CN1604511A

CN1604511A - Multi-antenna-OFDM communication system adaptive power allocation method

Info

Publication number: CN1604511A
Application number: CN 200410068029
Authority: CN
Inventors: 周军; 章飞舟; 张海滨; 宋文涛
Original assignee: Shanghai Jiao Tong University
Current assignee: Shanghai Jiao Tong University
Priority date: 2004-11-11
Filing date: 2004-11-11
Publication date: 2005-04-06

Abstract

It is a multi-antenna crossing frequency division multiplexing communication system self-adapting power alignment method used in wireless communication field, which comprises the following steps: the transmitting end first aligns transmitting power of each transmitting power based on injection algorism in the communication system; it must close negative power sub-channel and process the transmit bit number for integral process in that each sub-channel only transmit integral bits; it then again aligns the remain power to final alignment results based on greed algorism according to bit transmission maximum rule.

Description

The method that the multi-antenna OFDM communication system adaptive power distributes

Technical field

The present invention relates to the method that a kind of adaptive power distributes, specifically is the method that a kind of multi-antenna OFDM communication system adaptive power distributes, and is used for wireless communication field.

Background technology

Many antennas (MIMO) system is a kind of efficient wireless transmitting system that grows up recent years, compare with traditional single antenna (SISO) system, its distinguishing feature is to adopt a plurality of emissions and reception antenna to constitute a plurality of mutual incoherent wireless channels between transmitter and receiver.OFDM (OFDM) is a kind ofly to change high-speed serial data into the low-speed parallel data, and modulates parallel modulation system respectively with a plurality of mutually orthogonal subcarriers.At present, the MIMO-OFDM system that both is combined has obtained in fields such as WLAN (wireless local area network) using widely.

In the MIMO-OFDM system, because each subcarrier uses identical modulation, for reaching the requirement of error rate of system, the transmission symbol number that need reduce all subcarriers can cause overall bit-error rate to increase greatly to avoid the decline of channel partial-depth, thereby has reduced the availability of frequency spectrum of system.The adaptive power distribution technique can improve the availability of frequency spectrum of system, reduces the total transmitting power and the error rate.Existing multicarrier adaptive power allocation algorithm is under the prerequisite of given certain system restriction condition (for example each subcarrier error rate is certain) according to the algorithm purpose, make Bit Allocation in Discrete reach the maximized problem of speed, total be exactly the certain situation of transmitting power under, make the speed maximum of whole channel.

At present, distribute about adaptive power, the main algorithm of proposition comprises: the greedy algorithm of water-filling algorithm, discontinuous granularity optimum (Hughes-Hartogs algorithm), minimum bit-error rate method.Greedy algorithm has provided the optimal solution of the transmitted bit number in the integer range, and complexity is very high.Water-filling algorithm is based on the different situations of Lagrangian and current each subchannel, and the subchannel big in signal to noise ratio distributes more signal energy; The subchannel little in signal to noise ratio distributes less signal energy.It is constant that such distribution will keep signal energy and noise energy sum in each subchannel, all has been assigned with up to all channel energy.This algorithm has provided the optimal solution that the transmitted bit number distributes in the real number scope, but when the transmitted bit number was non-integer, the realization of system had very big difficulty.

Find by prior art documents, people such as R F H Fischer are at Proc.IEEE Gobecom[C] the minimum bit-error rate method that proposes in " the A new loading algorithm for discrete multitonetransmission[A] " that deliver on the .1996 pp724-728 (international IEEE " Global Communications System " proceeding in 1996) (a kind of new multicarrier bit loading algorithm "), this algorithm improves on the basis of water-filling algorithm.The key of algorithm is to be target with the error sign ratio minimum, has sacrificed a part of systematic function to reduce algorithm complex.Therefore, this algorithm remains sub-optimal algorithm.

