symbol timing detecting method used for a receiver used in digital communication, wherein each symbol comprises a plurality of symbols. The receiver receives a transmission signal having a frame structure in which a known symbol sequence having a length of M symbols is arranged at the head of the reception signal as a reception signal, and synchronously detects the reception signal to form a baseband signal, and then N times per symbol. (N is a natural number) Oversampling into a digital signal, calculating a correlation of the digital signal with the known symbol sequence, calculating a correlation sequence, and detecting a symbol timing from an absolute value of the correlation sequence. In the method, the absolute value of the correlation sequence is calculated as the number of samples per symbol N
In a cycle of N, the set is divided into N sets of numerical sequences, and the sorted N
Calculating a ratio of two absolute values having a large absolute value to the absolute value of the correlation sequence for each of the set of numerical sequences;
The correlation sequence having the largest ratio is searched, and the timing at which the searched correlation sequence exists is used as the symbol timing.

More specifically, in the symbol timing detection method according to the present invention, the ratio of two absolute values having a large absolute value to the absolute value of the correlation sequence is calculated for each of the sorted N sets of numerical sequences. The maximum value of the absolute value is calculated for each of the allocated N sets of numerical sequences, and the maximum value of the absolute value and the second largest value within the range of one symbol before and after the sample point giving the maximum value of the absolute value are calculated. Calculating the ratio of the absolute value of the maximum value and the second largest value of the absolute value to the absolute value of the correlation sequence, searching for the correlation sequence having the highest ratio from each of the N sets of numerical sequences; The timing at which the calculated correlation sequence exists is used as the symbol timing.

In the symbol timing detecting method according to the present invention, the maximum absolute value and the ratio of the second largest value to the absolute value of the correlation sequence are calculated before and after the sample point giving the maximum absolute value. (M-1) Within the range of the symbol, the maximum value of the absolute value and the ratio of the second largest value to the absolute value of the correlation sequence are calculated, and the correlation in which the ratio is the largest from each of the N sets of numerical sequences is calculated. A sequence is searched, and the timing at which the searched correlation sequence exists is used as the symbol timing.

The symbol timing detection method according to the present invention is a symbol timing detection method used for a receiver used for digital communication, wherein a symbol having a predetermined number of bits has a length at the beginning of each frame. M
A transmission signal having a frame configuration in which a known symbol sequence of symbols is arranged is received and used as a reception signal, and the reception signal is detected by a synchronous detector to generate a baseband signal, and the baseband signal is converted into one by an A / D converter. Oversampling N times (N is a natural number) per symbol into a digital signal,
The digital signal is correlated with the known symbol sequence by a correlation calculation circuit to calculate a correlation sequence, and the correlation sequence is calculated by an absolute value calculation circuit to calculate the absolute value of the correlation sequence. Is stored in a register, at least one frame of the absolute value of the correlation sequence is read out from the register, the maximum value is searched by a maximum value searcher, and a sample point giving the maximum value is determined by symbol timing. In the symbol timing detection method,
The absolute values of the correlation sequences calculated by the absolute value calculation circuit are sequentially circulated by a circulation switcher and distributed to N sets of numerical sequences at a period of N samples per symbol, and the N sets of the allocated sets are arranged. Is stored in an N-set × F-stage two-dimensional register, and is read out from the N-set × F-stage two-dimensional register to N-set arithmetic circuits connected to the respective sets.
The maximum value of the absolute value and the ratio of the second largest value to the absolute value of the correlation sequence are calculated for each of the N sets of numerical sequences. The correlation sequence with the largest ratio is searched, and the timing at which the searched correlation sequence with the largest ratio exists is defined as the symbol timing.

