CN104243392A - Novel self-adaption modulation algorithm - Google Patents

Abstract

The invention discloses a novel self-adaption modulation algorithm used for modulating signals received by the receiving end and calculating the LLR. The novel self-adaption modulation algorithm comprises the steps that (1), a point x1<1>, closest to yk, on a constellation set xi<1> is found; (2), the distance D1<1> between the yk and the point x1<1> is calculated; (3), the p0th point, closest to the yk, on the constellation set xi<1> is supposed to be x1<P0>, the p0 is increased progressively from two, the point x1<P0> is sought in the constellation set constantly, and the distance D1<P0> between the point yk and the point x1<P0> is calculated until that the difference between D1<P0> and D1<1> is greater than Rth, wherein Rth is a defined threshold value in a receiver, and comparison is needed p0 times additionally; (4), the position range of the yk is determined; (5), the first p1 D1<P1> is calculated in sequence according to the steps (2) and (3) in the constellation set xi<1>; (6), the LLR is calculated according to the formula. The algorithm takes complexity and performance balance into consideration at the same time, remedies the situation that the complexity of an optimal modulation algorithm is over high, and also solves the problem that a Max-log algorithm is possibly poor in performance.

Description

NEW ADAPTIVE demodulating algorithm

Technical field

The present invention relates to mobile communication system wireless transmission link technology, in particular, relate to a kind of adaptive demodulation algorithm balancing complexity and performance.

Background technology

Day by day in short supply along with radio spectrum resources, and the exponential growth of multi-medium data amount in wireless network, how effectively to improve spectrum utilization efficiency is the technological difficulties that third generation mobile communication system wireless transmission link technology must solve.Optimizing physical level coded modulation technology is the important channel realizing spectral efficient under limited bandwidth transmission, and the object of its research maximally utilises transfer resource, selects optimum transmission mechanism, to approach shannon limit.

Bit Interleaved Coded Modulation technology (BICM Bit-interleaved Coded Modulation) is as a kind of combined coding modulation scheme, by introducing interleaver between coding and modulation, make coding can independent design with modulation, so have simplicity of design, realize advantage easily, and its performance is similar to multilevel Codes, so be not only widely used in various business communication system, but also it is the key technology in wireless mobile communications link of future generation.

Quadrature amplitude modulation (Quadrature Amplitude Modulation) is a kind of method two kinds of amplitude-modulated signals (ASK and PSK) being convergeed to a channel, therefore can double extremely efficient bandwidth.Quadrature amplitude modulation is used to pulse amplitude modulation, particularly applies at wireless network.Quadrature amplitude modulated (QAM) signal has the carrier wave of two same frequencys, but phase 90 degree (1/4th cycles, from integration term).A signal is I signal, and another signal is Q signal.From mathematical angle, a signal can be expressed as sine, another is expressed as cosine.Two kinds of modulated carrier waves are mixed when launching.After arriving destination, carrier wave is separated, and then data are extracted respectively mixes mutually with raw modulation information.This modulation system usually have binary system QAM (4QAM), quaternary QAM (l6QAM), octal system QAM (64QAM) ... corresponding spacing wave vector end-points distribution map is called planisphere, have 4 respectively, 16,64 ... individual constellation point.

At transmitting terminal, information bit sequence, after encoder and interleaver, obtains bit sequence b={b to be modulated ₀..., b _k... b _k, wherein b _kfor a kth bit, sequence b again through the number of constellation point be M=2 ^mqAM digital modulator produce N=K/m transmission symbol x:x={x ₀..., x _k..., x _n-1, wherein N is number of symbols, and m is order of modulation, and M is the number of constellation point, x _kfor a kth transmission symbol.Be the QAM modulation of M for constellation point number, if point-to-point transmission beeline is d, then the coordinate of each point can be expressed as $(\&PlusMinus;\frac{2p+1}{2}d\&PlusMinus;\frac{2p+1}{2}d\&CenterDot;j),p=0,\&CenterDot;\&CenterDot;\&CenterDot;,\sqrt{M}/2-1.$

