CN1610285A - Method for estimating communication channel in OFDM system - Google Patents

Abstract

The channel estimation method in OFDM system includes: computing in available algorithm to obtain pilot subcarrier channel estimating value sequence Hp1; inserting NumPilotVSC 0's into Hp1 to obtain sequence Hp2; performing IFFT transformation on Hp2 to obtain CIR1; performing left half window processing to the specific part of CIR1 to obtain CIR2; insterting 0's specific positions of CIR2 to obtain CIR3; performing FFT transformation on CIR3 to obtain CFR1; right shifting CFR1 circularly to obtain all the subcarrier channel estimating values CFR2 in OFDM system. The other method has partial steps in the foregoing method altered. The present invention has improved channel estimating capacity in OFDM system, no increased complexity and high practical value.

Description

The method of channel estimating in the ofdm system

Technical field

The present invention relates to adopt the method for the channel estimating in the ofdm system of Comb Pilot symbol.

Background technology

The OFDM technology has advantages such as spectral efficient, anti-multipath effect, and therefore it will be a very promising technology in the radio communication in future.Adopt the ofdm system of qam mode to have higher data rate and spectrum efficiency, but must adopt coherent demodulation technology to realize the reception of ofdm signal simultaneously, this just need carry out parameter Estimation and tracking to wireless channel.

In general, the auxiliary ofdm system of pilot tone has two types pilot frequency mode, is exactly block pilot frequency mode and Comb Pilot pattern.For block pilot frequency mode, pilot sub-carrier is inserted in the subcarrier of all OFDM with certain cycle in time domain.Such pilot frequency mode is applicable in the wireless channel environment of slow variation, as the HIPERLAN/2 system.For the Comb Pilot pattern, pilot sub-carrier is inserted in the subcarrier of each OFDM symbol with certain interval.Such pilot frequency mode is applicable in the fast-changing wireless channel environment.This pilot frequency mode of two types is shown in Fig. 1 (a), Fig. 1 (b).

The Comb Pilot channel estimating is made of the algorithm of pilot sub-carrier channel estimation method and channel interpolation.In general, in the channel estimating apparatus of Comb Pilot type, several methods that can finish the channel interpolation are arranged now.These methods are: linear interpolation method, the second order interpolation method, the low-pass filtering interpolation method, spline method and traditional time domain interpolation method, document 1 (" channel estimation technique of arranging based on pilot tone in the ofdm system ", Sinem Coleri, Mustafa Ergen see reference, IEEE about broadcast communication can report, September the 48th in 2002, No. 3, volume, Sinem Coleri, Mustafa Ergen, " Channel Estimation Techniques Basedon Pilot Arrangement in OFDM Systems ", IEEE Transactions On Broadcasting, Vol.48, No.3, September, 2002).

The time domain interpolation algorithm of mentioning in list of references 1 is traditional time domain interpolation algorithm, for traditional time domain interpolation algorithm, using as LS, LMMSE, after the MMSE scheduling algorithm obtains the channel estimating of pilot sub-carrier, with IDFT conversion (inverse discrete fourier transform) the pilot sub-carrier channel estimating is transformed in the time domain, obtains being similar to the time-domain signal of channel impulse response (CIR).Insert behind this signal some 0 after, carry out DFT again and be transformed into frequency domain, just finished simultaneously interpolation.

List of references 2 (" the new channel estimation methods that is used for OFDM of the mobile communication system of handling based on pilot signal and transform domain ", Yuping Zhao, Aiping Huang, IEEE VTC ' 97,1997, pp.2089-2094; Yuping Zhao, AipingHuang, " A Novel Channel Estimation Method for OFDM Mobile Communication Systems Based onPilot Signals and Transform-Domain Processing ", IEEE VTC ' 97,1997, pp.2089-2094.) mentioned the transform domain interpolation algorithm of traditional channel estimating, transform domain interpolation method for traditional channel estimating, passing through as LS, LMMSE, after the MMSE scheduling algorithm obtains the channel estimating of pilot sub-carrier, with the DFT conversion pilot sub-carrier channel estimating is transformed into the transform domain frequency domain of frequency domain (or be called), the signal of acquisition transform domain again.Behind the signal centre position of zero insertion after the conversion, carry out the IDFT conversion again, change back frequency domain, the while has just been finished interpolation process.

We know, when ofdm system designed, the maximum delay of the multipath wireless channel that the length of Cyclic Prefix in the ofdm system (CP) should be used than ofdm system was big.Traditional transform domain channel estimating is compared better performances in the less relatively multipath channel models (such as maximum delay is than half also little situation of CP) at the CP of maximum delay and ofdm system, but, for those maximum delays compare with CP relatively large multipath channel models (such as, maximum delay is littler than CP length, but near CP length), traditional transform domain performance for estimating channel can reduce greatly.

Because be subjected to the influence of virtual subnet carrier wave, the performance of traditional time-domain channel estimating method can not satisfy the requirement of OFDM channel estimating.

Summary of the invention

Channel estimation methods in the ofdm system that the present invention proposes, when improving performance, the not extra complexity that increases.