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Before describing an embodiment of the present invention, a basic description will first be given of a symbol timing detecting method of the present invention. The basic block diagram for detecting the symbol timing shown in FIG. 1 is used. From the transmitting side, as shown in FIG. 7, for each frame, a known symbol sequence と す る u (k) as a reference for a length M symbol
| K = 0, 1,..., M−1} are transmitted at the beginning. On the receiving side, the received signal is processed as required, and a known symbol sequence {u (k) | k = 0, 1,... (Although not shown in FIG. 1, in the A / D converter, N
Times over-sampled), the first delay element 103 of the M-1 N-sample delay elements 103 and the M multiplication circuits 104 connected in cascade through the input terminal 101 of the correlation operation circuit 102. Is input to the first multiplication circuit 104. The M-stage N-sample delay element 103 connected in cascade gives a delay of N samples to the input digital signal oversampled N times per symbol. Assuming that the signal sequence y (n) is input to 103, the output of the first-stage delay element 103 is y (n-N) N samples before, and the output of the second-stage delay element 103 is 2
The output of y (n−2N) before N samples and the output of the last M−1 stage delay element 103 are y (n) before (M−1) N samples.
− (M−1) N).

In the following description, in order to set the output of the last (M-1) th delay element 103 to y (n), as shown in FIG. To y
(N + (M-1) N) and output is y (n + (M-2)
N), the output of the second delay element 103 is y (n + (M−
3) N) (the same applies hereinafter). First delay element 103
, And the output of each delay element 103 are connected to a multiplication circuit 104, and the product of each of the coefficients of the known symbol sequence {u} separately input is calculated.
-1 adder circuits 105, and in each of the M-1 adder circuits 105, the product of the output of each delay element 103 and each coefficient of the known symbol sequence {u} is integrated, and the input signal y (n ) And the known symbol sequence {u} are calculated as a correlation sequence r (n). This calculation is represented by the following equation (1).

(Equation 1)

In the correlation operation circuit 102, the correlation sequence r (n) calculated by the equation (1) is used as an absolute value operation circuit 111.
And the absolute value | r calculated by the absolute value calculation circuit 111
(N) | is set to N series (the number of sets matched with the cycle N) by the circulation switching unit 106 at N cycles. R ₀ = ｛| r (0) |, | r (N) |,... | r ((F−1) N) |｝, R ₁ = {| r (1) |, | r (N + 1) |,... | r ((F−1) N + 1) |｝,. ... R _N-1 = {| r (N-1) |, | r (2N-1), ... | r (FN-1) |}, and N sets × F stages (per frame) Is stored in the two-dimensional register 107 of the number of stages matched with the number of symbols F of (1). For example, if N = 4, n =
If it is 5, the second stage of the first set of the two-dimensional register 107 has n
If = 12, they are stored in the third row of the fourth set. For each of the N sets of numerical sequences R ₀ , R ₁ ,... R _N−1 distributed to the two-dimensional register 107 by the circulation switch 106, the ratio of the two main force waves is calculated.

The calculation of the ratio in the case of two main force waves is shown in FIG.
2 will be described. Each absolute value 1201 of the correlation sequence shown in FIG. 12 is an absolute value of the correlation sequence obtained by the circulation switch 106 (see FIG. 1) sorting the absolute value | r (n) | input from the absolute value calculation circuit 111. . First, the circulation switch 10
6, the correlation sequence r (n) (n =
0, 1,... FN-1) is the maximum sample point n
Within a range 1203 of one symbol before and after _max 1202, the maximum value r _{max, i} 1204 of each absolute value 1201 of the correlation sequence
And the second largest value r _{2nd, i} 1205 (two main waves) are searched, and the sample points where these maximum value and the second largest value exist are defined as m _{max, i} and m _{2nd, i} , respectively. . Next, the sum S _i of the absolute value 1201 of the correlation sequence is calculated in a range 1206 of (M−1) symbols before and after m _{max, i} , and further, the sum (r
_{max, i} + r _{2nd, i} ) and the sum S of the absolute value 1201 of the correlation sequence
The ratio of the _i, to calculate the _{T i = (r max, i} + r 2nd, i) / S i. After calculating the ratio T _i between the two main waves and the absolute value 1201 of the correlation sequence for each of the numerical sequences R ₀ , R ₁ ,... R _N−1 , the maximum value of this ratio T _i is set to the maximum value. searching by the search unit 109 (see FIG. 1), the i at that time and i _0, the sample point remainder of dividing n by n becomes i ₀ to the symbol timing, and outputs the i ₀ from the output terminal 110.

