CN1655541A - Differential bit soft value exporting method in orthogonal spectrum expansion differential phase modulation - Google Patents

Abstract

The present invention relates to a sort of method for outputting the differential bit soft values in the concocting of the orthogonal spread spectrum, it comprises carrying out the N-nary orthogonal spread spectrum and M-nary differential phase concocting to the information bit at the sending terminal, the receiving end does sampling with the speed of the chip, it is characterized in that the procedure of disposing the sampled signal comprises the following steps: matching and correlating the local orthogonal codes whose number is N to the sampled signal, choosing the L orthogonal despread values ( z < n, i >' ) and ( z < n+1 , j >' ) whose module square value are the biggest, according to the k-th difference bit d ( n, k) in triangle Theta m of the differential phase is zero or one, separate the M triangle Theta m into two subsets which is comprised of triangle Theta k < 0 > and triangle Theta k < 1 >, then the bit soft value of the d < n, k > is represented by the formula on the upper right, calculate the bit soft value of d < n, k > when the k equals to 1, 2, ..., l o g 2 < M >, therefore can get all differential bit soft values of l o g 2 < M > and output them.

Description

Differential bit soft value exporting method in the orthogonal spectrum expansion differential phase modulation

Technical field

The present invention relates to the efficient modulation-demodulation technique in a kind of digital communicating field, be specifically in a kind of N ary quadrature spread spectrum and the M system differential phase complex modulated differential bit soft value exporting method (wherein, N and M are 2 power series, value is 2,4,8,16...).

Background technology

Orthogonal spectrum expansion differential phase modulation (NOrth-MDPSK) is a kind of modulation technique of spectral efficient, it combines N ary quadrature spread spectrum (NOrth) and M system differential phase modulation (MDPSK), constitute hybrid multiplex modulation system, separate that timing is corresponding to comprise orthogonal spectrum expansion and MDPSK demodulation.Modulation and coding and decoding cooperatively interact in the practical communication system, and coding and decoding method commonly used has the Viterbi Soft decision decoding of convolutional encoding, and the Turbo coding and decoding, and the bit soft value input is one of essential characteristic of decoding.Therefore, when NOrth-MDPSK combines with coding and decoding, must consideration how to carry out the NOrth-MDPSK demodulation, obtain preferable bit soft value, make coding and decoding really embody high coding gain, this is the key issue that practical application need solve.

In the existing NOrth-MDPSK bit soft value exporting method, two max methods are adopted in the orthogonal spectrum expansion demodulation, utilize all the orthogonal dispreading values in the current spread symbol, calculate the quadrature bit soft value one by one.The differential ference spiral method is then adopted in the MDPSK demodulation, all orthogonal dispreading values based on former and later two spread symbol, determine two orthogonal code sequence numbers (this is equivalent to the hard decision demodulation of orthogonal spectrum expansion) of mould value maximum respectively, calculate differential bit soft value according to the orthogonal dispreading value of sequence number correspondence then.This differential ference spiral mode utilizes the hard decision demodulation result of orthogonal spectrum expansion to determine differential bit soft value in essence, makes the differential bit soft value orthogonal spectrum expansion hard decision demodulation result that places one's entire reliance upon; In case and any one hard decision demodulation mistake occurs in former and later two spread symbol, all will influence the MDPSK demodulation greatly, the high probability mistake appears in differential bit soft value, even through decoding, the overall performance of NOrth-MDPSK system is also limited greatly.

Summary of the invention

Purpose of the present invention is exactly in order to address the above problem, and proposes the differential bit soft value exporting method of a kind of orthogonal spectrum expansion differential phase that can improve the entire system performance and can reduce the algorithm implementation complexity in modulating.

The technology of the present invention solution:

Differential bit soft value exporting method in a kind of orthogonal spectrum expansion differential phase modulation, it is included in transmitting terminal information bit is carried out N ary quadrature spread spectrum and the modulation of M system differential phase, obtains sending signal by the chip formed filter; This is sampled in the optimum position with spreading rate after sending signal process channel and chip matched filter, forms sampled signal, it is characterized in that comprising the steps: in the processing to this sampled signal

It is relevant that a, sampled signal and local all orthogonal codes are mated, and in the cycle, obtains N orthogonal dispreading value { z in n and n+1 spread symbol respectively _{N, i}And { z _{N+1, j}, wherein n is an integer, 1≤i, j≤N;

B, from N orthogonal dispreading value { z _{N, i}And { z _{N+1, j}In, choose L orthogonal dispreading value of mould value square maximum respectively, be designated as { z _{N, i}' and { z _{N+1, j}', 1≤L≤N wherein;