The method of channel estimating in a kind of ofdm system, the length of FFT/IFFT (NFFT) equals the number (NumNonnegaPilot) of non-negative pilot sub-carrier, the number (NumPilotVSC) of pilot frequency locations in the virtual subnet carrier wave and number (NumNegaPilot) sum of negative pilot sub-carrier divided by pilot interval (PilotInterval) in the described ofdm system, and described method comprises:

A) utilize existing algorithm to obtain channel estimating value sequence Hp on the pilot sub-carrier ₁, Hp ₁Comprise the channel estimation value of non-negative pilot sub-carrier and the channel estimation value of negative pilot sub-carrier;

B) at channel estimating value sequence Hp ₁In the channel estimation value of non-negative pilot sub-carrier and the channel estimation value of negative pilot sub-carrier between insert NumPilotVSC 0, obtain sequence Hp ₂

C) to sequence Hp ₂Carrying out length is the IFFT conversion of NFFT/PilotInterval, Hp ₂Be transformed into time domain, obtain sequence C IR1;

D) specific part of the sequence C IR1 that obtains in the step c) is carried out left half-window and handle, obtain sequence C IR2;

E) specific location in sequence C IR2 inserts 0, obtains sequence C IR3, inserts 0 number and makes that the number of element is NFFT among the sequence C IR3;

F) sequence C IR3 is carried out the FFT conversion that length is NFFT, CIR3 is transformed into frequency domain, obtain sequence C FR1;

G) sequence C FR1 is carried out ring shift right, obtain all subcarrier channel estimation CFR2 in the ofdm system.

In the described step a), existing algorithm is LS algorithm, LMMSE algorithm or MMSE algorithm, channel estimating value sequence Hp ₁The number of middle element is NFFT/PilotInterval-NumPilotVSC.

In the described step b), sequence Hp ₂The number of middle element is NFFT/PilotInterval.

In the described step g), be that sequence C FR1 is carried out ring shift right InitPilot (the sub-carrier positions sequence number of first non-negative pilot sub-carrier) position.

In the described step d), the specific part of sequence C IR1 obtains by following steps:

D1) if the maximum delay of multipath channel (MAXDELAY) the unknown, then the length setting of windowing is the length (CPLENGTH) of Cyclic Prefix in the NFFT/PilotInterval-OFDM system, just CPLENGTH+1 element among the sequence C IR1 is begun to carry out windowing to a last element;

D2) if the maximum delay of multipath channel (MAXDELAY) is known; then the length setting of windowing is NFFT/PilotInterval-min (MAXDELAY+GUARDLENGTH; CPLENGTH); just to the min (MAXDELAY+GUARDLENGTH among the sequence C IR1; CPLENGTH)+1 an element begins to carry out windowing to a last element; function m in (MAXDELAY+GUARDLENGTH wherein; CPLENGTH) that less among both number is got in expression; parameter GUARDLENGTH represents the length on the protection border relevant with the maximum delay of channel, and this parameter can be by emulation or rule of thumb obtained.

In the described step d), described left half-window can adopt the left half-window of Gaussian window, Hamming window, Hanning window or rectangular window.

In the described step e), the ad-hoc location of sequence C IR2 obtains by following steps:

E1) if MAXDELAY the unknown then inserts 0 after CPLENGTH element in sequence C IR2;

E2) if MAXDELAY is known, then in sequence C IR2 min (MAXDELAY+GUARDLENGTH CPLENGTH) inserts 0 after the individual element.

The method of channel estimating in a kind of ofdm system, the length of FFT/IFFT (NFFT) equals the number (NumNonnegaPilot) of non-negative pilot sub-carrier, the number (NumPilotVSC) of pilot frequency locations in the virtual subnet carrier wave and number (NumNegaPilot) sum of negative pilot sub-carrier divided by pilot interval (PilotInterval) in the described ofdm system, and described method comprises:

A) utilize existing algorithm to obtain channel estimating value sequence Hp on the pilot sub-carrier ₁, Hp ₁Comprise the channel estimation value of negative pilot sub-carrier and the channel estimation value of non-negative pilot sub-carrier;

B) at channel estimating value sequence Hp ₁In the channel estimation value of non-negative pilot sub-carrier and the channel estimation value of negative pilot sub-carrier between insert NumPilotVSC 0, obtain sequence Hp ₂

C) to sequence Hp ₂Carrying out length is the FFT conversion of NFFT/PilotInterval, Hp ₂Be transformed into transform domain, obtain sequence C IR1;

D) specific part of the sequence C IR1 that obtains in the step c) is carried out right half-window and handle, obtain sequence C IR2;

E) specific location in CIR2 inserts 0, obtains sequence C IR3, inserts 0 number and makes that the number of element is NFFT among the sequence C IR3;

F) sequence C IR3 is carried out the IFFT conversion that length is NFFT, CIR3 is transformed into frequency domain, and the result of IFFT conversion be multiply by pilot interval (PilotInterval) in the frequency domain, obtain sequence C FR1;

G) sequence C FR1 is carried out ring shift right, obtain all subcarrier channel estimation CFR2 in the ofdm system.

In the described step a), existing algorithm is LS algorithm, LMMSE algorithm or MMSE algorithm, channel estimating value sequence Hp ₁The number of middle element is NFFT/PilotInterval-NumPilotVSC.

In the described step b), sequence Hp ₂The number of middle element is NFFT/PilotInterval.

In the described step g), be that sequence C FR1 is carried out ring shift right InitPilot (the sub-carrier positions sequence number of first non-negative pilot sub-carrier) position.