Next, an embodiment of the symbol timing detecting method according to the present invention based on the above basic principle will be described. FIG. 2 shows an embodiment of a symbol timing detection circuit used in a receiver and using the symbol timing detection method of the present invention. As shown in FIG. 7, the transmitter transmits a transmission signal having a frame structure in which a reference known symbol sequence is arranged at the head for each frame. In the receiver, the received signal received is subjected to required processing to become a received signal having a frame configuration in which a known symbol sequence (constituted by M symbols) serving as a reference is placed at the head of each frame (constituted by F symbols). Are input to a symbol timing detection circuit using the symbol timing detection method of the present invention shown in FIG.

The received signal is input to the synchronous detector 202 via the input terminal 201. The synchronous detector 202 synchronously detects and extracts the baseband signal 203 from the received signal input via the input terminal 201, and synchronously detects and extracts the extracted baseband signal 203 from the low-pass filter 20.
Output to 5 The low-pass filter 205 removes the high-frequency component of the baseband signal 203 extracted by synchronous detection input from the synchronous detector 202 and outputs the baseband signal from which the high-frequency component has been removed to the A / D converter 206. . The A / D converter 206 samples the baseband signal from which the high-frequency component input from the low-pass filter 205 has been removed at every 1 / N of a symbol interval, converts it into a digital signal, and converts the converted digital signal. Is output to the correlation operation circuit 102.

The correlation operation circuit 102 includes an A / D converter 20
6 and a known symbol sequence {u} separately used as a reference for a length M symbol that is input.
(0), u (1),..., U (M−1)} are calculated, and the calculated correlation sequence {r (n)} is calculated.
Is output to an absolute value calculation circuit (for example, a squaring circuit) 111. The absolute value calculation circuit 111 calculates the absolute value {| r (n) of the correlation sequence {r (n)} input from the correlation calculation circuit 102.
Is calculated, and the calculated absolute value {| r (n) |} is output to the circulation switch 106. The circulation switching unit 106 converts the absolute value {| r (n) |} input from the absolute value calculation circuit 111 into N cycles and modulo n by N to obtain N sets (the number of sets matched with the cycle N). ), And outputs to the two-dimensional register 107 of N sets × F stages (the number of stages in one set with the number of stages matched with the number of symbols F per frame). The two-dimensional register 107 stores a numerical sequence R ₀ = {r (0), r (N),... R ((F−1) N)}, which is distributed to N sets input from the circulation switch 106. R ₁ = {r (1), r (N + 1),... R ((F−1) N + 1)},..., R _N−1 = {r (N−1), r (2N−) 1),... R (FN-1)} are stored.

