CN100527721C - A method of bit and power distribution for adaptive modulation of OFDM system - Google Patents

Abstract

This invention relates to a distribution algorithm approach of bit and power, which is used to design self-adapting modulation plan in OFDM system. First, it groups all subsidiary communication channels with same space, and then in each group, it distributes bit and power at the same time; wherein the distribution process in each group comprises six steps: ordering according to subsidiary channel gain, dividing subsidiary channel cluster, generating sampling subsidiary communication channel, distributing bit according to greed distribution method, copying bit number and computing the needed power. Said algorithm approach has low complex rate, and the complex rate can be adjusted flexibly; besides, the distributing result from said algorithm approach is very near to the best distribution.

Description

A kind of bit and power distribution method that is used for the ofdm system Adaptive Modulation

Technical field

The present invention is a kind of adaptive bit and power distribution method, the bit and the power division that are used for each subchannel of ofdm communication system, determine the modulation system of each subchannel and the average power that transmits, belong to the adaptive technique field in the multi-carrier communications systems.

Background technology

OFDM (orthogonal frequency division multiplexing, OFDM) is a kind of multi-carrier communication technology that can effectively improve communication system performance in the frequency selective fading environment.OFDM is divided into the enough narrow subchannel of a plurality of bandwidth to whole transmission channel, comes parallel transmission information by these subchannels.Because each subchannel presents the flat fading characteristic, can carry out the demodulation of signal easily at receiving terminal, saved the channel equalization technique of the complexity that single-carrier system institute must employing, effectively reduce system complexity.Each sub-channel spectra of OFDM is overlapping, has the higher availability of frequency spectrum.At present, the OFDM technology is adopted by a lot of communication technology standards, as ADSL, DVB, DAB, IEEE802.11a, HIPERLAN/2, IEEE802.16 or the like.In next generation mobile communication system (Beyond 3G), the OFDM technology also is the strongest competitor.

Adaptive Modulation is a kind of along with channel state variations, change the technology of communication system modulation system and signal transmission power automatically.It can effectively improve the transmission rate of system and the reliability of system transmissions.In ofdm system, the channel gain difference of each subchannel therefore can be according to the gain of each subchannel, for each subchannel distributes suitable bit number and power, just allow each subchannel adopt suitable modulation system and signal transmitting power, so that systematic function reaches best.At present there be multiple adaptive bit and the power distribution method that is used for ofdm system to be suggested.Have plenty of in these methods under the certain condition of total transmitting power and error rate of system, make system transmissions speed reach maximum; Have plenty of under the certain condition of system transmissions speed and the error rate, make needed total transmitting power minimum.In addition, the method that also has is at first all subchannels to be divided into groups, adjacent several subchannels are divided into one group, distribute the bit and the power of similar number then for each subchannel in each grouping, needed signaling consumption during with reduction adaptation transmitter modulation intelligence.

Some communication services in the practical application, as browsing page, file in download etc., message transmission rate changes.These business do not require real-time.And for some other communication services, as multimedia audio, video traffic, desired message transmission rate is fixed.Simultaneously, for satisfying quality of service requirement, error rate of system must be lower than certain value.For the communication service of these fixed rates, the target of bit and power division is under the requirement of the certain data rate and the error rate, to make the needed transmitting power minimum of communication system.The present invention designs at the communication service of these fixed rates just.

Greedy distribution method is a kind of bit and power distribution method commonly used that is applicable to the multicarrier system of fixed data rate.This method can reach optimum bit and power distribution result.Its main thought is that it is zero that all subchannel bit numbers at first are set, and all bits to be allocated, distributes to subchannel successively then.Divide timing at every turn, at first find that subchannel of required increase power minimum, distribute one or more bits for then this subchannel.Each bit number (Bit Allocation in Discrete step-length) that distributes is determined by the system modulation mode.Circulation is gone down like this, is assigned with up to all bits to finish.Needed transmitting power when calculating each subchannel at last and guaranteeing certain error rate.Existing bit and power distribution algorithm based on greedy distribution method, assigning process is to carry out on all subchannels, so complexity is than higher.Particularly when subchannel number and bit number to be allocated all when higher, the operation times in needed cycle-index and the each circulation is all bigger, is not suitable for the central application of reality.

In fact, in ofdm system, the subchannel that those channel gains are more approaching finally might be assigned with the bit of similar number.Based on this rule, we can give their allocation bit the consideration of putting together of the more approaching subchannel of these channel gains simultaneously, rather than once only give a sub-channel allocation bit.Like this, the operand of the needed cycle-index of allocation algorithm and each circulation will reduce greatly, and assigning process is greatly simplified.In addition, for further reducing the complexity of algorithm, can be divided into all subchannels several groups, bit and power division process can be carried out in these several groups simultaneously.Assigning process in each grouping is further simplified, and the requirement of hardware complexity has further been reduced.And because the assigning process in each grouping can carry out simultaneously, needed total time of whole allocation process is greatly diminished.Though allocation result does not reach optimal result, when the combination of subchannel with divide into groups when suitable, allocation result can approach optimal result.So just accommodate actual central application.