In the described step d), the specific part of sequence C IR1 obtains by following steps:

D1) if the maximum delay of multipath channel (MAXDELAY) the unknown, then the length setting of windowing is the length (CPLENGTH) of Cyclic Prefix in the NFFT/PilotInterval-OFDM system, just second element among the sequence C IR1 is begun partly to carry out windowing to NFFT/PilotInterval-CPLENGTH+1 element;

D2) if the maximum delay of multipath channel (MAXDELAY) is known; then the length setting of windowing is NFFT/PilotInterval-min (MAXDELAY+GUARDLENGTH; CPLENGTH); just second element among the sequence C IR1 begun the (MAXDELAY+GUARDLENGTH to NFFT/PilotInterval-min; CPLENGTH)+1 an element partly carries out windowing; function m in (MAXDELAY+GUARDLENGTH wherein; CPLENGTH) that less among both number is got in expression; parameter GUARDLENGTH represents the length on the protection border relevant with the maximum delay of channel, and this parameter can be by emulation or rule of thumb obtained.

In the described step d), described right half-window can adopt the right half-window of Gaussian window, Hamming window, Hanning window or rectangular window.

In the described step e), the ad-hoc location of sequence C IR2 obtains by following steps:

E1) if MAXDELAY the unknown then inserts 0 after NFFT/PilotInterval-CPLENGTH+1 element from sequence C IR2;

E2) if MAXDELAY is known, then from sequence C IR2 NFFT/PilotInterval-min (MAXDELAY+GUARDLENGTH CPLENGTH)+1 inserts 0 after the element.

The present invention has improved the method for channel estimating in traditional ofdm system, when improving performance, does not increase its complexity.

Description of drawings

Fig. 1 (a) is the block pilot frequency mode schematic diagram in the auxiliary ofdm system of pilot tone;

Fig. 1 (b) is the Comb Pilot pattern diagram in the auxiliary ofdm system of pilot tone;

Fig. 2 is the left half-window schematic diagram of the Hanning window that uses in the embodiment of the invention;

Fig. 3 is the schematic diagram of the operating process of the embodiment of the invention;

Fig. 4 adopts the channel estimation value of conventional method and the comparison schematic diagram of the least mean-square error of the channel estimation value that adopts the embodiment of the invention;

Fig. 5 adopts the channel estimation value of conventional method and the comparison schematic diagram of the bit error rate of the channel estimation value that adopts the embodiment of the invention;

Fig. 6 is the right half-window schematic diagram of the Hanning window of another embodiment of the present invention use;

Fig. 7 is the schematic diagram of the operating process of another embodiment of the present invention;

Fig. 8 adopts the channel estimation value of conventional method and the comparison schematic diagram of the least mean-square error of the channel estimation value that adopts another embodiment of the present invention;

Fig. 9 adopts the channel estimation value of conventional method and the comparison schematic diagram of the bit error rate of the channel estimation value that adopts another embodiment of the present invention;

Figure 10 adopts the channel estimation value of conventional method and the comparison schematic diagram of the least mean-square error of the channel estimation value that adopts another embodiment of the present invention;

Figure 11 adopts the channel estimation value of conventional method and the comparison schematic diagram of the bit error rate of the channel estimation value that adopts another embodiment of the present invention.

Embodiment

Below in conjunction with drawings and Examples, the present invention is done concrete introduction:

The invention provides the method for channel estimating in a kind of ofdm system, this method is the improvement of traditional time domain channel estimator, the length of FFT/IFFT (NFFT) equals the number (NumNonnegaPilot) of non-negative pilot sub-carrier, the number (NumPilotVSC) of pilot frequency locations in the virtual subnet carrier wave and number (NumNegaPilot) sum of negative pilot sub-carrier divided by pilot interval (PilotInterval) in this ofdm system, the schematic diagram of the operating process of this embodiment describes the processing procedure of each step among this embodiment in detail as shown in Figure 3 below in conjunction with Fig. 3:

In the step 310, utilize existing algorithm to obtain channel estimating value sequence Hp on the pilot sub-carrier ₁, Hp ₁Comprise the channel estimation value of non-negative pilot sub-carrier and the channel estimation value of negative pilot sub-carrier.Here, can adopt existing channel estimation method such as LS algorithm, LMMSE algorithm or MMSE algorithm, obtain the channel estimating value sequence on the pilot sub-carrier

${\mathrm{<figure-callout\; id="1"\; label="H\; P"\; filenames="A20031010816600211.png"\; state="\{\{state\}\}">H</figure-callout>}}_P=[H_{m_0}{\mathrm{<figure-callout\; id="1"\; label="H\; H"\; filenames="A20031010816600211.png"\; state="\{\{state\}\}">H</figure-callout>}}_{H_{}}...H_{m_{\mathrm{NumNonnegaPilot}-1}}{H}_{n_{\mathrm{NumNegnPilot}-1}}...H_{n_1}{H}_{n_0}],$ M wherein _i=InitPilot+iPilotInterval and n _i=InitPilot-(i+1) PilotInterval represents the subscript of non-negative pilot sub-carrier and negative pilot sub-carrier respectively, i can get 0,1,2,3...... InitPilot represents the subcarrier sequence number of first non-negative pilot sub-carrier, for example can be 2, PilotInterval represents the pilot interval in the frequency domain, for example can be 8.Channel estimating value sequence Hp ₁The number of middle element is pilot interval (the PilotInterval)-NumPilotVSC in the NFFT/ frequency domain.