C, according to a M system differential phase modulated M differential phase Δ θ _m, Δ θ _m=2 π m/M, 0≤m≤M-1 is to { z _{N, i}' and { z _{N+1, j}' carry out combined crosswise, constitute L * L combination of M group z _{N, i}'+z _{N+1, j}' * exp (j Δ θ _m), and carry out mould value square operation, and obtain L * L mould square value of M group | z _{N, i}'+z _{N+1, j}' * exp (j Δ θ _m) | ²

D, according to the differential phase Δ θ of M system differential phase modulation _mOn k differential bit d _{N, k}Be 0 or 1, $1\&le;k\&le;{\log }_2^M,$ M is organized L * L mould square value be divided into two classes, constitute L * L the mould square value of M/2 group under 0 bit and the 1 bit condition respectively, and from this two classes mould square value, select maximum separately

With

Δ θ wherein _k ⁰With Δ θ _k ¹Expression { Δ θ _mAccording to d _{N, k}Be two subclass that 0 bit and 1 bit are divided;

E, calculating

With

Difference, this difference is as d _{N, k}Bit soft value Λ (d _{N, k}), i.e. Λ (d _{N, k}) satisfy following formula:

$\mathrm{\&Lambda;}\left(d_{n,k}\right)=\underset{i,j\&Element;\{1....L\},\mathrm{\&Delta;}{\mathrm{\&theta;}}_m\&Element;\mathrm{\&Delta;}{\mathrm{\&theta;}}_{\mathrm{<figure-callout\; id="1"\; label="k"\; filenames="A20051005201000091.png"\; state="\{\{state\}\}">k</figure-callout>}}^1}{\max }{|z_{n,1}^{\&prime;}+z_{n+1,j}^{\&prime;}*\exp (-\mathrm{j\&Delta;}{\mathrm{\&theta;}}_m)|}^2-\underset{i,j\&Element;\{1....L\},\mathrm{\&Delta;}{\mathrm{\&theta;}}_m\&Element;\mathrm{\&Delta;}{\mathrm{\&theta;}}_k^0}{\max }{|z_{n,1}^{\&prime;}+z_{n+1,j}^{\&prime;}*\exp (-\mathrm{j\&Delta;}{\mathrm{\&theta;}}_m)|}^2;$

F, repeating step d～e, d when calculating k respectively is worth for other _{N, k}Bit soft value, up to obtaining all log ₂ ^MIndividual differential bit soft value and output.

The present invention is directed to the shortcoming of existing differential ference spiral method in the NOrth-MDPSK modulation,, proposed a brand-new differential bit soft value exporting method, be called the MAP algorithm based on maximum a posteriori probability (MAP) criterion.This algorithm makes full use of all orthogonal dispreading values of former and later two spread symbol, calculates differential bit soft value.The MAP algorithm has avoided that orthogonal spectrum expansion hard decision demodulation result has improved NOrth-MDPSK entire system performance to the influence of differential bit soft value in the existing differential ference spiral method.Simultaneously, in order to reduce the implementation complexity of MAP algorithm, give and simplify the MAP algorithm.In above-mentioned technical solution, L=N represents the MAP algorithm, and L＜N represents to simplify the MAP algorithm, and L=1 then deteriorates to existing differential ference spiral method fully.

Description of drawings

Fig. 1 is a NOrth-MDPSK baseband modulation and demodulation general structure schematic diagram.

Fig. 2 is 1 differential bit soft value export structure schematic diagram of NOrth-BDPSK.

Fig. 3 is 2 differential bit soft value export structure schematic diagrames of NOrth-QDPSK.

Fig. 4 is the performance comparison curve chart of the soft value output algorithm of 32Orth-BDPSK difference.

Fig. 5 is the performance comparison curve chart of the soft value output algorithm of 32Orth-QDPSK difference.

Embodiment

As shown in Figure 1, transmitting terminal is according to K _N=log ₂The N bit is gathered W={W from orthogonal intersection ₁, W ₂..., W _NOrthogonal intersection W of middle selection _n, through the K of Gray (Gray) mapping _M=log ₂The M information bit carries out the modulation of MDPSK differential phase, and two kinds of compound backs of modulation are by chip formed filter g _T(t), obtain baseband transmit signals.This sends signal through channel and chip matched filter g _R(t) after, sample in the optimum position, form sampled signal with spreading rate.Can carry out orthogonal spectrum expansion based on sampled signal and separate mediation MDPSK demodulation, obtain K _N+ K _MIndividual bit soft value.

