Background technology
From the second generation (2G) to the third generation (3G), forth generation (4G) cell mobile communication systems, the complexity of system progressively improves, no matter also progressively improving for base station or for the requirement of terminal physical layer handling capacity, especially with Long Term Evolution (LongTermEvolution, LTE) system is in the accurate 4G system of representative, can under the prerequisite of guaranteed performance, adopt the physical layer process algorithm optimized, the competitiveness for improving product is very helpful.
OFDM (OrthogonalFrequencyDivisionMultiplexing, OFDM) technology become mainstream technology in current long evolving system, and MMSE (MinimumMeanSquareError, least mean-square error) channel estimation method can make full use of the correlation information of OFDM time domain and frequency domain, obtains excellent performance.In a wireless communication system, be very important module for detecting the channel estimation module of useful signal, LTE system is no exception.
The difference of the channel estimation method of signal directly affects the complexity in the process of physical layer, quite large on the impact of algorithm design.MMSE channel estimation method, because can well utilize the relevant information of time-frequency domain, is a kind of channel estimation method effectively going for OFDM technology.A kind of way of realization of this algorithm carries out MMSE filtering respectively at time-domain and frequency-domain, obtains the channel estimation results in the least mean-square error meaning of time domain and frequency domain respectively.Ofdm system frequency domain relevant information needs to be obtained by power delay spectrum, and concrete Signal estimation algorithm is as follows:
The fundamental formular that channel estimation method uses is:
wherein β=E{|X
k|
2e{1/|X
k|
2, X
kfor constellation symbol,
for the result of LS (leastSquare, minimum variance) channel estimating,
for the correlation matrix of pilot tone and frequency pilot sign,
for the correlation matrix of data and frequency pilot sign.Wherein,
with
acquisition and the frequency domain power delay profile of channel have direct relation.
When obtaining correlation matrix, can first according to R
f=F
hpF obtains R
f, then according to R
fin element obtain
with
wherein P is channel delay power spectrum, about R
f=F
hthe proof of PF is as follows:
Being provided with symbol time is T
use, then Δ f=1/T is had
use, suppose that multidiameter delay is the integral multiple (not being the situation of sampling time integral multiple for time delay, because energy is mainly distributed in its integral multiple sampling time tap closed on, so still can use above-mentioned hypothesis) in sampling time, namely
substitute into above formula can obtain:
i
lrepresent the tap position in l footpath.
Further, if definition channel delay power spectral vectors p, meet
With seasonal k=u-v, then have
If definition F matrix, wherein
then have:
wherein,
represent F matrix u row, and
represent F
hv is capable for matrix,
And frequency domain correlation matrix can be expressed as:
Application
note k=u-v.Be not difficult to obtain: R
f=F
hpF, this conclusion is often directly applied in numerous document.
The present inventor finds, for ofdm system, when frequency domain carries out spectral pattern estimation, the existence of noise has totally unfavorable impact to channel impulse response length.Carry out for effective multi-path information over-evaluating or underestimate the correlation that all can affect time domain, the quantity of less estimation effective diameter can cause over-evaluating of correlation, causes the deterioration of performance; The information of more estimation effective diameter can cause underestimating of correlation, causes the reduction of gain.
In fact spectral pattern comprises several aspect: one is Channel Impulse length, and two is envelope shapes, and three is tap intervals.Four is relative power value.Therefore, spectral pattern estimates more complicated, is subject to more multifactorial impact, and when SNR (signal to noise ratio) changes, the part of inverting in the fundamental formular that channel estimation method uses needs to upgrade; When multidiameter delay power spectrum (PDP) changes, the correlation matrix in the fundamental formular that channel estimation method uses and the part of inverting of response need to upgrade, and will bring certain difficulty like this to realization.
In general, ideally, the spectral pattern of channel delay power spectrum is defined as:
Cannot obtain in desirable spectral pattern reality, generally for the realization being convenient to product, reduce complexity, the mode of presetting power delay spectrum can be adopted under the prerequisite that performance meets the demands to generate frequency domain relevant information, such as adopt square spectrum, exponential spectrum etc.