In the step 320, at channel estimating value sequence Hp ₁In the channel estimation value of non-negative pilot sub-carrier and the channel estimation value of negative pilot sub-carrier between insert NumPilotVSC 0, obtain sequence Hp ₂Here, between the channel estimation value of the channel estimation value of non-negative pilot sub-carrier and negative pilot sub-carrier, insert NumPilotVSC 0, with Hp ₁Expand to the length of NFFT/PilotInterval, obtain sequence Hp ₂, i.e. sequence Hp ₂The number of middle element is NFFT/PilotInterval, and the result is expressed as $H_{P2}=[H_{m_0}{\mathrm{<figure-callout\; id="1"\; label="H\; m"\; filenames="A20031010816600211.png"\; state="\{\{state\}\}">H</figure-callout>}}_{m_{}}...H_{m_{\mathrm{NumNonnegaPilot}-1}}000...000H_{n_{\mathrm{NumNegaPilot}-1}}...H_{n_1}{H}_{n_0}].$

In the step 330, to sequence Hp ₂Carrying out length is the IFFT conversion of NFFT/PilotInterval, Hp ₂Be transformed into time domain and obtain sequence C IR1, CIR1 can be expressed as CIR ₁=[C ₀C ₁C ₂... C _{NFFT/PilotInterval-2}C _{NFFT/PilotInterval-1}].

In the step 340, judge whether the maximum delay (MAXDELAY) of multipath channel is known:

If the maximum delay of multipath channel (MAXDELAY) the unknown, then execution in step 342, CPLENGTH+1 element among the sequence C IR1 begun to carry out windowing to a last element, the length setting of windowing just is the length (CPLENGTH) of Cyclic Prefix in the NFFT/PilotInterval-OFDM system, and windowing partly is $[C_{\mathrm{CPLength}}...C_{\frac{\mathrm{NFFT}}{\mathrm{PilotInterval}}-2}{C}_{\frac{\mathrm{NFFT}}{\mathrm{PilotInterval}}-1}],$ Obtain sequence C IR2.

Execution in step 352 then, insert 0 after CPLENGTH element in sequence C IR2, just insert 0 between CPLENGTH element and CPLENGTH+1 element, obtain sequence C IR3.

If (MAXDELAY) is known for the maximum delay of multipath channel, then execution in step 341, to the min (MAXDELAY+GUARDLENGTH among the sequence C IR1, CPLENGTH)+1 an element begins to carry out windowing to a last element, the length setting of windowing just is NFFT/PilotInterval-min (MAXDELAY+GUARDLENGTH, CPLENGTH), windowing partly is $[C_{\min (\mathrm{MaxDelay}+\mathrm{GuardLength},\mathrm{CPLength})}...C_{\frac{\mathrm{NFFT}}{\mathrm{PilotInterval}}-2}{C}_{\frac{\mathrm{NFFT}}{\mathrm{PilotInterval}}-1}],$ Obtain sequence C IR2; function m in (MAXDELAY+GUARDLENGTH wherein; CPLENGTH) that less among both number is got in expression; parameter GUARDLENGTH represents the length on the protection border relevant with the maximum delay of channel; this parameter can be by emulation or is rule of thumb obtained, and for example can be set as 0.1*MAXDELAY.

Execution in step 351 then, min (MAXDELAY+GUARDLENGTH in sequence C IR2, CPLENGTH) insert 0 after the individual element, just at min (MAXDELAY+GUARDLENGTH, CPLENGTH) individual element and min (MAXDELAY+GUARDLENGTH, CPLENGTH)+1 insert 0 between the element, obtain sequence C IR3.

Among the figure, X represent min (MAXDELAY+GUARDLENGTH, CPLENGTH).

Here, the performance for estimating channel that left half-window can obtain according to emulation is selected the left half-window of Gaussian window, Hamming window, Hanning window or rectangular window for use, but is not limited to this a few class windows.For example can adopt the left half-window of Hanning window shown in Figure 2, the abscissa among Fig. 2 is represented sampled point, and ordinate is represented amplitude.Can suppress the noise effect of time-domain signal by step 341 or 342.

After the ad-hoc location of sequence C IR2 inserts 0, the extended length of sequence C IR2 to NFFT, is obtained sequence C IR3, the number of element is NFFT among this sequence C IR3, can be expressed as CIR3=[W ₀W ₁W ₂... ..W _{InsertingPosition-1}000.....000W _{InsertingPosition}... ..W _{NFFT/PilotInterval-1}], wherein InsertPosition represents the subscript of insertion position.Can eliminate the influence of virtual subnet carrier wave in channel estimating by step 351 or 352.

In the step 360, sequence C IR3 is carried out the FFT conversion that length is NFFT, CIR3 is transformed into frequency domain, obtain sequence C FR1, sequence C FR1 can be expressed as CFR ₁=[F ₀F ₁F ₂F _NFFT-2F _NFFT-1].

In the step 370, sequence C FR1 is carried out ring shift right, obtain all subcarrier channel estimation CFR2 in the ofdm system.Here, be that sequence C FR1 is carried out ring shift right InitPilot (the sub-carrier positions sequence number of first non-negative pilot sub-carrier) position, the result of channel estimating is mapped on the correct subcarrier, obtain sequence C FR2, be expressed as CFR ₂=[F _{NFFT-InitPilot}F _{NFFT-InitPilot+1}F _NFFT-1F ₀F ₁F _{NFFT-InitPilot-2}F _{NFFT-InitPilo-1}], CFR2 is the end product of the channel estimating of all subcarriers in the ofdm system.

Below by emulation example explanation effect of the present invention.

In this example, simulation parameter is as follows:

NFFT=2048, PilotInterval=8, the number 127 of virtual subnet carrier wave, InitPilot=2, CPLENTH=210, sample frequency is 20.592MHz, carrier frequency is 3.2GHz, and multipath channel models is the COST207 channel model, and translational speed is 60kmph, modulator approach 16QAM, no chnnel coding.Wherein the tapped delay line of COST207 channel model (TDL) parameter sees Table 1.