Summary of the invention

The present invention is a kind of adaptive bit and power distribution algorithm that is applicable to ofdm communication system, is used for designing the adaptive modulation scheme of ofdm system.The objective of the invention is according to difference channel condition constantly, for each subchannel in the ofdm communication system is determined the optimum modulation system and the average power that transmits.This optimization Algorithm target is: under the certain condition of system data rates and the error rate, determine the bit number of each subchannel optimum, make the needed total transmitted power of system minimum.This optimization aim and constraints can be summarized as follows:

${\left[\underset{i=1}{\overset{N_C}{\mathrm{\Σ}}}{P}_i\right]}_{\mathrm{Min}},$

Satisfy simultaneously $\underset{i=1}{\overset{N_C}{\mathrm{\Σ}}}{b}_i=R_b,$ P _i≥0，b _i≥0，i＝1，2，…，N _C。

P wherein _iBe that i subchannel is the needed signal transmission power of the error rate that reaches certain in the system, b _iIt is the bit number of distributing to i subchannel.N _CIt is the number of all subchannels.R _bBe the bit number (being bit number to be allocated in each assigning process) that each OFDM symbol is comprised, determine by system data rates.

The physical significance of some parameters of using in the arthmetic statement and the relation between the parameter at first are described below:

N _C---need the subchannel sum of allocation bit and power in the ofdm system

R _b---the bit number (total number of bits to be allocated of each assigning process) that each OFDM symbol is comprised

N _U---the number of all subchannel groupings

N _C'---the subchannel number (N that comprises in each subchannel grouping _C'=N _C/ N _U)

The subchannel number that comprises in L---each subchannel bunch

N _S---the number (N of subchannel in the grouping of each subchannel bunch _S=N _C/ (LN _U))

R _b'---total number of bits (R to be allocated in each grouping _b'=R _b/ N _U)

R _b ^*---the actual required total number of bits (R that distributes in each grouping _b ^*=R _b'/L)

A kind of bit and power distribution algorithm that is used for the ofdm system Adaptive Modulation proposed by the invention, its key step can be sketched and be: at first, according to the system applies environment, and the number of the self adaptation distribution processor that adopts of system, determine the grouping number N of subchannel _U, and the subchannel bunch subchannel number L that comprises in each grouping; Secondly, to all N _CThe individual subchannel that needs allocation bit and power according to they positions in frequency spectrum, is uniformly-spaced divided, and is divided into N _UIndividual grouping comprises the subchannel of similar number in each grouping; Then, in each grouping, carry out the distribution of bit and power respectively; In each grouping, carry out the branch timing of bit and power, at first according to the size of the amplitude of subchannel gains, to the subchannel in this grouping by from small to large or order from big to small sort; In each grouping, with the subchannel after the ordering, the mode by dividing continuously is divided into N then _SIndividual sub-cluster of channels (N _S=N _C/ (LN _U)), the subchannel number that comprises in each subchannel bunch is L; Next for each subchannel clusters into a sampling subchannel, the scheme that generates the sampling subchannel can be selected " least gain " scheme or " reciprocal average " scheme; Then the method for distributing by greediness on all sampling subchannels is carried out Bit Allocation in Discrete, and when carrying out Bit Allocation in Discrete by the method for greediness distribution, the bit number of actual required distribution is R _b/ (LN _U); After Bit Allocation in Discrete is intact,, duplicate to each subchannel in the subchannel of correspondence bunch with the bit number that distributes on each sampling subchannel; Calculate the needed signal power of each subchannel at last respectively.The assigning process of bit and power finishes.

Shown in Figure 1 in the flow chart of algorithm of the present invention such as the Figure of description.

Each step to algorithm of the present invention is elaborated below.

At first, according to the system applies environment, and the number of the self adaptation distribution processor that adopts of system, determine the grouping number N of subchannel _U, and the subchannel bunch subchannel number L that comprises in each grouping.

The number of the self adaptation distribution processor in the system requires decision by the system hardware complexity.These processors have identical functions, and can parallel running.Bit and power division in each grouping are finished by a distribution processor respectively.So grouping number N of subchannel _UBe not more than the number of the self adaptation distribution processor in the system.N _UWith the selection of L, should be according under the different application environment, the relation between the performance of algorithm and the complexity of algorithm is determined.The criterion of being followed is under the prerequisite that reaches the systematic function requirement, to make the complexity of algorithm minimum.The complexity of algorithm is by N _UWith this two parameters decision of L.N _UBig more with L, the complexity of algorithm is low more, and the needed time of assigning process is also short more.The relation of the complexity of the performance of system and algorithm under the varying environment, can by in advance to system under specific environment, at different N _UObtain with the performance simulation under the L numerical value.In addition, the grouping number N of subchannel _UThe total N that must be able to divide exactly subchannel _C, to guarantee to have in each grouping the subchannel of similar number.And L can be divided exactly N _C/ N _U, to guarantee to comprise in each subchannel bunch the subchannel of similar number.In general, N _C, N _UWith L all be 2 integral number power.