The N sets of numerical sequences R ₀ , R ₁ ,..., R _N-1 stored in the N sets × F-stage two-dimensional register 107 are N
The data is read out to N sets of arithmetic circuits 108 connected to each set of the set × F-stage two-dimensional registers 107. N sets of arithmetic circuits 108 read out from the two-dimensional register 107 and have N sets of numerical sequences R ₀ , R ₁ ,.
For each of _N-1 , the ratio Ti (i = 0, 1,... N-1) between the maximum value of the received signal and the two main force waves indicating the second largest value and the absolute value of the correlation sequence is calculated. I do. The calculation of the ratio Ti between the two main waves and the absolute value of the correlation sequence is shown in FIG.
, The correlation sequence r (n) (n = 0, 1,... F) before being allocated to N sets by the circulation switch 106.
In a range 1203 of one symbol before and after the point n _max 1202 at which N−1) becomes the maximum, the maximum value r _{max, i} 1204 which is the main wave in the absolute value 1201 of the correlation sequence and the second largest value r _{2nd, i} 1205, and the positions where these maximum value and the second largest value exist are defined as m _{max, i} , m
_{2nd, i} . Next, before and after centering on m _{max, i} (M−
1) The absolute value 120 of the correlation sequence in the symbol range 1206
Calculating a first sum S _i, further major wave 2 wave 1204,1
The ratio Ti = (r _{max, i} + r _{2nd, i} ) between the sum (r _{max, i} + r _{2nd, i} ) of 205 and the sum S _i of the absolute value 1201 of the correlation sequence
/ _Si is calculated. N sets of each numerical series R ₀ , R ₁ ,.
The maximum value search unit 109 searches for the maximum value of the ratio Ti between the two main waves calculated for each of R _N-1 and the absolute value 1201 of the correlation sequence, and sets i at which Ti is maximum to i ₀ , The sampling point at which the remainder of dividing n by N becomes i ₀ is set as the symbol timing, and i ₀ is output from the output terminal 110.

The diagrams of the correlation sequences shown in FIGS. 5 and 6 show the number of symbols per frame F = 192 as described above.
Symbol, known symbol length M = 16 symbols, 1
This is an example of a model of two main force waves each of which is oversampled N = 4 times per symbol, has the same amplitude and phase for the delayed wave and the direct wave, and has a delay time of one symbol. By the circulation switch 106, four correlation sequences R in N = 4 periods
If distributed to ₀ to R _3, the distribution shown by the black point in FIG. 13-16. The broken line in the figure is the correlation sequence shown in FIG. 6 before being sorted. R ₀ of the correlation sequence shown in FIG. 13 has a remainder of 0 obtained by dividing n by N = 4.
(N) | is a correlation sequence extracted (the same applies hereinafter).

A range 1203 of one symbol before and after a point n _max 1202 where the correlation sequence before being sorted has the maximum value.
When two main waves of the correlation sequence shown in FIGS. 13 to 16 are searched, r _{max, 0} = 57.36 (m _{max, 0} = 60) for the correlation sequence R ₀ shown in FIG. ), R
_{2nd, 0} = 7.56 (m _{2nd, 0} = 64), and r _{max, 1} = 51.65 (m _{max, 1} = 6 for the correlation sequence R _{1 in} FIG. 14)
1), r _{2nd, 1} = 19.10 (m _{2nd, 1} = 57), FIG.
For the correlation sequence R ₂ of 5, r _{max, 2} = 36.62 (m
_{max, 2} = 62), r _{2nd, 2} = 35.78 (m _{2nd, 2} = 5)
_{8), r max, 3 =} 4 for correlation sequence R ₃ in FIG. 16
9.82 (m _{max, 3} = 59), r _{2nd, 3} = 18.21
(M _{2nd, 3} = 63).

Next, m _{max, i} (i = 0,1,2,3)
13 to 13 in the range of M-1 = 9 symbols before and after
When the sum of the correlation sequences shown in FIG. 16 is calculated, S ₀ = 575.5 for the correlation sequence R _{0 in} FIG. 13, S ₁ = 577.6 for the correlation sequence R _{1 in} FIG. 14, and the correlation sequence R in FIG.
₂ , S ₂ = 572.9, and the correlation sequence R _{3 in} FIG.
Is S ₃ = 583.0. Therefore, as shown in FIG. 17, the ratio Ti of the two main waves is T ₀ = 0.1128 for the correlation sequence R _{0 in} FIG. 13, T ₁ = 0.1225 for the correlation sequence R _{1 in} FIG. T ₂ = .1264 for correlation sequence R ₂ in FIG. 15, T ₃ = 0.1167 next for correlation sequence R ₃ in FIG. 16, T ₂
Is the maximum, so the sampling point (n = 2, 6, 10,...) Where the remainder of dividing n by 4 becomes 2 is set as the symbol timing, and 2 is output from the output terminal 110.

