JPS6238049A - Unique word detector - Google Patents

Abstract

PURPOSE:To obtain an excellent unique word detecting characteristic by using a soft deciding data for an input reception signal, calculating the multi-dimension Euclid distance to a unique (UW) word pattern while keeping the soft deciding data as it is so as to detect the UW. CONSTITUTION:Soft dividing reception signals PR51, QR52 demodulated by a soft diciding demodulator are inputted to the shift register 57 of correlation devices P53, Q54 at every clock respectively. The difference between stored UW patterns P55, Q56 and the content of the shift register 57 is calculated by a clock subtractor 58 in the correlation devices P53, Q54, the result is squared by a square device 64 and inputted to an adder 59. The adder 59 totalizes the squared samples and inputs the total sum SIGMA to a comparator 60. A permissible Euclid distance epsilon is set in advance in the comparator 60 and when the relation of SIGMA<=epsilon<2> exists, a pulse is outputted and inputted to a AND circuit 62. When the pulse is outputted while a UW detection gate signal forecasting the arrival time of the UW in advance is enabled in a terminal station, the pulse passes through the AND circuit 62 and is outputted as a UW detection pulse 63.

Description

[Detailed Description of the Invention] [Industrial Application Field] This invention is applicable to the time division multiple access (TDMA) system of satellite communication and the Single Channel Per Carri system.
The present invention relates to a unique word prompt detector for use in frame synchronization of digital signal transmission in the er (5CPC) system and the like.

[Conventional technology]

Figure 2 shows an example of the configuration of a conventional unique word detector. In the figure, (1) is a hard decision demodulator (not shown) that
or the received signal Pa demodulated to 1, (2) the received signal Qi demodulated to 0 or 1 by a hard decision demodulator, (3) the correlator P storing the W pattern P, (4) the so-called pattern Correlator Q that stores Q, (5) is stored J pattern P, (6)
is the storage pattern Q, (7) is the (M/2) stage shift register, (8) is the exclusive OR circuit (Ex-OR), (9
) is an adder, and αO is a comparator 1. which compares with a certain value ε and outputs a pulse when it is smaller than or equal to ε. Qυ
is a door detection gate signal, (2) is an AND circuit, and α3 is an output LIW detection pulse signal.

Next, the operation will be explained. The received signal Fil (s and QR(2) demodulated to O or l by the hard decision demodulator)
Shift registers (7) of correlators P(3) and Q(4), respectively.
) is input every clock. Correlators P(3) and Q(4)
, the storage source patterns P(5) and Q(6) and the contents of the shift register (7) are Ex-OR(s
), and if the respective bits match, 0 is output; if they do not match, 1 is output. In the adder circuit (9), each Ex
- Count the number of 1s in the output of OR (8), that is, the number of mismatched bits, and calculate the total number of mismatched bits Σ to the comparator 0.
Compare with value C which is 0. Then, if Σ≦ε, a pulse is output. Predict the arrival time of 4 in advance at the terminal station■
If this pulse is output when the detection gate signal is enabled, it passes through the AND circuit 0o and is output as the door detection pulse αj.

[Problem that the invention seeks to solve]

As described above, the conventional J detector has already hard-determined the input received signal P and the parrot to be 0 or l, so it is susceptible to reception noise and detects an opening at the time when it should detect a < or a parrot. There were problems such as an increase in the probability of the so-called failure to open UW.

This invention was made to solve the above-mentioned problems, and aims to provide a door detector that can reduce the detection probability.

[Means for solving problems]

In the multidimensional soft-decision UW door detector according to the present invention, the output signal of the soft-decision demodulator is used as the input signal, the shift register of the correlator is a shift register having a number of bits sufficient to accommodate this soft-decision data, and the storage pattern is , which has a size corresponding to this soft decision data.

[Effect]

In the multidimensional soft decision detector of the present invention, the received input signal is soft decision data, and the multidimensional Euclidean distance between the soft decision data and the penalty pattern is calculated.