Second of algorithm goes on foot and is, all are needed the subchannel of allocation bit and power, according to they positions in frequency spectrum, uniformly-spaced divides, and is divided into N _UIndividual grouping comprises the subchannel of similar number in each grouping.Shown in Figure 2 in mode that channel is uniformly-spaced divided such as the Figure of description.The subchannel number N that each grouping is comprised _C' calculate according to following formula

N _C′＝N _C/N _U

The effect that subchannel is divided into groups is that after grouping, the carrying out that the distribution of bit and power can walk abreast in these groupings finished the needed time of assigning process like this and can effectively reduce.On the other hand, after grouping, subchannel number in each grouping and bit number to be allocated are and drop to original 1/N _U, the complexity of each distribution processor hardware can reduce greatly like this, finishes the assigning process required time and also can significantly reduce, and is convenient to actual central application.Adopting uniformly-spaced during to channel packet, the reason of dividing mode is, because stipulating the total number of bits to be allocated in each grouping in the algorithm all equates---for this result being met or approaching optimum bit allocation result, the transmission characteristic of each grouping sub-channels on the whole should be more approaching.When adopting the mode of uniformly-spaced dividing, with adjacent, be in the several subchannels (they have similar transmission characteristic) within the correlation bandwidth, be divided in the different groupings and go, can guarantee that the subchannel in each grouping has close overall transfer characteristic.The number N of grouping _UThe constraints that should satisfy is: N _UWith the product of subchannel bandwidth, should be less than the coherence bandwidth of channel.Simultaneously, N _UShould be able to divide exactly N _C, the subchannel of similar number is arranged to guarantee each grouping.Can guarantee that each transmission packets characteristic is more approaching on the whole so on the one hand, also simplify the complexity of assigning process on the other hand.

After grouping was finished, with total number of bits to be allocated, average mark divided into groups to each.To need the bit number R that distributes in each grouping _b' calculate according to following formula

R _b′＝R _b/N _U

In the 3rd step of algorithm, in each grouping, carry out the distribution of bit and power exactly simultaneously respectively.In the grouping of each subchannel, the process of bit and power division mainly comprises six steps: ordering, divide subchannel bunch, and generate the sampling subchannel, by greedy distribution method allocation bit, duplicate bit number, calculate the subchannel power demand.

In each grouping, carry out the branch timing of bit and power, at first will the subchannel in each grouping be sorted according to the size of the amplitude of subchannel gains.Ordering can be carried out according to order from big to small, also can carry out according to order from small to large.After channel estimating, subchannel grouping, the channel gain set of the subchannel in i the grouping that obtains can be expressed as { H _{I, j}| j=1,2 ..., N _C', i=1 wherein, 2 ..., N _UAfter ordering, obtaining i the sets of sub-channels in the grouping can be expressed as $\left\{H_{i,j}^{\′}\right|j=\mathrm{1,2},\·\·\·,N_C^{\′}\},$ Following relationship is satisfied in this set

$\left|H_{i,1}^{\′}\right|\≤\left|H_{i,2}^{\′}\right|\≤\·\·\·\≤\left|H_{i,N_C\′-1}^{\′}\right|\≤\left|H_{i,N_C\′}^{\′}\right|$

Perhaps

$\left|H_{i,1}^{\′}\right|\≥\left|H_{i,2}^{\′}\right|\≥\·\·\·\≥\left|H_{i,N_C\′-1}^{\′}\right|\≥\left|H_{i,N_C\′}^{\′}\right|$

Function wherein | x| represents the range value of plural x.

After ordering is finished, need carry out the division of subchannel bunch.When dividing subchannel bunch, be in each grouping, the subchannel after the ordering is divided into N _SIndividual sub-cluster of channels.What divide subchannel bunch employing is the mode of dividing continuously.Promptly in the subchannel after ordering, adjacent several subchannels are divided into cluster.Each subchannel bunch comprises the subchannel of similar number.Divide in the mode of subchannel bunch such as the Figure of description shown in Figure 2.The total N of the subchannel that is divided into bunch _S, calculate according to following formula

N _S＝N _C′/L＝N _C/(LN _U)

The mode of adopt this ordering earlier, dividing again continuously can make the range value of each subchannel gains in each subchannel bunch very approaching.In algorithm, the bit number that each subchannel in each subchannel bunch finally is assigned with is identical.And carrying out the branch timing according to the bit and the power distribution algorithm of optimum, the subchannel that the amplitude of those channel gains is very approaching finally probably is assigned with identical bit number.The mode of therefore adopt this ordering earlier, dividing again continuously can make the bit and the power distribution result that finally obtain very approaching with the bit and the power distribution result of optimum.When guaranteeing algorithm performance, subsequent bit and power division process are simplified like this.

Ordering and dividing after subchannel bunch finishes need be clustered into " the sampling subchannel " of a correspondence for each subchannel.The gain of sampling subchannel is the function of the gain of all subchannels in the corresponding subchannel bunch.The scheme that generates the sampling subchannel have two kinds available: " least gain " scheme and " reciprocal average " scheme.