The present invention calculates differential bit soft value at sampled signal, and its specific implementation step is as follows:

1, sampled signal and local all orthogonal codes are mated relevantly, in the cycle, obtain N orthogonal dispreading value { z in n and n+1 spread symbol respectively _{N, i}And { z _{N+1, j}, wherein n is an integer, 1≤i, j≤N;

2, from N orthogonal dispreading value { z _{N, i}And { z _{N+1, j}In, choose L orthogonal dispreading value of mould value square maximum respectively, be designated as { z _{N, i}' and { z _{N+1, j}', 1≤L≤N wherein;

3, according to a M system differential phase modulated M differential phase Δ θ _m, Δ θ _m=2 π m/M, 0≤m≤M-1 is to { z _{N, i}' and { z _{N+1, j}' carry out combined crosswise, constitute L * L combination of M group z _{N, i}'+z _{N+1, j}' * exp (j Δ θ _m), and carry out mould value square operation, and obtain L * L mould square value of M group | z _{N, i}'+z _{N+1, j}' * exp (j Δ θ _m) | ²

4, the differential phase Δ θ that modulates according to M system differential phase _mOn k differential bit d _{N, k}Be 0 or 1, $1\&le;k\&le;{\log }_2^M,$ M is organized L * L mould square value be divided into two classes, constitute L * L the mould square value of M/2 group under 0 bit and the 1 bit condition respectively, and from this two classes mould square value, select maximum separately

With Δ θ wherein _k ⁰With Δ θ _k ¹Expression { Δ θ _mAccording to d _{N, k}Be two subclass that 0 bit and 1 bit are divided;

5, calculate

With Difference, this difference is as d _{N, k}Bit soft value Λ (d _{N, k}), i.e. Λ (d _{N, k}) satisfy following formula:

$\mathrm{\&Lambda;}\left(d_{n,k}\right)=\underset{i,j\&Element;\{1,..,L\},\mathrm{\&Delta;}{\mathrm{\&theta;}}_m\&Element;\mathrm{\&Delta;}{\mathrm{\&theta;}}_{\mathrm{<figure-callout\; id="1"\; label="k"\; filenames="A20051005201000091.png"\; state="\{\{state\}\}">k</figure-callout>}}^1}{\max }{|z_{n,1}^{\&prime;}+z_{n+1,j}^{\&prime;}*\exp (-\mathrm{j\&Delta;}{\mathrm{\&theta;}}_m)|}^2-\underset{i,j\&Element;\{1,..,L\},\mathrm{\&Delta;}{\mathrm{\&theta;}}_m\&Element;\mathrm{\&Delta;}{\mathrm{\&theta;}}_k^0}{\max }{|z_{n,1}^{\&prime;}+z_{n+1,j}^{\&prime;}*\exp (-\mathrm{j\&Delta;}{\mathrm{\&theta;}}_m)|}^2;$

6, repeating step 4～5, d when calculating k respectively is worth for other _{N, k}Bit soft value, up to obtaining all log ₂ ^MIndividual differential bit soft value and output.

Above-mentioned steps explanation: the present invention utilizes all orthogonal dispreading values of former and later two spread symbol and possible MDPSK differential phase, unites coupling, and according to 0,1 bit distribution characteristic of differential phase, calculates differential bit soft value.The present invention has successfully avoided the hard decision demodulation of orthogonal spectrum expansion in the former Difference Solution accent method, helps promoting the overall performance of NOrth-MDPSK system.

The embodiment of the invention:

Be example with L=N below, provided the computational process and the export structure of the differential bit soft value of NOrth-BDPSK (M=2) and NOrth-QDPSK (M=4).When L=N, { z _{N, i}And { z _{N, i}', { z _{N+1, j}And { z _{N+1, j}' identical, the 2nd step can omit fully in the above-mentioned performing step.The export structure of differential bit soft value as shown in Figures 2 and 3, concrete computational process is as follows:

One, calculates the orthogonal dispreading value

Suppose the sampled signal row vector r in n and n+1 spread symbol cycle _n, r _N+1Expression, orthogonal intersection is a column vector, then sampled signal and local all orthogonal intersections mate the despread values z after relevant _{N, i}, z _{N+1, j}Be expressed as:

z _n，i＝r _n·W _i， z _n+1，j＝r _n+1·W _j， 1≤i，j≤N

Two, N * N mould square value of structure M group.

NOrth-BDPSK: differential phase Δ θ _mBe 0, π, 2 groups of N * N mould square values are according to following rule structure:

|z _n，i+z _n+1，j| ²， |z _n，i-z _n+1，j| ²， 1≤i，j≤N

NOrth-QDPSK: differential phase Δ θ _mBe 0, pi/2, π, 3 pi/2s, 4 groups of N * N mould square values are according to following rule structure:

|z _n，i+z _n+1，j| ²，|z _n，i-z _n+1，j| ²，|z _n，i+jz _n+1，j| ²，|z _n，i-jz _n+1，j| ²

Subscript i in the following formula, j represents sequence number, 1≤i, j 〉=N, symbol j represents imaginary unit.