Consider additive white Gaussian noise (AWGN) channel, channel model be expressed as:

y _k＝x _k+n _k

Wherein y _kfor kth sends signal, x _kfor a kth Received signal strength, n _kfor multiple Gaussian noise, its average is zero, and variance is

At receiving terminal, need to carry out demodulation according to Received signal strength, calculate log-likelihood LLR _i,k, be specially, definition log-likelihood LLR is

${\mathrm{LLR}}_{i,k}=\ln \frac{p(b_{i,k}=0|y_k)}{p(b_{i,k}=1|y_k)}=\ln \frac{\underset{x_k\&Element;{\mathrm{\&chi;}}_i^0}{\mathrm{\&Sigma;}}p\left(x_k\right|y_k)}{\underset{x_k\&Element;{\mathrm{\&chi;}}_i^1}{\mathrm{\&Sigma;}}p\left(x_k\right|y_k)}$

Suppose to send signal equiprobability, utilize bayesian criterion, log-likelihood formula is converted into maximum-likelihood criterion:

${\mathrm{LLR}}_{i,k}=\ln \frac{\underset{x_k\&Element;{\mathrm{\&chi;}}_i^0}{\mathrm{\&Sigma;}}p\left(y_k\right|x_k)}{\underset{x_k\&Element;{\mathrm{\&chi;}}_i^1}{\mathrm{\&Sigma;}}p\left(y_k\right|x_k)}$

Wherein, p (y _k| x _k) be channel transition probability, b _i,kfor i-th bit in a kth symbol, with represent b in constellation point respectively _i,kequal the assemble of symbol of 0,1.

Because noise is Gaussian Profile, channel transition probability is so the demodulation formula of optimum is:

${\mathrm{LLR}}_{i,k}=\ln \frac{\underset{x_k\&Element;{\mathrm{\&chi;}}_i^0}{\mathrm{\&Sigma;}}\exp [-\frac{{(y_k-x_k)}^2}{2{\mathrm{\&sigma;}}_n^2}]}{\underset{x_k\&Element;{\mathrm{\&chi;}}_i^1}{\mathrm{\&Sigma;}}\exp [-\frac{{(y_k-x_k)}^2}{2{\mathrm{\&sigma;}}_n^2}]}=\ln \frac{\underset{k_0=0}{\overset{M/2-1}{\mathrm{\&Sigma;}}}\exp [-\frac{{(y_k-x_{k_0})}^2}{2{\mathrm{\&sigma;}}_n^2}]}{\underset{k_1=0}{\overset{M/2-1}{\mathrm{\&Sigma;}}}\exp [-\frac{{(y_k-x_{k_1})}^2}{2{\mathrm{\&sigma;}}_n^2}]}$

As can be seen from formula, the LLR calculating each bit needs M subtraction, M time square, M+1 division, M exponential depth, M sub-addition and 1 logarithm operation.Because exponent arithmetic complexity is high within hardware, so usually adopt following Max-log algorithm:

According to mathematical formulae: lLR formula can be converted into:

${\mathrm{LLR}}_{i,k}\&ap;\underset{x_k\&Element;{\mathrm{\&chi;}}_i^0}{\max }[-\frac{{(y_k-x_k)}^2}{2{\mathrm{\&sigma;}}_n^2}]-\underset{x_k\&Element;{\mathrm{\&chi;}}_i^1}{\max }[-\frac{{(y_k-x_k)}^2}{2{\mathrm{\&sigma;}}_n^2}]=\frac{1}{2{\mathrm{\&sigma;}}_n^2}[\underset{x_k\&Element;{\mathrm{\&chi;}}_i^1}{\min }{(y_k-x_k)}^2-\underset{x_k\&Element;{\mathrm{\&chi;}}_i^0}{\min }{(y_k-x_k)}^2]$

For this kind of algorithm, still need M subtraction, M time square, compare for M-2 time and a division.Compared to optimal algorithm, complexity has had larger lifting.