???Tap ???No.	Relative delay (nsec)	Average power (dB)	Doppler frequency spectrum
				????1	????0	?????0.0	The typical case
????2	????2000	????-6.0	The typical case
				????3	????4000	????-12.0	The typical case
????4	????6000	????-18.0	The typical case
				????5	????8000	????-24.0	The typical case
????6	????10000	????-30.0	The typical case

Table 1

In emulation, suppose MAXDELAY the unknown of multipath channel.Fig. 4 adopts the channel estimation value of conventional method and the comparison schematic diagram of the least mean-square error of the channel estimation value that adopts channel estimation methods of the present invention to above-mentioned emulation example, on behalf of above-mentioned emulation example, the curve that wherein has " △ " symbol adopt the least mean-square error of the channel estimation value of channel estimation methods of the present invention, and on behalf of above-mentioned emulation example, the curve that has " zero " symbol adopt the least mean-square error of the channel estimation value of conventional method.Fig. 5 adopts the channel estimation value of conventional method and the comparison schematic diagram of the bit error rate of the channel estimation value that adopts channel estimation methods of the present invention to above-mentioned emulation example, on behalf of above-mentioned emulation example, the curve that wherein has " △ " symbol adopt the bit error rate of the channel estimation value of channel estimation methods of the present invention, and on behalf of above-mentioned emulation example, the curve that has " zero " symbol adopt the bit error rate of the channel estimation value of conventional method.By Fig. 4 and Fig. 5 as can be seen, the least mean-square error of channel estimation value of the present invention and bit error rate compared with prior art reduce greatly, and performance for estimating channel improves greatly.

The present invention also provides the method for channel estimating in the another kind of ofdm system, this method is the improvement of traditional transform domain channel estimation method, the length of FFT/IFFT (NFFT) equals the number (NumNonnegaPilot) of non-negative pilot sub-carrier in the described ofdm system divided by pilot interval (PilotInterval), number (NumNegaPilot) sum of number of the pilot frequency locations in the virtual subnet carrier wave (NumPilotVSC) and negative pilot sub-carrier, the schematic diagram of the operating process of this embodiment describes the processing procedure of each step among this embodiment in detail as shown in Figure 7 below in conjunction with Fig. 7:

In the step 710, utilize existing algorithm to obtain channel estimating value sequence Hp on the pilot sub-carrier ₁, Hp ₁Comprise the channel estimation value of negative pilot sub-carrier and the channel estimation value of non-negative pilot sub-carrier.Here, existing algorithm is LS algorithm, LMMSE algorithm or MMSE algorithm, obtains the channel estimating value sequence on the pilot sub-carrier $H_P$ $1_{}=[H_{m_0}{\mathrm{<figure-callout\; id="1"\; label="H\; m"\; filenames="A20031010816600211.png"\; state="\{\{state\}\}">H</figure-callout>}}_{m_{}}...H_{m_{\mathrm{NumNonnegaPilot}-1}}{H}_{n_{\mathrm{NumNegaPilot}-1}}...H_{n_1}{H}_{n_0}],$ M wherein _i=InitPilot+iPilotInterval and n _i=InitPilot-(i+1) PilotInterval represents the subscript of non-negative pilot sub-carrier and negative pilot sub-carrier respectively, i can get 0,1,2...., InitPilot represents the subcarrier sequence number of first non-negative pilot sub-carrier, for example can be 2, PilotInterval represents the pilot interval in the frequency domain, for example can be 8.Channel estimating value sequence Hp ₁The number of middle element is pilot interval (the PilotInterval)-NumPilotVSC in the NFFT/ frequency domain.

In the step 720, at channel estimating value sequence Hp ₁In the channel estimation value of non-negative pilot sub-carrier and the channel estimation value of negative pilot sub-carrier between insert NumPilotVSC 0, obtain sequence Hp ₂Here, between the channel estimation value of the channel estimation value of non-negative pilot sub-carrier and negative pilot sub-carrier, insert NumPilotVSC 0, with Hp ₁Expand to the length of NFFT/PilotInterval, obtain sequence Hp ₂, i.e. sequence Hp ₂The number of middle element is NFFT/PilotInterval, and the result can be expressed as $H_{P2}=[H_{m_0}{\mathrm{<figure-callout\; id="1"\; label="H\; m"\; filenames="A20031010816600211.png"\; state="\{\{state\}\}">H</figure-callout>}}_{m_{}}...H_{m_{\mathrm{NumNonnegaPilot}-1}}000...000H_{n_{\mathrm{NumNegaPilot}-1}}...H_{n_1}{H}_{n_0}].$

In the step 730, to sequence Hp ₂Carrying out length is the FFT conversion of NFFT/PilotInterval, Hp ₂Be transformed into transform domain, obtain sequence C IR1, CIR1 can be expressed as CIR1=[C ₀C ₁C ₂... C _{NFFT/PilotInterval-2}C _{NFFT/PilotInterval-1}].

In the step 740, judge whether the maximum delay (MAXDELAY) of multipath channel is known:

If the maximum delay of multipath channel (MAXDELAY) the unknown, then execution in step 742, second element among the sequence C IR1 begun partly to carry out windowing to NFFT/PilotInterval-CPLENGTH+1 element, the length of windowing just is the length (CPLENGTH) of Cyclic Prefix in the NFFT/PilotInterval-OFDM system, to the part of windowing among the CIR1 is $[{\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="A20031010816600211.png"\; state="\{\{state\}\}">C</figure-callout>}}_1...C_{\frac{\mathrm{NFFT}}{\mathrm{PilotInterval}}\mathrm{CPLength}-1}{C}_{\frac{\mathrm{NFFT}}{\mathrm{PilotInterval}}-\mathrm{CPLength}}].$

Execution in step 752 then, from sequence C IR2, insert 0 after NFFT/PilotInterval-CPLENGTH+1 element, just between NFFT/PilotInterval-CPLENGTH+1 element and NFFT/PilotInterval-CPLENGTH+2 element, insert 0, obtain sequence C IR3.