Summary of the invention

The object of the invention is to overcome the deficiencies in the prior art, has proposed the method that a kind of multi-antenna OFDM communication system adaptive power distributes.Make its basic thought, utilize Lagrangian constant method,, make under the certain prerequisite of gross power, average error rate, transmit total bit number big a kind of sub-optimal algorithm of trying one's best each subchannel allocation bit number and power according to water-filling algorithm.Simulation result proves that the performance of this method and greedy algorithm is close, but computational complexity reduces greatly, has very strong using value.

The present invention is achieved by the following technical solutions, in multi-antenna OFDM communication system, transmitting terminal at first carries out primary distribution according to the water-filling algorithm basic thought to the transmitting power of each subchannel, because each subchannel transmission power must be for just and in actual channel transmitting the integer bit, simultaneously the bit number of transmission is carried out the integer processing so must close the negative power subchannel, count the criterion of maximum subsequently according to bit transfer, based on greedy algorithm, dump power is carried out sub-distribution again, obtain final allocation result.

Described MIMO-OFDM wireless transmitting system, at transmitting terminal, at first the serial data with input obtains parallel data through serial to parallel conversion, distribute power and bit number for each subchannel according to the adaptive power allocation algorithm to the estimated result of channel subsequently, through ovennodulation, after precoding and the OFDM, signal is transferred to radio frequency, and send signal by different transmitting antennas; At receiving terminal, the baseband signal that receives is separated mediation filtering by OFDM, estimate this channel characteristics constantly according to training sequence subsequently, subchannel Bit Allocation in Discrete that provides according to transmitting terminal and power allocation information be the signal demodulation, at last the information after the demodulation obtained dateout after by parallel serial conversion.

Below the present invention is further illustrated, concrete steps are as follows:

1) according to Lagrangian the power of system's sub-carriers is carried out initialization process

p_{k, n} = \frac{- n_{0} (k, n) * Γ}{{| λ_{k, n} |}^{2}} + \frac{P}{N * M} + \frac{n_{0} (k, n) * Γ}{N * M} Σ_{k = 1}^{M} Σ_{n = 1}^{N} \frac{1}{{| λ_{k, n} |}^{2}} - - - (3)

λ in the formula _{K, n}The channel characteristic value of representing n subcarrier correspondence of k antenna, P represents transmitting antenna gross power, n ₀(k, n) the multiple Gaussian noise of the independent same distribution of n subcarrier correspondence of k antenna

2) for p arbitrarily _{K, n}＜0, represent that this subcarrier respective channels condition is not suitable for transmission signals, it is changed to zero, i.e. p _{K, n}=0.

3) calculate sub-carrier signal-noise ratio SNR _{K, n}, calculate the bit number of this subcarrier allocation according to formula (4), to  _{K, n}The rounding operation that rounds up gets b _{K, n}, and will be higher than b _MaxB _{K, n}Be made as b _Max

{\hat{b}}_{k, n} = \log_{2} (1 + \frac{{SNR}_{k, n}}{Γ}) - - - (4)

SNR in the formula _{K, n}Represent n subcarrier correspondence of k antenna signal to noise ratio,  _{K, n}The subchannel bit number of representing n subcarrier correspondence of k antenna, b _{K, n}' expression  _{K, n}The value that obtains after rounding.

4) according to existing b _{K, n}' sequence is utilized

{SNR}^{'}_{k, n} = \frac{(2^{{b_{k, n}}^{'}} - 1) * n_{0} (k, n) * Γ}{{| λ_{k, n} |}^{2}} - - - (5)

Calculate the actual gross power P ' that sends, and calculated difference power Δ P=P-P '.

5), form following power increment table to each subcarrier:

{Δp}_{k, n} (b) = p_{k, n} (b) - p_{k, n} (b - 1) = \frac{2^{b_{k, n} - 1} * n_{0} (k, n) * Γ}{{| λ_{k, n} |}^{2}} - - - (6)

P in the formula _{K, n}(b) the corresponding power of k antenna n of expression sub-carrier transmit b bit, Δ p _{K, n}(b) the power difference of k antenna n of expression sub-carrier transmit b or b-1 bit.