Should be noted that and work as time larger, the difference between distinct symbols is less, adopts Max-log algorithm to bring performance loss.

Above-mentioned traditional optimal demodulation algorithm can obtain optimum performance in theory, but due to its inner index operation too much, especially when order of modulation is higher, cause its algorithm complex too high.And Max-log algorithm complex is lower, but for the poor scene of channel circumstance, its approximate inequality has larger error, so can bring larger performance loss.

Summary of the invention

For the technical problem existed in above-mentioned prior art, the invention provides a kind of NEW ADAPTIVE demodulating algorithm, consider the balance of complexity and performance simultaneously, compensate for the situation that optimal demodulation algorithm complex is too high, also solve the situation of the poor-performing that Max-log algorithm may occur.

For achieving the above object, the technical solution adopted in the present invention is as follows:

A kind of NEW ADAPTIVE demodulating algorithm, carries out demodulation for the signal received receiving terminal, calculates log-likelihood ${\mathrm{LLR}}_{i,k}=\ln \frac{p(b_{i,k}=0|y_k)}{p(b_{i,k}=1|y_k)}=\ln \frac{\underset{x_k\&Element;{\mathrm{\&chi;}}_i^0}{\mathrm{\&Sigma;}}p\left(x_k\right|y_k)}{\underset{x_k\&Element;{\mathrm{\&chi;}}_i^1}{\mathrm{\&Sigma;}}p\left(x_k\right|y_k)},$ Wherein: $p\left(y_k\right|x_k)=\frac{1}{\sqrt{2\mathrm{\&pi;}{\mathrm{\&sigma;}}_n^2}}{e}^{-\frac{{(y_k-x_k)}^2}{2{\mathrm{\&sigma;}}_n^2}}$ For channel transition probability, b _i,kfor i-th bit in a kth symbol, with represent b in constellation point respectively _i,kequal the assemble of symbol of 0,1, y _k=x _k+ n _kfor kth sends signal, x _kfor a kth Received signal strength, n _kfor multiple Gaussian noise, its average is zero, and variance is bit sequence b={b to be modulated ₀..., b _k... b _k, wherein b _kfor a kth bit, sequence b again through the number of constellation point be M=2 ^mqAM digital modulator produce N=K/m transmission symbol x:x={x ₀..., x _k..., x _n-1, wherein N is number of symbols, and m is order of modulation, and M is the number of constellation point, x _kfor a kth transmission symbol, be the QAM modulation of M for constellation point number, if point-to-point transmission beeline is d, then the coordinate of each point can be expressed as $(\&PlusMinus;\frac{2p+1}{2}d\&PlusMinus;\frac{2p+1}{2}d\&CenterDot;j),p=0,\&CenterDot;\&CenterDot;\&CenterDot;,\sqrt{M}/2-1,$

It is as follows that described adaptive demodulation algorithm comprises step:

1) find respectively with y in constellation set _knearest point

2) calculation level y _kto point distance $D_1^1=\frac{{(y_k-x_1^1)}^2}{2{\mathrm{\&sigma;}}_n^2};$

3) establish distance y in constellation set _knearest p ₀individual point is p ₀increase progressively from 2, planisphere is constantly found a little and calculation level y _kto point distance until meet rear stopping calculates, wherein R _thfor the threshold value defined in receiver, need p altogether ₀compare for+1 time;

4) y is determined _kposition range;

5) for constellation set, according to step 2), 3) calculate p successively ₁individual

6) according to following formulae discovery LLR:

${\mathrm{LLR}}_{i,k}\&ap;D_1^1-D_0^1,p_0=p_1=1$

Described step 1) concrete grammar as follows:

constellation set has individual, on planisphere, find y successively _kthe grid at place, constellation set needs secondaryly to compare, according to the interval division of grid, can determine

For constellation set individual, on planisphere, adopt plane binary chop method to find y successively _kthe grid at place.

Described step 3) in threshold value R _thcan free setting.