If (MAXDELAY) is known for the maximum delay of multipath channel, then execution in step 741, second element among the sequence C IR1 begun the (MAXDELAY+GUARDLENGTH to NFFT/PILOTINTERVAL-min, CPLENGTH)+1 an element partly carries out windowing, the length of windowing just is NFFT/PilotInterval-min (MAXDELAY+GUARDLENGTH, CPLENGTH), second element among the sequence C IR1 begun the (MAXDELAY+GUARDLENGTH to NFFT/PILOTINTERVAL-min, CPLENGTH)+1 an element partly carries out windowing, to the part of windowing among the CIR1 is $[{\mathrm{<figure-callout\; id="1"\; label="C"\; filenames="A20031010816600211.png"\; state="\{\{state\}\}">C</figure-callout>}}_1...C_{\frac{\mathrm{NFFT}}{\mathrm{PilotInterval}}-\min (\mathrm{MaxDelay}+\mathrm{GuardLength},\mathrm{CPLength}-1)}{C}_{\frac{\mathrm{NFFT}}{\mathrm{PilotInterval}}-\min (\mathrm{MaxDelay}+\mathrm{GuardLength},\mathrm{CPLength})}],$ Function m in (MAXDELAY+GUARDLENGTH wherein; CPLENGTH) that less among both number is got in expression; parameter GUARDLENGTH represents the length on the protection border relevant with the maximum delay of channel; this parameter can be by emulation or is rule of thumb obtained, and for example can be set as 0.1*MAXDELAY.Notice that first element of the part of windowing here is C ₁, rather than C ₀

Execution in step 751 then, NFFT/PilotInterval-min (MAXDELAY+GUARDLENGTH from sequence C IR2, CPLENGTH)+1 insert 0 after the element, just at NFFT/PilotInterval-min (MAXDELAY+GUARDLENGTH, CPLENGTH)+1 an element and NFFT/PilotInterval-min (MAXDELAY+GUARDLENGTH, CPLENGTH)+2 insert 0 between the element, obtain sequence C IR3.

Among the figure, X represents NFFT/PilotInterval-CPLENGTH+1, Y represent NFFT/PILOTINTERVAL-min (MAXDELAY+GUARDLENGTH, CPLENGTH)+1.

Here, the performance for estimating channel that right half-window can obtain according to emulation is selected the right half-window of Gaussian window, Hamming window, Hanning window or rectangular window for use, but is not limited to this a few class windows.For example can adopt the right half-window of Hanning window shown in Figure 6, the abscissa among Fig. 6 is represented sampled point, and ordinate is represented amplitude.Can suppress the noise effect of transform-domain signals by step 741 or 742.

Here, insert 0 at the ad-hoc location of sequence C IR2, the extended length of sequence C IR2 to NFFT, is obtained sequence C IR3, the number of element is NFFT among this sequence C IR3, can be expressed as CIR3=[W ₀W ₁W ₂... ..W _{InsertingPosition-1}000.....000W _{InsertingPosition}... ..W _{NFFT/PilotInterval-1}], wherein InsertPosition represents the subscript of insertion position.

Can eliminate maximum delay and the CP length situation that performance reduces in the big multipath channel of comparing by step 751 or 752.

In the step 760, sequence C IR3 is carried out the IFFT conversion that length is NFFT, CIR3 is transformed into frequency domain, and the result of IFFT conversion be multiply by pilot interval (PilotInterval) in the frequency domain, obtain sequence C FR1, be expressed as CFR ₁=[F ₀F ₁F ₂F _NFFT-2F _NFFT-1].

In the step 770, sequence C FR1 is carried out ring shift right, obtain all subcarrier channel estimation CFR2 in the ofdm system.Here, be that sequence C FR1 is carried out ring shift right InitPilot (first non-negative pilot sub-carrier position number) position, the result of channel estimating is mapped to correctly to subcarrier, obtain sequence C FR2, can be expressed as CFR ₂=[F _{NFFT-InitPilot}F _{NFFT-InitPilot+1}F _NFFT-1F ₀F ₁F _{NFFT-InitPilot-2}F _{NFFT-InitPilot-1}], CFR2 is exactly the road estimated value of all subcarriers in the ofdm system.

Below by emulation example explanation effect of the present invention.

In this embodiment, simulation parameter is as follows:

NFFT=2048, PilotInterval=8, the number 127 of virtual subnet carrier wave, InitPilot=2, CPLENTH=210, sample frequency is 20.592MHz, carrier frequency is 3.2GHz, and multipath channel models is the COST207 channel model, and translational speed is 60kmph, modulator approach 16QAM, no chnnel coding.Wherein the tapped delay line of COST207 channel model (TDL) parameter sees Table 1.