6) according to this power increment table and greedy algorithm, at first find out the subchannel of power increment maximum in all subchannels, and the bit number that this subchannel transmitted reduced accordingly, upgrade the total transmitted bit number of difference power Δ P and corresponding change simultaneously, until difference power Δ P for just.

7) according to the power increment table, find out the subchannel of power increment minimum in all subchannels, and the bit number that this subchannel transmitted increased accordingly, upgrade the total transmitted bit number of difference power Δ P and corresponding change simultaneously, in current subchannel, till the power Minimum Increment, obtain ultimate sequence b _{K, n}

The inventive method also is applicable to the adaptive bit allocation algorithm, promptly under the certain condition of the total bit number of emission, makes the minimized problem of transmitting power.Specifically describe as follows:

Target function

\min (Σ_{k = 1}^{M} Σ_{n = 1}^{N} p_{k, n})

Constraints

Σ_{k = 1}^{M} Σ_{n = 1}^{N} b_{k, n} = B,

p_{k, n} = (2^{b_{k, n}} - 1) \cdot Γ \cdot n_{0}

p _k，n≤p _max

b _k，n≥0，k＝1，2…，M，n＝1，2…，N

B wherein _{K, n}Be the bit number of the N subcarrier of k antenna, p _{K, n}Be the power of the N subcarrier of k antenna, B is a total emission power, and Γ is an adjustable parameter, represents the different coded systems and the relevant error rate.

According to the said method step 1), can get:

b_{k, n} = \log_{2} \frac{2^{\frac{B}{M}} \cdot {| λ_{k, n} |}^{2}}{\sqrt[MN]{{\overset{n = 1,2, . . . N}{Π}}_{k, n}^{k = 1,2, M} {| λ_{k, n} |}^{2}}} - - - (7)

For b arbitrarily _{K, n}＜0, represent that this subcarrier respective channels condition is not suitable for transmission signals, it is changed to zero, i.e. b _{K, n}=0.This moment, total bit number of transmission should be greater than the total bit number B of initial setting.Form the power increment table according to above-mentioned formula (6), subsequently according to this table, at first find out the subchannel of power increment maximum in all subchannels, and the bit number that this subchannel transmitted reduced accordingly, change total transmitted bit number simultaneously, equal bits of original until the transmitted bit number and count B.

With the adaptive power algorithm is example, the present invention is based on the basic thought of water-filling algorithm, and power is carried out sub-distribution just, is (O (Nlog based on complexity subsequently ₂(N)) greedy algorithm) is reallocated to dump power.Under reality, dump power is far smaller than initial power, so the present invention can reduce operation time greatly.Simultaneously, be optimal algorithm, so the present invention can approach optimal algorithm on performance based on water-filling algorithm and greedy algorithm.

The simulation result proof is under speed maximization criterion, and under the condition of error rate basically identical, the present invention is computation amount not only, and reach the performance of greedy algorithm substantially, the present invention generally is applicable to the multi-channel wireless communication system, as mimo system, and ofdm system etc.

Description of drawings

Fig. 1 is a MIMO-OFDM emitting-receiving system block diagram

Fig. 2 is an OFDM emitting-receiving system block diagram

Fig. 3 is a MIMO emitting-receiving system block diagram

Fig. 4 is Γ=10 ²The time greedy algorithm and the transmitted bit number of the present invention and the error rate comparison

Fig. 5 is Γ=10 ¹The time greedy algorithm and the transmitted bit number of the present invention and the error rate comparison

Fig. 6 is under the different capacity initial condition, and greedy algorithm and operation time of the present invention are relatively

Embodiment

As shown in Figure 1, the MIMO-OFDM system, at transmitting terminal, at first the serial data with input obtains parallel data through serial to parallel conversion, distribute power and bit number for each subchannel according to the adaptive power allocation algorithm to the estimated result of channel subsequently, through ovennodulation, after precoding and the OFDM, signal is transferred to radio frequency, and send signal by different transmitting antennas.