" least gain " scheme is to get in each subchannel bunch, and the minimum value in the gain range value of all subchannels is as the gain of the sampling subchannel of this subchannel bunch correspondence.If in i the grouping, the subchannel gains amplitude in the individual sub-cluster of channels of m can be expressed as $\left\{\right|H_{(m-1)+P}^i\left||p=\mathrm{1,2},\·\·\·,L\right\},$ I=1 wherein, 2 ..., N _U, m=1,2 ..., N _SThe sampling subchannel gains of m the sub-cluster of channels correspondence in i grouping then

Can be expressed as

${\left|H\right|}_m^i=\underset{\∀p=\mathrm{1,2},...,L}{\min }\left|H_{(m-1)L+p}^i\right|$

Function m in[X wherein] expression gets the minimum value element among the set X.

" reciprocal average " scheme is to get in each subchannel bunch, and the mean value of the gain range inverse of all subchannels is as the reciprocal value of the gain of the sampling subchannel of this subchannel bunch correspondence.If in i the grouping, the subchannel gains amplitude in the individual sub-cluster of channels of m can be expressed as $\left\{\right|H_{(m-1)+P}^i\left||p=\mathrm{1,2},\·\·\·,L\right\},$ I=1 wherein, 2 ..., N _U, m=1,2 ..., N _SThe sampling subchannel gains of m the sub-cluster of channels correspondence in i grouping then

Can be expressed as

${\left|H\right|}_m^i=L/\underset{p=1}{\overset{\mathrm{<figure-callout\; id="1"\; label="L"\; filenames="C200510000082E00151.png,C200510000082E00171.png"\; state="\{\{state\}\}">L</figure-callout>}}{\mathrm{\&Sigma;}}}\frac{1}{\left|H_{(m-1)L+p}^i\right|}$

Select " least gain " scheme still to select " reciprocal average " scheme in the practical application, should take all factors into consideration the algorithm performance requirement and the hardware complexity requirement decides.The computational complexity of " reciprocal average " scheme will be higher than " least gain " scheme, but performance is better than " least gain " scheme.If in the practical application, very high to the algorithm performance requirement, and allow higher computational complexity, can select " reciprocal average " scheme.If in the practical application, it is not very high that algorithm performance is required, but only allows lower computational complexity, can select " least gain " scheme.

After generation sampling subchannel finishes, carry out Bit Allocation in Discrete according to greedy allocation algorithm.Carrying out the process of Bit Allocation in Discrete according to greedy distribution method, is to carry out on the sampling subchannel in each grouping respectively, rather than on all subchannels, carry out.And bit in each grouping and power division process can walk abreast and carry out.

In each grouping, when carrying out Bit Allocation in Discrete according to greedy allocation algorithm, the sum of the subchannel of required consideration is the total N of sampling subchannel _S(or sum of each grouping sub-channels bunch).Because in the final allocation result, the bit number that all subchannels in each subchannel bunch distribute is identical.Therefore, the N in each grouping _SWhen carrying out Bit Allocation in Discrete on the individual sampling subchannel, the total number of bits R of actual required distribution ^*For

R ^*＝R _b′/L＝R _b/(LN _U)

This shows that according to the algorithm that the present invention proposes, when carrying out bit and power division according to greedy distribution method, the subchannel number of actual required consideration only is all number of subchannels purpose 1/ (LN _U), and the bit number of required distribution also is 1/ (LN of all total number of bits to be allocated _U).Therefore actual bit and power division process complexity greatly reduce.

When carrying out Bit Allocation in Discrete, consider that the modulation system that adopts in the middle of the real system is quadrature amplitude modulation (QAM, Quadrature Amplitude Modulation) according to greedy distribution method.In the bit allocation procedures, calculate j subchannel and send b _jNeeded average power signal P during individual bit _jThe time, calculate according to following formula

$P_j=\frac{{\mathrm{KP}}_j^n(2^{b_j}-1)}{{\left|H_j\right|}^2}(*)$

Wherein Be j the noise average power in the subchannel.K is a constant, and only the target error rate by system determines.K value under the different target error rate can be selected with reference to following table

Target error rate P e	10 -2	10 -3	10 -4	10 -5	10 -6	10 -7
							K	1.9	3.3	4.8	6.2	7.7	9.1

During according to greedy distribution method allocation bit, need the max. number of bits of determining that Bit Allocation in Discrete step-length and subchannel are allowed.These are by modulation system decision available in the real system.For example, if all available modulation systems of system be 0, BPSK, QPSK, 8QAM, 16QAM, 32QAM, 64QAM, 128QAM} (wherein 0 this subchannel of expression does not send signal), the bit number that the symbol after each modulation of their correspondences comprises is respectively 0,1,2,3,4,5,6,7.The Bit Allocation in Discrete step-length is 1 so, and the max. number of bits of each subchannel is 7.Similarly, if all available modulation systems of system be 0, QPSK, 16QAM, 64QAM}, the bit number that comprises of symbol after each modulation of their correspondences is respectively 0,2,4,6.The Bit Allocation in Discrete step-length is 2 so, and the max. number of bits of each subchannel is 6.

On the sampling subchannel of each grouping, the process of carrying out Bit Allocation in Discrete according to greedy distribution method is as follows:

1. carry out initialization, it is 0 that the current allocation bit number of all sampling subchannels is set.Even b _m=0, m=1,2 ..., N _SCalculate all total number of bits R to be allocated ^*=R _b/ (LN _U).