In emulation, suppose MAXDELAY the unknown of multipath channel.Fig. 8 adopts the channel estimation value of conventional method and the comparison schematic diagram of the least mean-square error of the channel estimation value that adopts channel estimation methods of the present invention to above-mentioned emulation example, on behalf of above-mentioned emulation example, the curve that wherein has " △ " symbol adopt the least mean-square error of the channel estimation value of channel estimation methods of the present invention, and on behalf of above-mentioned emulation example, the curve that has " zero " symbol adopt the least mean-square error of the channel estimation value of conventional method.Fig. 9 adopts the channel estimation value of conventional method and the comparison schematic diagram of the bit error rate of the channel estimation value that adopts channel estimation methods of the present invention to above-mentioned emulation example, on behalf of above-mentioned emulation example, the curve that wherein has " △ " symbol adopt the bit error rate of the channel estimation value of channel estimation methods of the present invention, and on behalf of above-mentioned emulation example, the curve that has " zero " symbol adopt the bit error rate of the channel estimation value of conventional method.By Fig. 8 and Fig. 9 as can be seen, to the COST207 channel, the least mean-square error of channel estimation value of the present invention and bit error rate compared with prior art reduce greatly, and performance for estimating channel improves greatly.

For the COST207 channel that uses, maximum delay is 10 μ s, and CP length is 10.198 μ s, so maximum delay is relatively near the length of CP.We may safely draw the conclusion, and channel estimating of the present invention is compared with traditional channel estimating, relatively under the multipath channel models condition near CP length, can improve the performance of system at maximum delay greatly.

In this example, the simulation parameter of another emulation example is as follows:

NFFT=2048, PilotInterval=8, the number 127 of virtual subnet carrier wave, InitPilot=2, CPLENTH=210, sample frequency is 20.592MHz, carrier frequency is 3.2GHz, and multipath channel models is the channel model of the vehicle-mounted A of UMTS, and translational speed is 60kmph, modulator approach 16QAM, no chnnel coding.Wherein the TDL parameter of the channel model of the vehicle-mounted A of UMTS sees Table 2.

The tap sequence number	Relative delay (nsec)	Average power (dB)	Doppler frequency spectrum
				??1	????0	?????0.0	The typical case
??2	????310	????-1.0	The typical case
				??3	????710	????-9.0	The typical case
??4	????1090	????-10.0	The typical case
				??5	????1730	????-15.0	The typical case
??6	????2510	????-20.0	The typical case

Table 2

The MAXDELAY that supposes multipath is known, and the GUARDLENGTH parameter value is set at 0.2*MAXDELAY.Referring to Figure 10 and Figure 11, Figure 10 adopts the channel estimation value of conventional method and the comparison schematic diagram of the least mean-square error of the channel estimation value that adopts channel estimation methods of the present invention to above-mentioned emulation example, on behalf of above-mentioned emulation example, the curve that wherein has " △ " symbol adopt the least mean-square error of the channel estimation value of channel estimation methods of the present invention, and on behalf of above-mentioned emulation example, the curve that has " zero " symbol adopt the least mean-square error of the channel estimation value of conventional method.Figure 11 adopts the channel estimation value of conventional method and the comparison schematic diagram of the bit error rate of the channel estimation value that adopts channel estimation methods of the present invention to above-mentioned emulation example, on behalf of above-mentioned emulation example, the curve that wherein has " △ " symbol adopt the bit error rate of the channel estimation value of channel estimation methods of the present invention, and on behalf of above-mentioned emulation example, the curve that has " zero " symbol adopt the bit error rate of the channel estimation value of conventional method.From Figure 10 and 11, find out, compare less relatively multipath channel for those maximum delays with CP length, the least mean-square error of channel estimation value of the present invention and bit error rate compared with prior art obviously reduce, and performance for estimating channel also obviously increases.

In a word, compared with prior art, the improvement that the present invention is bigger the channel estimating performance of ofdm system, and can not increase complexity.Therefore, has very high practical value.

The above embodiment is two embodiment of the present invention, and is not limited to this, and in the situation that does not exceed spiritual scope of the present invention, the many variations of being done is implemented, and all belongs to scope of the present invention.

Claims (10)

1. the method for channel estimating in the ofdm system, the length of FFT/IFFT (NFFT) equals the number (NumNonnegaPilot) of non-negative pilot sub-carrier, the number (NumPilotVSC) of pilot frequency locations in the virtual subnet carrier wave and number (NumNegaPilot) sum of negative pilot sub-carrier divided by pilot interval (PilotInterval) in the described ofdm system, it is characterized in that described method comprises:

A) utilize existing algorithm to obtain channel estimating value sequence Hp on the pilot sub-carrier ₁, Hp ₁Comprise the channel estimation value of non-negative pilot sub-carrier and the channel estimation value of negative pilot sub-carrier;

B) at channel estimating value sequence Hp ₁In the channel estimation value of non-negative pilot sub-carrier and the channel estimation value of negative pilot sub-carrier between insert NumPilotVSC 0, obtain sequence Hp ₂

C) to sequence Hp ₂Carrying out length is the IFFT conversion of NFFT/PilotInterval, Hp ₂Be transformed into time domain, obtain sequence C IR1;

D) specific part of the sequence C IR1 that obtains in the step c) is carried out left half-window and handle, obtain sequence C IR2;

E) specific location in sequence C IR2 inserts 0, obtains sequence C IR3, inserts 0 number and makes that the number of element is NFFT among the sequence C IR3;

F) sequence C IR3 is carried out the FFT conversion that length is NFFT, CIR3 is transformed into frequency domain, obtain sequence C FR1;

G) sequence C FR1 is carried out ring shift right, obtain all subcarrier channel estimation CFR2 in the ofdm system.

2. the method for claim 1 is characterized in that, in the described step a), existing algorithm is LS algorithm, LMMSE algorithm or MMSE algorithm, and the number of element is NFFT/PilotInterval-NumPilotVSC among the channel estimating value sequence Hp1;

In the described step b), sequence Hp ₂The number of middle element is NFFT/PilotInterval;

In the described step f), be that sequence C FR1 is carried out ring shift right InitPilot (the sub-carrier positions sequence number of first non-negative pilot sub-carrier) position.