At receiving terminal, the baseband signal that receives is separated mediation filtering by OFDM, estimate this channel characteristics constantly according to training sequence subsequently, subchannel Bit Allocation in Discrete that provides according to transmitting terminal and power allocation information be the signal demodulation, at last the information after the demodulation obtained dateout after by parallel serial conversion.

As shown in Figure 2, at transmitting terminal, at first the serial data with input obtains parallel data through serial to parallel conversion, give each subcarrier allocation power according to the adaptive power allocation algorithm to the estimated result of channel subsequently, then successively through ovennodulation, precoding, IFFT, add cyclic prefix CP, be modulated to intermediate frequency and radio frequency.

At receiving terminal, received signal after removing cyclic prefix CP, successively by FFT and filtering, demodulation, and string changes and can obtain dateout.

As shown in Figure 3, at transmitting terminal, at first the serial data with input obtains parallel data through serial to parallel conversion, distributes power for each subchannel according to the adaptive power allocation algorithm to the estimated result of channel subsequently, launches through after the precoding then.

At receiving terminal, the data that receive estimate this channel characteristics constantly according to training sequence subsequently earlier by filtering, and the channel information of renewal adaptive power allocation algorithm obtains dateout with the information after the demodulation after by parallel serial conversion then.

Fig. 4 and Fig. 5 represent respectively when Γ=10 ²And Γ=10 ¹The time greedy algorithm and the transmitted bit number of the present invention and the error rate comparison

Fig. 6 is that greedy algorithm and operation time of the present invention are relatively under the expression different capacity initial condition

Operation time of the present invention is greater than greedy algorithm when the power initial value is very little, this is because the difference power Δ P that obtains after the first power division when initial power is very little when will be greater than the initial value power P, thus next difference power Δ P is carried out operation time of greedy algorithm will be direct greater than the computing of the initial value power P directly being carried out greedy algorithm.Increase to certain the time when the initial value power P subsequently, the difference power Δ P that obtains will be much smaller than the initial value power P.

Content below in conjunction with accompanying drawing and the inventive method provides following examples:

Method of the present invention can be carried out the adaptive power Bit Allocation in Discrete to the subchannel of multiple channel wireless communication system, both can be applied to the MIMO-OFDM system, also can be applied to MIMO, the OFDM wireless local area network (WLAN) system.Can illustrate with the system functional block diagram in following embodiment one and the example two.

Embodiment one: the MIMO-OFDM system

As shown in Figure 1.Consider a MIMO-OFDM wireless communication system that M transmitting antenna and N root reception antenna are arranged.Total transmitting power of supposing system is P.At transmitting terminal, determine the value of adjustable parameter Γ according to the error rate of system's specific requirement; Then, the channel matrix that receiving terminal is fed back to carries out singular value decomposition, obtains channel characteristic value, premultiplication matrix and right multiply matrix; Calculate the initial value of each sub-channel power distribution and Bit Allocation in Discrete again by formula (3) (4), close the minus subchannel of initial value, and utilize greedy algorithm to dump power sub-distribution again, and all assigned up to all power, generate complete bit and power division table.System to each Channel Modulation, generates complete MIMO-OFDM symbol according to the bit number of pairing each subchannel of this form and power, with right multiply matrix to this encoding symbols.At last, the symbol that encodes is modulated through OFDM, added Cyclic Prefix, and be modulated to radio-frequency transmissions in wireless channel.

At receiving terminal, the signal that receives is modulated to base band, and carries out corresponding symbol, frequency, regularly synchronous and OFDM rectification.Then, according to estimating good channel condition information (identical), adopt and the similar method generation of transmitting terminal premultiplication matrix, right multiply matrix, characteristic value and bit power allocation table with the channel condition information that feeds back to transmitting terminal.Take advantage of with the premultiplication matrix right side and to separate the OFDM symbol that mixes up, antithetical phrase channel-decoding.According to the power of each subchannel of bit power allocation table normalization, and separate and be mapped as concrete bit stream.At last, the bit stream that several are parallel is merged into serial bit stream, sends follow-up baseband module to, obtains dateout.