[Embodiment J of the Invention An embodiment of the invention will be described below with reference to the drawings. 1st
In the figure, hυ is a soft-decision received signal pR demodulated by a soft-decision demodulator (not shown), iz is a soft-decision received signal QR demodulated by a soft-decision demodulator, and phantom is a correlator that stores a pattern P. P, i41 is a correlator Q that stores the pattern Q,
arp is a storage pattern P having a magnitude corresponding to the soft decision reception signal PR, M is a storage pattern Q having a magnitude corresponding to the soft decision reception signal Qa, and t57) is a storage pattern P having a magnitude corresponding to the soft decision reception signal PR. The shift register has the number of bits that can be accommodated, the subtractor is the adder, the figure shows the comparator 2, IO is the door detection gate signal, the bag is the AND plane, the cocoon is the output j detection pulse signal, - is a square circuit.The operation will be explained next. The soft-decision received signals P-υ and Q, @ demodulated by the soft-decision demodulator are input to the shift registers of the correlators PM and Qr)4I, respectively, every clock. In the correlator pH and Qi41, the storage pattern P
The difference between M and Q- and the contents of the shift register .eta. is calculated in each tallock subtracter, then squared in a squarer and input to an adder. In the adder, the values of each squared sample are summed and the sum Σ is input to the comparator. In comparator 1, the allowable Euclidean distance ε
is set. Here, if Σ≦62, a pulse is output and input to the AND circuit bag. If this pulse is output when the door detection gate signal that predicts the arrival time of the UW is enabled at the terminal station, it passes through the piece circuit and is output as a door detection pulse.

Next, an analysis will be performed to explain that the present invention has better characteristics than the conventional method. It is assumed that the opening is made up of M bits. In practice, the W length M is often chosen as an even number, and here, M is analyzed as an even number.

The energy per bit of the transmitted signal is assumed to be E8.

Now, if there are two arbitrary codewords -6,a, and the Hamming distance between them is m, then the Euclidean distance D1 between them and the Euclidean distance D1 normalized by 2J'P:'s The distance d is D=la, -a, respectively. 1
=2rTS (1)d = D/2%
= m (2).

Further, it is assumed that the noise accompanying the received signal is white Gaussian, and its electric scum vector density is set to No.

First, find the detection probability. If the transmitting UW now exists at the origin O of the M-dimensional space, it is expressed by the receiving symposium.

Next, equation (3) is transformed into M-dimensional polar coordinates.

First, coordinate transformation (x1 = r2 cos $2 LX2 = r2 sin 3!12
(5) (xl, X2E (-ae, ao),
p2E(-π, π), r2Er O, w)), q(r2.yI2)=1J21p(X2.x,)-q
(r2)q(7212) (6) is obtained. Here,
IJ21 is a Jacobian, fJ21=lδ(x2
, Xl ) /a (r2.42)l = r2
It is expressed as (9). Next, coordinate transformation 1 “” (” e M) + rI, rl − for i=3 to M
I E (0+”) vX i E (−” e ”
) *Da, 6(-π/2.π/2)) Repeatedly. rl-1 is an (i-1) dimensional vector (
)J −1+ )J −2+・−, xl), and is orthogonal to xi. Now, p of rl
Assume that for d/, holds true. From formula (7),
! -8, equation (b) holds true (R2=2).

Next, we will prove that the formula Qυ also holds true for i≧3.

When we apply the coordinate transformation of equation α0 using equation (4) and αη, we get q
(rl, p,) = lJi I Q(rI-1) p
(x, ) get it. Here, IJ, 1 is the Jacobian, and I J+ l = I a (rI-1-xI)/
a(rt e j'+) i = r((represented by 13. Now R1"R1-1/J"; j CI
Q4 (to) C, = 1/8. If we set cos Ashi1dAshi, the formula (
b) is q(r, , temple) = q(r,) q(100f5q
(III+) = CIcos da I
(to). From this, it can be seen that the assumption of formula αη also holds true for 1E(8,M). By repeating equations (II) to (to) with i=8 to M, the following equation is obtained.

By the way, the equation (to) can be rewritten as follows, so it can be expressed as Equation 041 C.