Below for square spectrum, the method step carrying out channel estimating based on different spectral pattern is described:
The first step, generates square power spectrum:
Second step, generates channel related information R according to P
f=F
hpF, wherein
3rd step, takes out the Received signal strength r of pilot frequency locations
p, according to the pilot frequency sequence r that receiving end is known
seq, obtain the channel-estimation information of pilot point:
4th step, according to the channel-estimation information of pilot point
pass through formula
obtain the channel estimate matrix on final each frequency
Square spectrum is uniformly distributed spectrum in other words, realize simple, but under imperfect environment, because this spectral pattern well can not mate actual conditions, therefore, the robustness of carrying out channel estimating is not good.
And the parameter of exponential spectrum determines to bring extra complexity, based on the channel estimation methods of exponential spectrum and the step of square spectrum similar, difference is only that the generation formula of P is different, no longer too much describe at this, because this spectral pattern well can not mate actual conditions, so the robustness of its channel estimating is not good yet.
Embodiment
The embodiment of the present invention provides a kind of channel estimation methods and device, when determining channel related information, according to power delay spectrum spectrum shape be: before the middle base band time sampling point preset, time-domain response is steady state value, and after described middle base band time sampling point, time-domain response decayed with the base band time; Channel estimating is carried out again by this channel related information.Because the power delay spectrum used has attenuation process, better can mate actual conditions, so use this power delay spectrum to carry out robustness that channel estimating improves channel estimating.
As shown in Figure 1, the channel estimation methods that the embodiment of the present invention provides comprises:
Step S101, according to power delay spectrum determination channel related information, power delay spectrum spectrum shape is specially, before the middle base band time sampling point preset, time-domain response is steady state value, and after middle base band time sampling point, time-domain response decayed with the base band time;
Step S102, carry out channel estimating according to channel related information.
Wherein, middle base band time sampling point sets according to channel synchronization error usually, when synchronous error is larger, this middle base band time sampling point also needs the larger of setting, for making channel estimating can obtain better robustness, usually need the constant portion of time-domain response can contain whole channel synchronization errors.
In the part that time-domain response decayed with the base band time, can be designed to according to the decay of different rules, such as, can be linear attenuation, also can exponentially decay, can also decay according to logarithm waveform according to actual conditions.
Concrete, as shown in Figure 2, the calculation procedure that the power delay spectrum provided according to the embodiment of the present invention carries out channel estimation method is as follows:
Step S201, generating power delay profile P
pdp;
Step S202, according to P
pdpgenerate channel related information R
f=F
hp
pdpf, wherein
Step S203, Received signal strength r according to pilot frequency locations
pand the pilot frequency sequence r that receiving terminal is known
seq, determine the channel-estimation information of pilot point
Step S204, channel-estimation information according to pilot point
determine the channel estimate matrix on each frequency
wherein, β=E{|X
k|
2e{1/|X
k|
2, X
kfor constellation symbol,
for the result of LS channel estimating,
for the correlation matrix of pilot tone and frequency pilot sign,
for the correlation matrix of data and frequency pilot sign,
with
according to R
fdetermine.
When using linear attenuation, the spectrum shape of power delay spectrum as shown in Figure 3, in step s 201, determines power delay spectrum P
pdpspecifically comprise the steps:
First base band sampled point StarPoint=α L in the middle of setting
cp, wherein L
cpfor CP (time baseband sampling point) length, α is Optimal Parameters; Setting end point time-domain response value size is β;
Then can determine to compose interval
Can determine that time-domain response value is: P (i)=1, i=1,2 ..., α L
cp; P (i)=1-(step) (i-1), i=α L
cp+ 1, α L
cp+ 2 ..., L
cp;
Calculating is normalized to P (i), P
av(i)=P (i)/mean (P), i=1,2 ..., L
cp+ 1, wherein, the mean value that mean (P) is P (i);
Determine P
pdp=P
av.
Wherein, α value and β value can be chosen as required, such as α=0.08, β=0.02.
When channel estimation parameter does not change, determine the channel related information that channel related information uses when being the estimation of last secondary channel; When channel estimation parameter changes, redefine channel related information according to power delay spectrum.That is, if α is fixed value, then the result of calculation R after first two steps
fcan prestore, thus avoid carrying out R when carrying out channel estimating at every turn
fdouble counting.If α needs to adjust according to measuring amount, then can according to adjustment mode cycle or triggering calculating R
f, such as, if periodically adjust α, then periodically can calculate R according to the cycle of adjustment α
fif not timing adjustment α, then can trigger R after adjustment α
fcalculating.