Implement arithmetic result of the present invention in the MIMO-OFDM system shown in Fig. 4,5,6, determined that again after the adjustable parameter Γ, along with the raising of initial gross power, the error rate remains unchanged substantially, the transmitted bit number constantly increases and can approach optimum greedy algorithm.Simultaneously, along with the raising of initial power, be exponential growth the operation time of greedy algorithm, and algorithm of the present invention the time ask constant substantially.

Embodiment two: the OFDM WLAN (wireless local area network)

As shown in Figure 2, system framework is similar to top MIMO-OFDM system.

Total transmitting power of supposing system is P.At transmitting terminal, determine the value of adjustable parameter Γ according to the error rate of system's specific requirement; Then, the channel matrix that receiving terminal is fed back to carries out singular value decomposition, obtains channel characteristic value, premultiplication matrix and right multiply matrix; Calculate the initial value of each sub-channel power distribution and Bit Allocation in Discrete again by formula (3) (4), close the minus subchannel of initial value, and utilize greedy algorithm to dump power sub-distribution again, and all assigned up to all power, generate complete bit and power division table.Use right multiply matrix to this encoding symbols subsequently, last, the symbol that encodes is modulated through OFDM, add Cyclic Prefix, and be modulated to radio-frequency transmissions in wireless channel.

The same in the MIMO-OFDM system that coexists, in the OFDM WLAN (wireless local area network), the present invention also can obtain good performance.

Claims

1. A method for adaptive power allocation of a multi-antenna-OFDM communication system, characterized in that, in the multi-antenna-OFDM communication system, the transmitting end at first performs a power allocation for each according to the basic idea of the water injection algorithm The transmission power of subchannels is allocated once. Since the transmission power of each subchannel must be positive and only integer bits can be transmitted in the actual channel, it is necessary to close the negative power subchannel and carry out integer processing on the number of transmitted bits, and then according to the bit The criterion of the maximum number of transmissions is based on the greedy algorithm, and the remaining power is redistributed to obtain the final distribution result.

2. The method for adaptive power allocation in a multi-antenna-OFDM communication system according to claim 1, wherein it is further limited by the following steps:

1) Initialize the power of the subcarriers in the system according to the Lagrangian operator

{p p}_{k k,, n no} = = \frac{- - {n no}_{00} ((k k,, n no)) * * Γ Γ}{{| | {λ λ}_{k k,, n no} | |}^{22}} + + \frac{P P}{N N * * M m} + + \frac{{n no}_{00} ((k k,, n no)) * * Γ Γ}{N N * * M m} {Σ Σ}_{k k = = 11}^{M m} {Σ Σ}_{n no = = 11}^{N N} \frac{11}{{| | {λ λ}_{k k,, n no} | |}^{22}}

where λ _{k, n} represent the channel eigenvalue corresponding to the nth subcarrier of the kth antenna, P represents the total power of the transmitting antenna, n ₀ (k, n) the independent and identically distributed complex Gaussian distribution corresponding to the nth subcarrier of the kth antenna noise;

2) For any p _{k, n} <0, it means that the channel condition corresponding to the subcarrier is not suitable for transmitting signals, and it is set to zero, that is, p _{k, n} =0;

3) Calculate the subcarrier signal-to-noise ratio SNR _{k, n} , calculate the number of bits allocated to the subcarrier according to the following formula, perform rounding operations on  _{k, n} to get b _{k, n} , and set b higher than b _max _k,n is set to b _max ,

{\overset{^^}{b b}}_{k k,, n no} = = {log log}_{22} ((11 + + \frac{{SNR SNR}_{k k,, n no}}{Γ Γ}))

In the formula, SNR _{k, n} represents the signal-to-noise ratio corresponding to the nth subcarrier of the kth antenna,  _{k, n} represents the number of subchannel bits corresponding to the nth subcarrier of the kth antenna, b _{k, n} represents  _k, The value obtained after rounding _n ;