Now, it is assumed that 1m1(, (when it is an even season, the following holds true. In the case of 1o-2, R2-2 and the formula) holds true.Next, we will examine the case when 1=i(1+2). From the equation) and the gradient, we get (i,)-14)-2. From this, it can be seen that the formula ) also holds true for 1E(4,M). By the way, 10 is an even number, so 16-1! It is. Therefore, formula (a) is given.

If the allowable Euclidean distance in so-called detection is ε, then M
In the dimensional space, if 4 is received inside a hypersphere whose center is the position of transmission J and whose radius is ε, J will be detected, and if it goes outside, it will not be detected. Therefore, the 7 non-detection probability UWMP is expressed by the following equation.

MP = f q (rM) (1rMε (lsuEs) Next, calculate the false detection probability.If a certain arbitrary code word is a point A-(A, 0..., 0) in the M-dimensional space , and 4 exists at the origin O. If the Hamming distance between the code word A and 4 is f, then the Euclidean distance between them is A=2〃-鳳 (to).

The M-dimensional pd/ of the received symbol is p(Xw*Xw-1#-°・m Xi) = :"j
rp(Xi) 8 ml. If formulas (5) to die are used, 1. =, p(xl) q(rM-1) is derived as follows.

Next, variable transformation □(pME[
0, -), Mε(-π, π)), then p(pM, pM) = IJ≦I Q(rM-1)
obtain XM). Here; J≦1 is the Jacobian, I J,, I = l a ”M-1,XM)/a
(P,,@0M) I = P, represented by C1. Also, from equation (4) (to), it is above.

Now, calculate the probability that code word A is erroneously determined to be open. In the M-dimensional space, this code word A is located at the center 01
If it is received inside a hypersphere with radius ε, it will be falsely detected as 4. If this probability is referred to as L false detection probability UWF P, it is expressed as. t=PM//JT,
After converting the variables and rearranging them, it is finally expressed as the following formula.

Figure 8 shows an example of UWM in the case of length M = 20 bits.
The results of calculating P and UWFP using the formulas @ and @ are shown. In addition, the same figure also shows conventional UWMP and UWFP.
Also show and compare. It can be seen that while UWFP is comparable to the present invention and the conventional method, UWMP has better characteristics in the present invention.

〔Effect of the invention〕

As described above, according to the present invention, the so-called pattern is detected by using soft-decision data as the input received signal and calculating the multidimensional Euclidean distance between the so-called pattern and the soft-decision data. Therefore, there is an effect that good door detection characteristics can be obtained.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a unique word detector according to an embodiment of the present invention, FIG. 2 is a block diagram of a unique word detector using conventional hard decision data, and FIG. 3 is a block diagram of a unique word detector of the present invention. FIG. @υ is the soft-decision received signal Pl demodulated by the soft-decision demodulator,
Bell 2 receives the soft-decision received signal Q demodulated by the soft-decision demodulator,
The correlator P1- which stores UWP is the correlator Q which stores the cabinet pattern Q, - is the stored W pattern P whose size corresponds to the soft-decision received signal PR, and ■ is the soft-decision received signal Q
A so-called storage pattern Q having a size corresponding to 1, 6η is a shift register having a number of bits sufficient to accommodate the soft decision reception signal PR or QR, state is a subtracter, - is an adder,
Ton is the comparator, Ill is the door detection gate signal, the older sister is the melon circuit,
Akatsuki is the output L detection pulse signal, and (goods) is the square circuit.

Claims (1)

[Claims] The soft-decision reception signals P_R and Q_R are input signals, and an M-stage shift register capable of accommodating these signals, respectively, and storage UW patterns P and Q that accommodate unique word (UW) patterns P and Q composed of M bits are provided. , a subtracter that calculates the difference between the value of each stage of the shift register and the value corresponding to each bit of the UW pattern, a squarer that squares the difference signal of this output, and these M squarers. A unique word detector comprising an adder for summing the output signals, an AND circuit for passing the adder output to a value ε and the comparator output signal while the UW detection gate signal is enabled.