2. when calculating d bit of each sampling subchannel increase respectively, the average transmission power Δ P of required increase _m(d is the step-length of Bit Allocation in Discrete here, by the modulation system decision that real system adopted) calculated and carried out according to following formula

${\mathrm{\&Delta;P}}_m=\left\{\begin{array}{cc}{P}_m(b_m+d)-P_m\left(b_m\right),& (b_m+d\&le;D)\\ 0,& (b_m+d>D)\end{array}\right.,$

M=1 wherein, 2 ..., N _S, the maximum of the bit number that D can be assigned with for each subchannel (by the modulation system decision of the high-order that adopts in the real system).P _mWhen (b) every symbol transmission b bit of subchannel sampled in m of expression, needed average power signal calculated according to (*) formula.

3. find power increment value minimum and non-vanishing sampling subchannel m ₀, the current allocation bit number of the subchannel of should sampling increases d.Promptly

$b_{m_0}=b_{m_0}+d$

4. calculate the total number of bits of current distribution $R_{\mathrm{tot}}={\mathrm{\&Sigma;}}_{m=1}^{N_S}{b}_m.$ If R _Tot＜R ^*, turned back to for the 2nd step.If $R_{\mathrm{tot}}=R_b^{*},$ All bits to be allocated all assign, and bit allocation procedures finishes.

In each grouping, after the process of carrying out Bit Allocation in Discrete according to greedy distribution method finishes,, duplicate to all subchannels in the subchannel of correspondence bunch with the bit number that each sampling subchannel is assigned with.Like this, all subchannels in each subchannel bunch have identical bit number.All bits all assign.

After all Bit Allocation in Discrete finish, in each grouping, calculate the needed average power signal of each subchannel respectively.Average power signal calculates according to (*) formula.

So far, the assigning process of bit and power is all over.

After the assigning process of bit and power is finished, ofdm system is according to distribution result, on each subchannel, modulate, the power adjustment, and then through the IFFT conversion, add work such as Cyclic Prefix, obtain the OFDM symbol, and with the OFDM symbol-modulated to radio frequency, be transmitted in the channel and transmit.

The key step of whole bit and power division process can be summarized as follows:

The first step according to the system applies environment, and the number of the self adaptation distribution processor that adopts of system, is determined the grouping number N of subchannel _U, and the subchannel bunch subchannel number L that comprises in each grouping.

Second step, to all subchannels,, uniformly-spaced divide according to they positions in frequency spectrum, be divided into N _UIndividual grouping comprises the subchannel of similar number in each grouping.

In the 3rd step, in each grouping, carry out the distribution of bit and power respectively.Bit and power division process in each grouping are as follows:

(1) according to the size of the amplitude of grouping sub-channels gain, to the subchannel in this grouping by from small to large or order from big to small sort.

(2) subchannel after will sorting, the mode by dividing continuously is divided into N _SIndividual sub-cluster of channels (N _S=N _C/ (LN _U)), the subchannel number that comprises in each subchannel bunch is L.

(3) select " least gain " scheme or " reciprocal average " scheme, for each subchannel clusters into a sampling subchannel.

(4) method of distributing by greediness on all sampling subchannels is carried out Bit Allocation in Discrete.

(5) bit number that each sampling is distributed on subchannel duplicates to each subchannel in the subchannel of correspondence bunch.

(6) calculate the needed average power signal of each subchannel respectively.

The assigning process of bit and power finishes.

The present invention is except being used for fixing the communication system of data rate, the needed average power signal of system is minimized, also can be used under the certain condition of data rate and total power signal, make the surplus maximization (system margin is meant the slippage of system's patient signal to noise ratio of institute when keeping certain bit error rate performance, represents the antijamming capability of system) of system.Detailed process can be carried out like this: after bit and power division finish, calculate the current distribution power of each subchannel sum $R_{\mathrm{tot}}={\mathrm{\&Sigma;}}_{i=1}^{N_C}{P}_i,$ Calculate specified total power signal P then _TWith P _TotRatio k=P _T/ P _Tot, promptly $k=\frac{P_T}{{\mathrm{\&Sigma;}}_{i=1}^{N_C}{P}_i}.$ At last, the performance number that each subchannel is distributed

The k that is adjusted into original distribution performance number doubly, promptly $P_i^{\&prime;}={\mathrm{kP}}_i(i=\mathrm{1,2},\&CenterDot;\&CenterDot;\&CenterDot;,N_C),$ Be the final power that distributes of each subchannel.This moment, each had the subchannel of signal transmission to have identical surplus, and this surplus is maximized surplus.

The characteristics of maximum of the present invention are, assigning process can walk abreast in a plurality of groupings and carry out, and the assigning process required time equates with assigning process required time in the grouping.And in each grouping, Bit Allocation in Discrete is not to carry out on all subchannels, but on some sampling subchannels, carry out, the subchannel number and the bit number that need in the assigning process to consider significantly reduce, and the computational complexity in each grouping is very low, and the assigning process required time is very short.Therefore, compare with existing allocation algorithm, great advantage of the present invention is, algorithm complex is very low, and operation time is also very little.Help reducing hardware cost, improve the response time of the self adaptation distribution of hardware.Simultaneously, when the parameter selection was suitable, algorithm assigns result was very approaching with optimum bit and power distribution result, so the performance of algorithm is very near optimum performance.