Described step 4), according to fixed constellation point, reduce next closest approach scope, determine y _kposition range, according to from the close-by examples to those far off calculating and the distance of constellation point, thus significantly reduce number of comparisons.

NEW ADAPTIVE demodulation method proposed by the invention, compared with prior art, has following technique effect:

1. abandoning traditional needs calculates the distance of Received signal strength to all constellation point, but calculates beeline successively dynamically;

2. have employed the method compared and determine the constellation point that distance Received signal strength is nearest;

3. set threshold value to stop the calculating of beeline, threshold value can free setting;

4. algorithm can adapt to different channel circumstances automatically, reaches the balance of complexity and performance.

Accompanying drawing explanation

By reading the detailed description done non-limiting example with reference to the following drawings, other features, objects and advantages of the present invention will become more obvious:

Fig. 1 is the system model figure of the inventive method;

Fig. 2 is the inventive method one embodiment schematic diagram;

Fig. 3 (a) and Fig. 3 (b) are the algorithm schematic diagram of p0 when getting different value respectively.

Embodiment

Below in conjunction with specific embodiment, the present invention is described in detail.Following examples will contribute to those skilled in the art and understand the present invention further, but not limit the present invention in any form.It should be pointed out that to those skilled in the art, without departing from the inventive concept of the premise, some distortion and improvement can also be made.These all belong to protection scope of the present invention.

Fig. 1 is that the system model of the inventive method and specific algorithm are described in further detail.Specifically comprise:

At transmitting terminal, information bit sequence, after encoder and interleaver, obtains bit sequence b={b to be modulated ₀..., b _k... b _k, wherein b _kfor a kth bit, sequence b again through the number of constellation point be M=2 ^mqAM digital modulator produce N=K/m transmission symbol x:x={x ₀..., x _k..., x _n-1, wherein N is number of symbols, and m is order of modulation, and M is the number of constellation point, x _kfor a kth transmission symbol.Be the QAM modulation of M for constellation point number, if point-to-point transmission beeline is d, then the coordinate of each point can be expressed as $(\&PlusMinus;\frac{2p+1}{2}d\&PlusMinus;\frac{2p+1}{2}d\&CenterDot;j),p=0,\&CenterDot;\&CenterDot;\&CenterDot;,\sqrt{M}/2-1.$

Consider additive white Gaussian noise (AWGN) channel, channel model be expressed as:

y _k＝x _k+n _k

Wherein y _kfor kth sends signal, x _kfor a kth Received signal strength, n _kfor multiple Gaussian noise, its average is zero, and variance is

At receiving terminal, need to carry out demodulation according to Received signal strength, calculate log-likelihood LLR _i,k, be specially, definition log-likelihood LLR is

${\mathrm{LLR}}_{i,k}=\ln \frac{p(b_{i,k}=0|y_k)}{p(b_{i,k}=1|y_k)}=\ln \frac{\underset{x_k\&Element;{\mathrm{\&chi;}}_i^0}{\mathrm{\&Sigma;}}p\left(x_k\right|y_k)}{\underset{x_k\&Element;{\mathrm{\&chi;}}_i^1}{\mathrm{\&Sigma;}}p\left(x_k\right|y_k)}$

Suppose to send signal equiprobability, utilize bayesian criterion, log-likelihood formula is converted into maximum-likelihood criterion:

${\mathrm{LLR}}_{i,k}=\ln \frac{\underset{x_k\&Element;{\mathrm{\&chi;}}_i^0}{\mathrm{\&Sigma;}}p\left(y_k\right|x_k)}{\underset{x_k\&Element;{\mathrm{\&chi;}}_i^1}{\mathrm{\&Sigma;}}p\left(y_k\right|x_k)}$

Wherein, p (y _k| x _k) be channel transition probability, b _i,kfor i-th bit in a kth symbol, with represent b in constellation point respectively _i,kequal the assemble of symbol of 0,1.

The following detailed description of adaptive demodulation algorithm of the present invention.