Three, calculate differential bit soft value

NOrth-BDPSK: have only 1 differential bit, its soft value is calculated according to following formula:

$\underset{1\&le;i,j\&le;N}{\mathrm{<figure-callout\; id="1"\; label="max"\; filenames="A20051005201000091.png"\; state="\{\{state\}\}">max</figure-callout>}}{|z_{n,i}-z_{n+1,j}|}^2-\underset{1\&le;i,j\&le;N}{\mathrm{<figure-callout\; id="1"\; label="max"\; filenames="A20051005201000091.png"\; state="\{\{state\}\}">max</figure-callout>}}{|z_{n,i}+z_{n+1,j}|}^2$

NOrth-QDPSK: 2 differential bits are arranged, and soft value calculating formula is expressed as respectively:

The 1st differential bit soft value:

max{|z _n，i-z _n+1，j| ²，|z _n，i-jz _n+1，j| ²，1≤i，j≤N}-max{|z _n，i+z _n+1，j| ²，|z _n，i+jz _n+1，j| ²，1≤i，j≤N}

The 2nd differential bit soft value:

max{|z _n，i-z _n+1，j| ²，|z _n，i+jz _n+1，j| ²，1≤i，j≤N}-max{|z _n，i+z _n+1，j| ²，|z _n，i-jz _n+1，j| ²，1≤i，j≤N}

For the ease of the MAP algorithm among the present invention, simplification MAP algorithm and existing differential ference spiral algorithm are compared, Fig. 4～Fig. 5 has provided bit signal to noise ratio and the bit error rate correlation curve of 32Orth-BDPSK and 32Orth-QDPSK.Orthogonal intersection adopts the Hadamard sequences of N=32, and two max methods are adopted in the output of quadrature bit soft value.M=0,2,4 represent simple NOrth, NOrth-BDPSK and NOrth-QDPSK respectively; Coding and decoding measure employing code check is 2/3 Turbo product code, and channel condition is an awgn channel.Fig. 4～Fig. 5 shows:

MAP algorithm and simplification MAP algorithm all obviously are better than existing differential ference spiral algorithm among the present invention, when bit error rate is 10 ^-5The time, the MAP algorithm can provide the snr gain of 0.3～0.4dB.This proves absolutely the bright NOrth-MDPSK of the being conceived to differential bit soft value exporting method of we, has important function for improving the entire system performance.In addition, the simplification MAP algorithm of L=3 approaches the MAP algorithm very much, and this has directive significance to the practical application of simplifying the MAP algorithm.

In the above-described embodiments,, can also extend in other NOrth-MDPSK system though, the invention is not restricted to this system at 32Orth-MDPSK (M=2,4) system, as: N=64,128, M=8.

Claims (1)

1, the differential bit soft value exporting method in a kind of orthogonal spectrum expansion differential phase modulation, it is included in transmitting terminal information bit is carried out N ary quadrature spread spectrum and the modulation of M system differential phase, obtains sending signal by the chip formed filter; This is sampled in the optimum position with spreading rate after sending signal process channel and chip matched filter, forms sampled signal, it is characterized in that comprising the steps: in the processing to this sampled signal

It is relevant that a, sampled signal and local all orthogonal codes are mated, and in the cycle, obtains N orthogonal dispreading value { z in n and n+1 spread symbol respectively _{N, i}And { z _{N+1, j}, wherein n is an integer, 1≤i, j≤N;

B, from N orthogonal dispreading value { z _{N, i}And { z _{N+1, j}In, choose L orthogonal dispreading value of mould value square maximum respectively, be designated as { z _{N, i}' and { z _{N+1, j}', 1≤L≤N wherein;

C, according to a M system differential phase modulated M differential phase Δ θ _m, 0≤m≤M-1 is to { z _{N, i}' and { z _{N+1, j}' carry out combined crosswise, constitute L * L combination of M group z _{N, i}'+z _{N+1, j}' * exp (j Δ θ _m), and carry out mould value square operation, and obtain L * L mould square value of M group | z _{N, i}'+z _{N+1, j}' * exp (j Δ θ _m) | ²

D, according to the differential phase Δ θ of M system differential phase modulation _mOn k differential bit d _{N, k}Be 0 or 1, $1\&le;k\&le;{\log }_2^M,$ M is organized L * L mould square value be divided into two classes, constitute L * L the mould square value of M/2 group under 0 bit and the 1 bit condition respectively, and from this two classes mould square value, select maximum max separately _{Δ θ k0}And max _{Δ θ k1}

E, calculating max _{Δ θ k0}And max _{Δ θ k1}Difference, this difference is as d _{N, k}Bit soft value Λ (d _{N, k}), i.e. Λ (d _{N, k}) satisfy following formula:

F, repeating step d～e, d when calculating k respectively is worth for other _{N, k}Bit soft value, up to obtaining all log ₂ ^MIndividual differential bit soft value and output.

Images