When a maximum compensation delay time is one symbol and a Viterbi equalizer that removes non-main wave except the two main wave is used, as described above, as the ratio Ti of the two main wave becomes larger, Since the ratio of the non-mainstream wave is small, the influence of the removal of the non-mainstream wave component is reduced. The symbol timings detected by the prior art are n = 0, 4, 8,.
The remainder when divided by 0 is the same as the symbol timing when T ₀ is selected. On the other hand, in the above-described example according to the symbol timing detection method of the present invention, since T ₂ (maximum value) larger than T ₀ is selected, the influence of removal of the non-mainstream wave component is smaller than that of the symbol timing detected by the conventional symbol timing detection method. Are detected with less symbol timing. According to the symbol timing detection method of the present invention, the symbol timing at which the ratio of the two main wave components is maximized is detected, and when using a Viterbi equalizer that removes non-main component wave components other than the two main force components, The effect of component removal is minimal.

[0030]

According to the present invention, the symbol timing is determined so that the ratio of the two main waves in the correlation sequence is maximized, and the influence of the removal of the non-main wave component is reduced. A detection method can be provided.

[Brief description of the drawings]

FIG. 1 is a block diagram of a basic part for detecting a symbol timing by a symbol timing detection method according to the present invention.

FIG. 2 is a block diagram of a symbol timing detection circuit using the symbol timing detection method of the present invention.

FIG. 3 is a detailed block diagram of a correlation operation circuit.

FIG. 4 is a block diagram of a symbol timing detection circuit using a conventional symbol timing detection method.

FIG. 5 shows a correlation sequence of a received signal (two main wave models).

FIG. 6 is an enlarged view of the vicinity of the maximum value of the correlation sequence of the received signal shown in FIG. 5;

FIG. 7 is a diagram showing a symbol configuration of one frame of a transmission signal.

FIG. 8 shows an ideal correlation sequence of a received signal (two main wave models: one symbol delay).

FIG. 9 is a view for explaining a difference in a correlation sequence depending on a sample timing (a main wave two-wave model: one delay symbol).

FIG. 10 is a block diagram showing a concept of a maximum likelihood sequence estimator as a virtual circuit.

FIG. 11 is a view for explaining the limitation of the maximum compensation delay time of the equalizer and the removal of non-mainstream waves.

FIG. 12 is a view for explaining the calculation of the ratio of two main force waves.

FIG. 13 is a diagram showing a correlation sequence R ₀ obtained by allocating the correlation sequence of FIG. 6 by a circulation switch.

FIG. 14 is a diagram showing a correlation sequence R ₁ obtained by allocating the correlation sequence of FIG. 6 by a circulation switch.

FIG. 15 is a diagram showing a correlation sequence R ₂ obtained by allocating the correlation sequence of FIG. 6 by a circulation switch.

FIG. 16 is a diagram showing a correlation sequence R ₃ obtained by allocating the correlation sequence of FIG. 6 by a circulation switch.

FIG. 17 shows correlation sequences R ₀ , R ₁ , R ₂ ,
Shows the percentage Ti main wave 2 wave calculated for R _3.

[Explanation of symbols]

101, 201, 301, 401 ... input terminal, 102,
407: correlation operation circuit, 103, 302 ... delay element, 1
04, 303 ... multiplication circuit, 105, 304 ... addition circuit,
106: circulation switching device, 107: two-dimensional register, 108
... arithmetic circuit, 109, 410 ... maximum value searcher, 110,
305, 411: output terminal, 111, 408: absolute value calculation circuit, 202, 402: synchronous detector, 205, 405
... low-pass filters, 206 and 406 ... A / D converters
409 ... register.