1, find respectively with y in constellation set _knearest point method can be adopted as follows:

constellation set has individual, on planisphere, adopt plane dichotomy to find y successively _kthe grid at place, constellation set needs secondaryly to compare.According to the interval division of grid, can determine

For the example that Fig. 2 provides, ask the 1st LLR that bit is corresponding, by y _kreal part compare with 0: Re (y _k) > 0, then by y _kreal part compare with 4d: Re (y _k) < 4d, then by y _kreal part compare with 2d: Re (y _k) > 2d.In like manner by y _kimaginary part be defined as 2d < Im (y _k) < 4d, now can determine now compare 3+2=5 time altogether.

2, calculation level y _kto point distance $D_1^1=\frac{{(y_k-x_1^1)}^2}{2{\mathrm{\&sigma;}}_n^2}.$

3, establish distance y in constellation set _knearest p ₀individual point is p ₀increase progressively from 2, planisphere is constantly found a little and calculation level y _kto point distance until meet rear stopping calculates, wherein R _thfor the threshold value defined in receiver, need p altogether ₀secondaryly to compare.

Work as y _kwhen being positioned at position shown in Fig. 2, for p ₀=2,3,4,5 time points must be one of following four points: $(\mathrm{Re}\left(x_1^1\right),\mathrm{Im}\left(x_1^1\right)+d),(\mathrm{Re}\left(x_1^1\right),\mathrm{Im}\left(x_1^1\right)-d),(\mathrm{Re}\left(x_1^1\right)+d,\mathrm{Im}\left(x_1^1\right)),(\mathrm{Re}\left(x_1^1\right)-d,\mathrm{Im}\left(x_1^1\right)),$ As shown in Fig. 3 (a).In order to determine p ₀point when=2 only need to carry out comparing for twice, septal line was point slope is respectively ± two straight lines of 1, and for p ₀two points of=3,4, only need compare Re (y _k) and once, and remaining point is p ₀point when=4 determine that these four points compare 3 times altogether, average each point compares 0.75 time.

For p ₀=6,7,8,9 time points must be one of following four points: as shown in Fig. 3 (b), according to above-mentioned Re (y _k) and comparison, can p be determined ₀=6,7 time points be the left side in four points or right side, now once compare again and can determine all four points: compare Im (y _k) and

4, through comparing for m-1+2+1+1=m+3 time at most above, y is determined in detail _kposition range, for p ₀the point of > 9 do not need to compare and can find.

5, for constellation set, calculates p according to step 2 successively to step 3 method ₁individual wherein p ₁with p ₀may be unequal.

6, according to following formulae discovery LLR:

${\mathrm{LLR}}_{i,k}\&ap;D_1^1-D_0^1,p_0=p_1=1$

Below with two kinds of methods of prior art, done comparing of one effect to method provided by the present invention:

Analysis of complexity

Work as p ₀=p ₁when=1, the present invention program and Max-log algorithm very similar, but complexity is different, and the present invention needs 2m-2 time to compare, 2 subtractions, 2 times squares, and a division arithmetic, and it is compared as follows table:

?	Relatively	Addition	Subtraction	Square	Division	Index	Logarithm
								Max-log algorithm	M-2	0	M+1	M	1	0	0
Algorithm of the present invention	2m-2	0	2	2	1	0	0

Can be found out by contrast, algorithm complex of the present invention is lower than Max-log algorithm.

And for other situations, the contrast of three kinds of complexities is as following table:

Can find out that algorithm complex of the present invention is lower than optimal algorithm, and with Max-log algorithm compare needs according to p ₁and p ₀occurrence judge, on the whole algorithm of the present invention decrease subtraction and square number of times, but add number of comparisons, addition, the number of times of division and index.Work as p ₁and p ₀time less, algorithm complex of the present invention is better, and works as p ₁and p ₀time larger, algorithm complex is high, but the performance brought advantageously.Concrete condition needs to select threshold value R according to actual needs _thsize.