On the other hand, the complexity of algorithm can be regulated by the parameter in the adjustment algorithm.When guaranteeing that systematic function requires, can come the parameter of adjustment algorithm flexibly according to the quality of channel condition, so that the complexity of algorithm is minimum.Therefore has very high flexibility in actual applications.

For algorithm of the present invention being described in the advantage aspect the complexity, the complexity with algorithm of the present invention and existing two kinds of allocation algorithms compares below.Wherein a kind of algorithm is more famous Hughes-Hartogs algorithm.Another kind is a kind of algorithm simplified on Hughes-Hartogs algorithm basis that is proposed by S.K.Lai.Following table has been listed these several algorithms needed separately number of comparisons, division number of times and the number of times of tabling look-up in assigning process.

M is the sum that system can use modulation system in the last table.As can be seen from the above table, in the middle of system, the complexity of the inventive method is mainly by N _UProduct decision with L.N _UBig more with the product of L, the needed operation times of algorithm is few more, and the complexity of algorithm is low more.In addition, work as N _UWith the product of L fixedly the time, N _UBig more, the needed operation times of algorithm is also few more, and the complexity of algorithm is low more.Work as N _USelect when suitable with L size, two kinds of algorithms of the inventive method and other are compared, and needed computing number again can significantly reduce, and the complexity of algorithm can effectively reduce.Especially when the subchannel number of ofdm system and transmission rate were bigger, the effect that complexity reduces was more obvious.Therefore the present invention more helps in actual central application.

Description of drawings

Fig. 1 is the bit that proposes of the present invention and the flow chart of power distribution algorithm.

Fig. 2 is the schematic diagram that channel is uniformly-spaced divided, divided into groups.

Fig. 3 is in each grouping, the subchannel after the ordering is divided the schematic diagram of subchannel bunch.

Fig. 4 is the performance simulation figure as a result that adopts the algorithm of the present invention of " least gain " scheme and optimum allocation algorithm in open indoor environment.

Embodiment

In FDD (Frequency Division Duplexing (FDD)) communication system, the present invention can be placed on emission bring in realizations, as the part of transmitter function, also can be placed on reception bring in realization, as the part of receiver function.(annotate: when receiver emission data, when transmitter receives data, said here transmitter and receiver role exchange.)

If the present invention is placed on emission and brings in realization, transmitting terminal is pilot signal transmitted at first, after receiving terminal receives pilot signal, the channel gain of all subchannels is estimated, estimates each subchannel noise power-value simultaneously.After estimating to finish, receiving terminal feeds back to transmitting terminal to these estimated informations.Transmitting terminal demodulates after these channel informations, promptly carries out the distribution of bit and power, determine the modulation system of each subchannel and send signal power, then to the data of each subchannel modulate, the power adjustment, send signal in the channel.Simultaneously, transmitting terminal also will send to receiving terminal (adopting the known signal format of receiving terminal) to the distribution result of bit and power.Receiving terminal at first demodulate bit and power distribution result, obtain the information of adaptive modulation parameter, then the Adaptive Modulation signal that receives is carried out demodulation.

Bring in realization if the present invention is placed on to receive, transmitting terminal is pilot signal transmitted at first, after receiving terminal receives pilot signal, channel gain and each subchannel noise power-value of all subchannels is estimated.After estimating to finish, receiving terminal is according to these channel estimating information, carry out the distribution of bit and power, determine the needs of each subchannel adopted when transmitting terminal sent signal modulation system and signal power, then the result of bit and power division is fed back to transmitting terminal (adopting the known signal format of transmitting terminal).Bit and power distribution result that transmitting terminal feeds back according to receiving terminal, to the data of each subchannel modulate, the power adjustment, send signal in the channel.After receiving terminal receives signal, promptly can carry out correct demodulation (this moment, receiving terminal was known to the information of adaptive modulation parameter) to the received signal.

In TDD (time division duplex) communication system, except can adopting the application mode identical with the FDD system, can also adopt following working method: transmitting terminal is at first according to the data-signal or the pilot signal that send over from receiving terminal, channel is estimated, obtained the yield value and the noise power of each subchannel.Carry out the distribution of bit and power then, the needs of each subchannel adopt when determining to send signal modulation system and signal power.Then according to allocation result, modulate and transmit.Simultaneously, transmitting terminal also will send to receiving terminal to the information of Adaptive Modulation and (adopts the known signal format of receiving terminal, make correctly restituted signal of receiving terminal.

The cycle of bit and power adaptive adjustment should be chosen according to speed optimization of real channel variance.

Grouping number N _UWith the selection of the subchannel number L that comprises in each subchannel bunch, should be according under the different application environment, the relation between the performance of algorithm and the complexity of algorithm and the hardware complexity of real system require to determine.The criterion of being followed is under the prerequisite that reaches the systematic function requirement, to make the complexity of algorithm minimum.During practical application, should by measuring or emulation, draw N when different pieces of information speed at first at the channel transfer characteristic of applied environment _UAnd the relation between L and the systematic function.According to the systematic function requirement, take all factors into consideration the requirement of system hardware complexity then, be chosen at and satisfy under the systematic function requirement condition, make the minimum N of system complexity _UValue with L.