3. the method for claim 1 is characterized in that, in the described step d), the specific part of sequence C IR1 obtains by following steps:

D1) if the maximum delay of multipath channel (MAXDELAY) the unknown, then the length setting of windowing is the length (CPLENGTH) of Cyclic Prefix in the NFFT/PilotInterval-OFDM system, just CPLENGTH+1 element among the sequence C IR1 is begun to carry out windowing to a last element;

D2) if the maximum delay of multipath channel (MAXDELAY) is known; then the length setting of windowing is NFFT/PilotInterval-min (MAXDELAY+GUARDLENGTH; CPLENGTH); just to the min (MAXDELAY+GUARDLENGTH among the sequence C IR1; CPLENGTH)+1 an element begins to carry out windowing to a last element; function m in (MAXDELAY+GUARDLENGTH wherein; CPLENGTH) that less among both number is got in expression; parameter GUARDLENGTH represents the length on the protection border relevant with the maximum delay of channel, and this parameter can be by emulation or rule of thumb obtained.

4. as claim 1 or 3 described methods, it is characterized in that in the described step d), described left half-window can adopt the left half-window of Gaussian window, Hamming window, Hanning window or rectangular window.

5. the method for claim 1 is characterized in that, in the described step e), the ad-hoc location of sequence C IR2 obtains by following steps:

E1) if MAXDELAY the unknown then inserts 0 after CPLENGTH element in sequence C IR2;

E2) if MAXDELAY is known, then in sequence C IR2 min (MAXDELAY+GUARDLENGTH CPLENGTH) inserts 0 after the individual element.

6. the method for channel estimating in the ofdm system, the length of FFT/IFFT (NFFT) equals the number (NumNonnegaPilot) of non-negative pilot sub-carrier, the number (NumPilotVSC) of pilot frequency locations in the virtual subnet carrier wave and number (NumNegaPilot) sum of negative pilot sub-carrier divided by pilot interval (PilotInterval) in the described ofdm system, it is characterized in that described method comprises:

A) utilize existing algorithm to obtain channel estimating value sequence Hp on the pilot sub-carrier ₁, Hp ₁Comprise the channel estimation value of negative pilot sub-carrier and the channel estimation value of non-negative pilot sub-carrier;

B) at channel estimating value sequence Hp ₁In the channel estimation value of non-negative pilot sub-carrier and the channel estimation value of negative pilot sub-carrier between insert NumPilotVSC 0, obtain sequence Hp ₂

C) to sequence HP ₂Carrying out length is the FFT conversion of NFFT/PilotInterval, HP ₂Be transformed into transform domain, obtain sequence C IR1;

D) specific part of the sequence C IR1 that obtains in the step c) is carried out right half-window and handle, obtain sequence C IR2;

E) specific location in CIR2 inserts 0, obtains sequence C IR3, inserts 0 number and makes that the number of element is NFFT among the sequence C IR3;

F) sequence C IR3 is carried out the IFFT conversion that length is NFFT, CIR3 is transformed into frequency domain, and the result of IFFT conversion be multiply by pilot interval (PilotInterval) in the frequency domain, obtain sequence C FR1;

G) sequence C FR1 is carried out ring shift right, obtain all subcarrier channel estimation CFR2 in the ofdm system.

7. method as claimed in claim 6 is characterized in that, in the described step a), existing algorithm is LS algorithm, LMMSE algorithm or MMSE algorithm, channel estimating value sequence Hp ₁The number of middle element is NFFT/PilotInterval-NumPilotVSC;

In the described step b), sequence Hp ₂The number of middle element is NFFT/PilotInterval;

In the described step g), be that sequence C FR1 is carried out ring shift right InitPilot (the sub-carrier positions sequence number of first non-negative pilot sub-carrier) position.

8. method as claimed in claim 6 is characterized in that, in the described step d), the specific part of sequence C IR1 obtains by following steps:

D1) if the maximum delay of multipath channel (MAXDELAY) the unknown, then the length setting of windowing is the length (CPLENGTH) of Cyclic Prefix in the NFFT/PilotInterval-OFDM system, just second element among the sequence C IR1 is begun partly to carry out windowing to NFFT/PilotInterval-CPLENGTH+1 element;

D2) if the maximum delay of multipath channel (MAXDELAY) is known; then the length setting of windowing is NFFT/PilotInterval-min (MAXDELAY+GUARDLENGTH; CPLENGTH); just second element among the sequence C IR1 begun the (MAXDELAY+GUARDLENGTH to NFFT/PilotInterval-min; CPLENGTH)+1 an element partly carries out windowing; function m in (MAXDELAY+GUARDLENGTH wherein; CPLENGTH) that less among both number is got in expression; parameter GUARDLENGTH represents the length on the protection border relevant with the maximum delay of channel, and this parameter can be by emulation or rule of thumb obtained.

9. as claim 6 or 8 described methods, it is characterized in that in the described step d), described right half-window can adopt the right half-window of Gaussian window, Hamming window, Hanning window or rectangular window.

10. method as claimed in claim 6 is characterized in that, in the described step e), the ad-hoc location of sequence C IR2 obtains by following steps:

E1) if MAXDELAY the unknown then inserts 0 after NFFT/PilotInterval-CPLENGTH+1 element from sequence C IR2;

E2) if MAXDELAY is known, then from sequence C IR2 NFFT/PilotInterval-min (MAXDELAY+GUARDLENGTH CPLENGTH)+1 inserts 0 after the element.

Images