4) According to the existing b _{k, n} sequence, use

{SNR SNR}^{' '}_{k k,, n no} = = \frac{((22^{{b b}_{k k,, n no}^{' '}} - - 11)) * * {n no}_{00} ((k k,, n no)) * * Γ Γ}{{| | {λ λ}_{k k,, n no} | |}^{22}}

Calculate the actual total power P' sent, and calculate the difference power ΔP=P-P';

5) For each subcarrier, form the following power increment table:

{Δp Δp}_{k k,, n no} ((b b)) = = {p p}_{k k n no} ((b b)) - - {p p}_{k k,, n no} ((b b - - 11)) = = \frac{22^{{b b}_{k k n no} - - 11} * * {n no}_{00} ((k k,, n no)) * * Γ Γ}{{| | {λ λ}_{k k,, n no} | |}^{22}}

where p _kn (b) represents the corresponding power of b bits transmitted by the nth subcarrier of the kth antenna, and Δp _kn (b) represents the power difference of b or b-1 bits transmitted by the nth subcarrier of the kth antenna ;

6) According to the power increment table and the greedy algorithm, first find out the subchannel with the largest power increment among all subchannels, and reduce the number of bits transmitted by the subchannel accordingly, and update the difference power ΔP and the corresponding Change the total number of transmitted bits until the difference power ΔP is positive;

7) According to the power increment table, find out the subchannel with the smallest power increment among all subchannels, and increase the number of bits transmitted by the subchannel accordingly, and update the difference power ΔP and change the total transmission bits accordingly number until the minimum power increment in the current subchannel, to obtain the final sequence b _k,n .

3. The method for adaptive power allocation of multi-antenna-OFDM communication system according to claim 1 or 2, characterized in that it is applicable to the adaptive bit allocation algorithm at the same time, that is, when the total number of transmitted bits is certain Under the condition of , the transmission power is minimized, which is specifically described as follows:

objective function

\min (Σ_{k = 1}^{m} Σ_{no = 1}^{N} p_{k, no})

Restrictions

Σ_{k = 1}^{m} Σ_{no = 1}^{N} b_{k, no} = B,

{p p}_{k k,, n no} = = ((22^{{b b}_{k k,, n no}} - - 11)) \cdot &Center Dot; Γ Γ \cdot &Center Dot; {n no}_{00}

p _{k, n} ≤ p _max

b _{k, n} ≥ 0, k=1, 2..., M, n=1, 2..., N

where b _k,n is the number of bits of the Nth subcarrier of the kth antenna, pk _,n is the power of the Nth subcarrier of the kth antenna, B is the total transmission power, and Г is an adjustable parameter, indicating the encoding mode and the associated bit error rate;

According to step 1), get:

{b b}_{k k,, n no} = = {log log}_{22} \frac{22^{\frac{B B}{M m}} \cdot &Center Dot; {| | {λ λ}_{k k,, n no} | |}^{22}}{\sqrt[MN MN]{{\overset{n no = = 1,2 1,2,, . . . . N N}{Π Π}}_{k k,, n no}^{k k = = 1,2 1,2 . . M m} {| | {λ λ}_{k k,, n no} | |}^{22}}}

For any b _{k, n} < 0, it means that the channel condition corresponding to the subcarrier is not suitable for transmitting signals, and it is set to zero, that is, b _{k, n} = 0. At this time, the total number of bits transmitted should be greater than the initial set total The number of bits B, according to the formula

{Δp}_{k, no} (b) = p_{k, no} (b) - p_{k, no} (b - 1) = \frac{2^{b_{k, no} - 1} * {no}_{0} (k, no) * Γ}{{| λ_{k, no} |}^{2}}

Form a power increment table, and then according to the table, first find out the subchannel with the largest power increment among all subchannels, and reduce the number of bits transmitted by the subchannel accordingly, and change the total number of transmitted bits at the same time until The number of transmitted bits is equal to the initial number B of bits.