N _UWith the selection of L, general earlier definite N _UN _UMainly the hardware complexity by system determines.Owing to can handle a bit and a power division in the grouping with a self adaptation distribution processor in the middle of actual, so N _UMaximum generally can be taken as the number of self adaptation distribution processor in the system.The number of self adaptation distribution processor generally can be got the integral number power of 1,2,4,8 grades 2 in the system.

Determine N _UAfterwards, just can be by under specific environment, N _UAnd the relation between L and the systematic function is determined the value of L.According to N under specific environment _UAnd the relation curve between L and the systematic function, can find the value of the L that satisfies the maximum that systematic function requires, as the actual parameter value of using of algorithm.In general, the value of L can be taken as 1,2, the integral number power of 4,8 grades 2.

In the middle of the practical application, N _UCan adjust along with the variation of channel transfer characteristic with L.The criterion of adjusting also is under the prerequisite that satisfies the systematic function requirement, makes the complexity of algorithm minimum.For example, the frequency selectivity characteristic when channel dies down, when the correlated bandwidth of channel broadens, can increase N _UAnd L, to reduce system complexity.And, may need to reduce N when the frequency selectivity characteristic grow of channel, when the correlated bandwidth of channel narrows down _UWith L to satisfy the systematic function requirement.These adjustment need be estimated the transmission characteristic of channel.

Below in conjunction with an example, the implementation of algorithm is described.

The major parameter of an ofdm system is as follows:

Number of sub carrier wave (N C)	512
		Data rate	400Mbps
Target error rate	10 -5
		Each allocation bit sum (R b)	2560
Available modulation system	0，QPSK，16QAM，64QAM，256QAM

At first, need to determine N _UValue with L.

For reducing complexity, when generating the sampling subchannel, system adopts " least gain " scheme.Consider the application of this system in open indoor environment.When carrying out Computer Simulation, the channel model of employing is open indoor environment channel model.By Computer Simulation, obtain when " least gain " scheme of employing, system reaches target error rate 10 ^-5Ratio (the E of needed bit energy and noise power spectral density _b/ N ₀), with N _UAnd the relation curve between the L, as shown in Figure 4 in the Figure of description.The performance curve (with the curve of square frame mark) of system when also having marked the bit that reaches optimum and power distribution result among the figure.

According to the requirement of hardware complexity in the real system, the number of self adaptation distribution processor is 4 in the system, therefore can select N _U=4.Determined N _UValue after, again according to the systematic function requirement, determine the value of L.If system requirements is when reaching target error rate, needed E _b/ N ₀Value be no more than 19dB, N in last figure _UOn that of=4 curve, find and satisfy E _b/ N ₀The value of the L of the maximum during＜19dB is 2 ⁴=16, then select L=16.If require needed E _b/ N ₀Value be no more than 18.6dB, N in last figure _UOn that of=4 curve, find and satisfy E _b/ N ₀The value of the L of the maximum during＜18.6dB is 2 ³=8, then select L=8.The value of supposing system's selection L in the reality is 16.

Determined N _UAfter the value of L, system can carry out the distribution of bit and power.In the beginning of each assigning process,, obtain the range value of the channel gain of 512 subchannels at first by channel estimating.With these 512 channel amplitudes, the mode (referring to accompanying drawing 2) according to uniformly-spaced dividing is divided into 4 groups, and every group comprises N _CA '=512/4=128 subchannel.Then,, send to respectively in 4 self adaptation distribution processor, in 4 processors, carry out the distribution of bit and power simultaneously 128 channel amplitudes that 4 groupings are comprised separately.In each processor, at first 128 channel amplitudes are sorted, the channel amplitude after will sorting again, the mode (referring to accompanying drawing 3) according to continuous division is divided into N _S=128/16=8 sub-cluster of channels, each subchannel bunch comprise 16 subchannels.Cluster into the sampling subchannel of a correspondence then for each subchannel.According to " least gain " scheme, calculate the channel gain amplitude of each sampling subchannel.Next, on the sampling subchannel,, carry out Bit Allocation in Discrete according to greedy distribution method.Divide timing it is noted that and carry out Bit Allocation in Discrete in each processor, the bit number of actual required distribution is R _b ^*=2560/ (4 * 16)=40.Can use modulation system according to system, the step-length that can obtain Bit Allocation in Discrete is d=2, and the maximum permission of each subchannel bit number is 8.After Bit Allocation in Discrete finishes,, duplicate respectively to 16 subchannels in the subchannel of correspondence bunch with the bit number of 8 samplings on the subchannels.At last, be respectively each subchannel, calculate needed separately transmitted power.Like this, bit and power division process are finished.

Can see that from available modulation system the step-length of Bit Allocation in Discrete is d=2, available modulation system adds up to M=5, and the maximum permission of each subchannel bit number is D=8.Following table has provided N _U=4, during L=16, algorithm of the present invention and Hughes-Hartogs algorithm and S.K.Lai algorithm computational complexity separately.