Claims (5)

[Claims] 1. A method for detecting a symbol timing used in a receiver used for digital communication, comprising: transmitting a transmission signal having a configuration in which a symbol sequence known on the receiver side is arranged at a predetermined position in a frame. And calculates a correlation sequence between the received signal and the known symbol sequence. The correlation sequence is divided into a plurality of numerical sequences in a symbol cycle, and from among the plurality of distributed numerical sequences. A method for selecting a numerical sequence having the smallest numerical value amplitude, and using a timing at which the selected numerical sequence exists as a symbol timing. 2. A symbol timing detection method used in a receiver used for digital communication, comprising: a transmission signal having a frame configuration in which a known symbol sequence having a length of M symbols is arranged at the head of each frame including a plurality of symbols. The received signal is received by the receiver to obtain a received signal. The received signal is synchronously detected to be a baseband signal, and then oversampled N times (N is a natural number) per symbol to obtain a digital signal. The known symbol of the digital signal In a symbol timing detection method for calculating a correlation sequence by calculating a correlation with a sequence and detecting a symbol timing from an absolute value of the correlation sequence, the absolute value of the correlation sequence is calculated at a period of the number N of samples per symbol. Sorted into N sets of numerical series, and the absolute value of each of the sorted N sets of numerical series is large. A symbol sequence for calculating a ratio of two absolute values to the absolute value of the correlation sequence, searching for a correlation sequence having the largest ratio, and setting a timing at which the searched correlation sequence exists as a symbol timing. Timing detection method. 3. The symbol timing detection method according to claim 1, wherein the ratio of two absolute values having a large absolute value to the absolute value of the correlation sequence is calculated for each of the allocated N sets of numerical sequences. The maximum value of the absolute value is calculated for each of the N sets of numerical sequences, and the maximum value of the absolute value and the second largest value are searched within one symbol before and after the sample point that gives the maximum value of the absolute value. Calculating the proportion of the absolute value of the maximum value and the second largest value of the absolute value to the absolute value of the correlation sequence, searching for the correlation sequence having the largest ratio from each of the N sets of numerical sequences, A symbol timing detection method, wherein timing at which a sequence exists is used as symbol timing. 4. The symbol timing detecting method according to claim 2, wherein calculating the maximum value of the absolute value and the ratio of the second largest value to the absolute value of the correlation sequence comprises: Before and after (M-
1) Calculate the ratio of the maximum absolute value and the second largest value to the absolute value of the correlation sequence within the range of the symbol, and search the N sets of numerical sequences for the correlation sequence with the highest ratio. A timing at which the searched correlation sequence exists is used as a symbol timing. 5. A method for detecting a symbol timing used in a receiver used for digital communication, wherein a known symbol sequence having a length of M symbols is arranged at the head of each frame composed of a plurality of symbols of data having a predetermined number of bits. A transmission signal having a frame configuration is received and converted into a reception signal. The reception signal is detected by a synchronous detector to generate a baseband signal. The baseband signal is N times per symbol by an A / D converter (N is a natural number). ) Oversampling into a digital signal, calculating the correlation of the digital signal with the known symbol sequence in a correlation operation circuit to calculate a correlation sequence, and calculating the correlation sequence in an absolute value operation circuit to calculate the correlation sequence. Calculating an absolute value; storing at least one frame of the absolute value of the correlation sequence in a register; Also reads one frame, searches for a maximum value with a maximum value searcher, and uses a sample point giving the maximum value as a symbol timing, wherein the absolute value of the correlation sequence calculated by the absolute value calculation circuit is calculated. Sequentially circulated by the circulation switcher and distributed to N sets of numerical sequences at a period of N samples per symbol, and the allocated N sets of numerical sequences are stored in N sets × F-stage two-dimensional registers. The N sets of arithmetic circuits connected to each set are read out from the N sets × F stages of two-dimensional registers, and the maximum value and the second largest value of the absolute value are obtained for each of the N sets of numerical sequences. Calculates the percentage of the absolute value of the correlation sequence in the absolute value, and searches the maximum value searcher for the correlation sequence in which the ratio is the largest from the respective ratios of the N sets of numerical sequences. A symbol timing detection method, wherein a timing at which a correlation sequence having the largest combination exists is defined as a symbol timing.