Although content of the present invention has done detailed introduction by above-described embodiment, will be appreciated that above-mentioned description should not be considered to limitation of the present invention.After those skilled in the art have read foregoing, for multiple amendment of the present invention and substitute will be all apparent.Therefore, protection scope of the present invention should be limited to the appended claims.

Claims (5)

1. a NEW ADAPTIVE demodulating algorithm, carries out demodulation for the signal received receiving terminal, calculates log-likelihood ${\mathrm{LLR}}_{i,k}=\ln \frac{p(b_{i,k}=0|y_k)}{p(b_{i,k}=1|y_k)}=\ln \frac{\underset{x_k\&Element;{\mathrm{\&chi;}}_i^0}{\mathrm{\&Sigma;}}p\left(x_k\right|y_k)}{\underset{x_k\&Element;{\mathrm{\&chi;}}_i^1}{\mathrm{\&Sigma;}}p\left(x_k\right|y_k)},$ Wherein: $p\left(y_k\right|x_k)=\frac{1}{\sqrt{2\mathrm{\&pi;}{\mathrm{\&sigma;}}_n^2}}{e}^{-\frac{{(y_k-x_k)}^2}{2{\mathrm{\&sigma;}}_n^2}}$ For channel transition probability, b _i,kfor i-th bit in a kth symbol, with represent b in constellation point respectively _i,kequal the assemble of symbol of 0,1, y _k=x _k+ n _kfor kth sends signal, x _kfor a kth Received signal strength, n _kfor multiple Gaussian noise, its average is zero, and variance is bit sequence b={b to be modulated ₀..., b _k... b _k, wherein b _kfor a kth bit, sequence b again through the number of constellation point be M=2 ^mqAM digital modulator produce N=K/m transmission symbol x:x={x ₀..., x _k..., x _n-1, wherein N is number of symbols, and m is order of modulation, and M is the number of constellation point, x _kfor a kth transmission symbol, be the QAM modulation of M for constellation point number, if point-to-point transmission beeline is d, then the coordinate of each point is expressed as $(\&PlusMinus;\frac{2p+1}{2}d\&PlusMinus;\frac{2p+1}{2}d\&CenterDot;j),p=0,\&CenterDot;\&CenterDot;\&CenterDot;,\sqrt{M}/2-1,$ It is characterized in that, it is as follows that described adaptive demodulation algorithm comprises step:

1) find respectively with y in constellation set _knearest point

2) calculation level y _kto point distance $D_1^1=\frac{{(y_k-x_1^1)}^2}{2{\mathrm{\&sigma;}}_n^2};$

3) establish distance y in constellation set _knearest p ₀individual point is p ₀increase progressively from 2, planisphere is constantly found a little and calculation level y _kto point distance until meet rear stopping calculates, wherein R _thfor the threshold value defined in receiver, need p altogether ₀compare for+1 time;

4) y is determined _kposition range;

5) for constellation set, according to step 2), 3) calculate p successively ₁individual

6) according to following formulae discovery LLR:

${\mathrm{LLR}}_{i,k}\&ap;D_1^1-D_0^1,p_0=p_1=1$

2. NEW ADAPTIVE demodulating algorithm according to claim 1, is characterized in that, described step 1) concrete grammar as follows:

constellation set has individual, on planisphere, find y successively _kthe grid at place, constellation set needs at most secondaryly to compare, according to the interval division of grid, can determine

3. NEW ADAPTIVE demodulating algorithm according to claim 2, is characterized in that, for constellation set individual, on planisphere, adopt plane binary chop method to find y successively _kthe grid at place.

4. NEW ADAPTIVE demodulating algorithm according to claim 1, is characterized in that, described step 3) in threshold value R _thcan free setting.

5. NEW ADAPTIVE demodulating algorithm according to claim 1, is characterized in that, described step 4), according to fixed constellation point, reduce next closest approach scope, determine y _kposition range, according to from the close-by examples to those far off calculating and the distance of constellation point.