Certainly, the parameter related to can also be debugged further, thus optimize channel estimating performance, such as, can emulate for different α value in power delay spectrum, thus preferably optimum configurations under selection respective channel model.Fig. 3 is under the condition of LTE system 20M bandwidth, the emulation spectrum shape that CP length is 144, α when being 0.15.
When using exponential damping, the spectrum shape of power delay spectrum as shown in Figure 4, in step s 201, determines power delay spectrum P
pdpspecifically comprise the steps:
First base band sampled point StarPoint=α L in the middle of setting
cp, wherein L
cpfor CP length, α is Optimal Parameters; Setting end point time-domain response value size is β;
Can determine that time-domain response value is: P (i)=1, i=1,2 ..., α L
cp,
P (i)=exp (-β i), i=α L
cp+ 1, α L
cp+ 2 ..., L
cp, β=0.02 in Fig. 4.
Calculate P ' (i)=P (i)+χ, i=1,2 ..., L
cp+ 1, χ is Optimal Parameters, χ=0.2 in Fig. 4.
Calculating is normalized to P ' (i), P
av(i)=P (i)/mean (P), i=1,2 ..., L
cp+ 1, wherein, the mean value that mean (P) is P (i);
Determine P
pdp=P
av.
Wherein, α value and β value can be chosen as required, such as α=0.08, β=0.02.
When using according to logarithm waveform attenuating, the spectrum shape of power delay spectrum as shown in Figure 5, in step s 201, determines power delay spectrum P
pdpspecifically comprise the steps:
First base band sampled point StarPoint=α L in the middle of setting
cp, wherein L
cpfor CP length, α is Optimal Parameters, α=0.08 in Fig. 5;
Can determine that time-domain response value is: P (i)=1, i=1,2 ..., α L
cp,
P(i)=i
-0.2,i=α·L
cp+1,α·L
cp+2,…,L
cp;
Calculate P ' (i)=P (i)+χ, i=1,2 ..., L
cp+ 1, χ is Optimal Parameters, χ=0.2 in Fig. 5
Calculating is normalized to P ' (i), P
av(i)=P (i)/mean (P), i=1,2 ..., L
cp+ 1, wherein, the mean value that mean (P) is P (i);
Determine P
pdp=P
av.
Wherein, α value and β value can be chosen as required, such as α=0.08, β=0.02.
The embodiment of the present invention is also corresponding provides a kind of channel estimating apparatus, and as shown in Figure 6, this device comprises:
Channel related information determining unit 601, for according to power delay spectrum determination channel related information, power delay spectrum spectrum shape is specially, and before the middle base band time sampling point preset, time-domain response is steady state value, and after middle base band time sampling point, time-domain response decayed with the base band time;
Channel estimating unit 602, for carrying out channel estimating according to channel related information.
Wherein, channel related information determining unit 601 specifically for:
When channel estimation parameter does not change, determine the channel related information that channel related information uses when being the estimation of last secondary channel;
When channel estimation parameter changes, redefine channel related information according to power delay spectrum.
The circular that channel related information determining unit 601 carries out channel estimating with channel estimating unit 602 is identical with embodiment of the method, at this no longer repeated description.
The embodiment of the present invention provides a kind of channel estimation methods and device, when determining channel related information, according to power delay spectrum spectrum shape be: before the middle base band time sampling point preset, time-domain response is steady state value, and after described middle base band time sampling point, time-domain response decayed with the base band time; Channel estimating is carried out again by this channel related information.Because the power delay spectrum used has attenuation process, better can mate actual conditions, so use this power delay spectrum to carry out robustness that channel estimating improves channel estimating.
Obviously, those skilled in the art can carry out various change and modification to the present invention and not depart from the spirit and scope of the present invention.Like this, if these amendments of the present invention and modification belong within the scope of the claims in the present invention and equivalent technologies thereof, then the present invention is also intended to comprise these change and modification.