Allocation algorithm	The Hughes-Hartogs algorithm complex	The S.K.Lai algorithm complex	Algorithm complex of the present invention (in each grouping)
				Number of comparisons	654,080	9728	1036
The division number of times	1791	1284	131
				The number of times of tabling look-up	1791	1284	11

As seen from the above table, algorithm by the present invention's proposition, each needed computational complexity in self adaptation allocation process unit reduces greatly, then the hardware complexity of each self adaptation distribution processor can reduce greatly, the response time of self adaptation distribution processor significantly reduces, and systematic function approaches optimum performance.Therefore algorithm of the present invention is suitable for actual central application.

Claims (8)

1, a kind of bit and power distribution method that is used for the ofdm system Adaptive Modulation, key step comprises: at first, according to the subchannel total number N of system applies environment, required allocation bit and power _C, required distribution total number of bits R _b, and the number of the self adaptation distribution processor that adopts of system is determined the grouping number N of subchannel _U, and the subchannel bunch subchannel number L that comprises in each grouping; Secondly, to all N _CThe individual subchannel that needs allocation bit and power according to they positions in frequency spectrum, is uniformly-spaced divided, and is divided into N _UIndividual grouping comprises the subchannel of similar number in each grouping; Then, in each grouping, carry out the distribution of bit and power respectively; In each grouping, carry out the branch timing of bit and power, at first according to the size of the amplitude of subchannel gains, to the subchannel in this grouping by from small to large or order from big to small sort; In each grouping, with the subchannel after the ordering, the mode by dividing continuously is divided into N then _SIndividual sub-cluster of channels N _S=N _C/ (LN _U), the subchannel number that comprises in each subchannel bunch is L; Next cluster into a sampling subchannel for each subchannel, the scheme that generates the sampling subchannel requires to select " least gain " scheme and one of " reciprocal average " scheme according to practical application performance and complexity, wherein " least gain " scheme is to get in each subchannel bunch, minimum value in the gain range value of all subchannels, range value as the sampling subchannel gains of this subchannel bunch correspondence, " reciprocal average " scheme is to get in each subchannel bunch, the mean value of the inverse of the gain range of all subchannels is as the reciprocal value of the sampling subchannel gains amplitude of this subchannel bunch correspondence; Then the method for distributing by greediness on all sampling subchannels is carried out Bit Allocation in Discrete, and when carrying out Bit Allocation in Discrete by the method for greediness distribution, the bit number of actual required distribution is R _b/ (LN _U); After Bit Allocation in Discrete is intact,, duplicate, calculate the needed signal power of each subchannel at last respectively to each subchannel in the subchannel of correspondence bunch with the bit number that distributes on each sampling subchannel.

2, a kind of bit and power distribution method that is used for the ofdm system Adaptive Modulation according to claim 1, it is characterized in that: when needing the subchannel of allocation bit to divide into groups all, according to they positions in frequency spectrum, uniformly-spaced divide, be divided into several groupings, comprise the subchannel of similar number in each grouping.

3, a kind of bit and power distribution method that is used for the ofdm system Adaptive Modulation according to claim 1, it is characterized in that: when in each grouping, dividing subchannel bunch, need earlier the subchannel in the grouping to be sorted, ordering is the size according to the amplitude of subchannel gains, according to carrying out with one of them order from small to large from big to small.

4, a kind of bit and power distribution method that is used for the ofdm system Adaptive Modulation according to claim 1, it is characterized in that: when dividing subchannel bunch, be in each grouping, in the subchannel after the ordering, adjacent several subchannels are divided into cluster, and each subchannel bunch comprises the subchannel of similar number.

5, a kind of bit and power distribution method that is used for the ofdm system Adaptive Modulation according to claim 1, it is characterized in that: ordering and dividing after subchannel bunch finishes, need cluster into " the sampling subchannel " of a correspondence for each subchannel, the gain of sampling subchannel is the function of the gain of all subchannels in the corresponding subchannel bunch, the generation of this function requires to select " least gain " scheme and one of " reciprocal average " scheme according to practical application performance and complexity.

6, a kind of bit and power distribution method that is used for the ofdm system Adaptive Modulation according to claim 1, it is characterized in that: after generation sampling subchannel finishes, on the sampling subchannel in each grouping, carry out Bit Allocation in Discrete respectively according to greedy allocation algorithm.

7, a kind of bit and power distribution method that is used for the ofdm system Adaptive Modulation according to claim 1 is characterized in that: when carrying out Bit Allocation in Discrete on the sampling subchannel in each grouping, the total number of bits of actual required distribution is R in each grouping _b/ (LN _U), R wherein _bBe total number of bits to be allocated, L is each subchannel bunch number of subchannels that comprises, N _UIt is grouping number.

8, a kind of bit and power distribution method that is used for the ofdm system Adaptive Modulation according to claim 1, it is characterized in that: in each grouping, after the process of Bit Allocation in Discrete finishes, with each bit number of being assigned with of sampling subchannel, duplicate to all subchannels in the subchannel of correspondence bunch, calculate the needed average power signal of each subchannel then respectively.

Images