The application is dividing an application of application number is 01132898.3, the applying date is July 27 calendar year 2001 original bill application, and the priority date the earliest of this original bill application is on July 27th, 2000.
Background technology
In recent years, in mobile communication system, it is the transfer channel distribution method of purpose effectively to utilize frequency that people have proposed various, wherein part practicability.
Figure 15 shows that frequency division multiple access (Frequency Division MultipleAccess:FDMA), time-division multiple access (Time Division Multiple Access:TDMA) and road divide the channel configuration figure of multiple access various communication systems such as (Path Division Multiple Access:PDMA).
At first in conjunction with Figure 15, to FDMA, TDMA, PDMA does with brief description.Figure 15 (A) is depicted as the FDMA schematic diagram, and under the electric wave of the different frequency of f1~f4, user 1~4 analog signal is transmitted under the frequency division state, and each user's 1~4 signal is separated by frequency filter.
Figure 15 (B) is the TDMA schematic diagram, each user's digitized signal is under the electric wave of f1~f4 different frequency, transmitted by the time-division according to the regular hour (time slot), each user's signal is by separating the lock in time between frequency filter and base station and each the customer mobile terminal device.
On the other hand, in recent years, along with popularizing of portable telephone, in order to improve the utilization ratio of frequency, the PDMA mode has been carried.Shown in Figure 15 (C), this PDMA mode can be by 1 time slot of space segmentation same frequency in order to transmit a plurality of user data.Under this PDMA effect, each subscriber signal can utilize time synchronized between frequency filter and base station and each the customer mobile terminal device and the mutual interference blanking unit of adaptive array (adaptive array) etc. to separate.
The operation principle of this adaptive array radio base station is having explanation in the routine document down.
B.Widrow, et al.: " adaptive antenna system: Adaptive AntennaSystems ", Proc.IEEE, Vol.55, No.12, pp.2143-2159 (in December, 1967).
S.P.Applebaum: " adaptive array: Adaptive Arrays ", IEEE Trans.Antennas ﹠amp; Propag., vol.Ap-24, No.5, pp.585-598 (in September, 1976).
O.L.Frost, III: " linear all degree suppress the interior minimum square formation optimization of self adaptation: Adaptive Least Squares Optimization Subject to LinearEquality Constraints ", SEL-70-055, Technical Report, No.6796-2, Information System Lab., Stanford Univ, (in August, 1970).
B.Widrow and S.D.Stearns: " Adaptive Signal Processing: Adaptive SignalProcessing ", Prentice-Hall, Englewood Cliffs (1985).
R.A.Monzingo and T.W.Miller: " adaptive array introduction: Introduction to Adaptive Arrays ", John Wiley ﹠amp; Sons, New York (1980).
J.E.Hudson: " adaptive array principle: Adaptive Array Principles ", Peter Peregrinus Ltd, London (1981).
R.T.Compton, Jr. " adaptive antenna-notion and performance: AdaptiveAntennas-Concepts and Performance ", Prentice-Hall, EnglewoodCliffs (1988).
E.Nicolau and D.zaharia: " adaptive array: Adaptive Arrays ", Elsevier, Amsterdam (1989).
Figure 16 shows that the operation principle mould equation figure of this adaptive array radio base station.In Figure 16,1 adaptive array radio base station 1 is furnished with by n root antenna #1, #2, #3 ..., the array antenna 2 formed of #n, the zone that the 1st hatched example areas 3 its electric waves of expression among the figure can reach.On the other hand, the 2nd hatched example areas 7 is represented the electric wave effective coverage of other adjacent wireless base station 6.
In the 3rd zone, be the transceiver (arrow 5) of carrying out electric wave signal between mobile phone 4 and the adaptive array radio base station 1 in the terminal of user A.In the 7th zone, be the transceiver (arrow 9) of carrying out electric wave signal between mobile phone 8 and the wireless base station 6 in the terminal of other user B.
Here, if the electric wave signal frequency of the mobile phone 4 of the user A electric wave signal frequency with the mobile phone 8 of user B by chance is identical, then according to the position at user B place, the electric wave signal that is sent from the mobile phone 8 of user B just becomes unnecessary interference signal in zone 3, will be blended in the mobile phone 4 and the electric wave signal between the adaptive array radio base station 1 of user A.
Thus, adaptive array radio base station 1 is received is mixed signal from user A and user B both sides, if any processing is not done in base station 1 to this, what then exported will be user A and user B both sides' mixed signal, and the conversation of the feasible user A that can normally carry out originally is interfered.
The structure of existing adaptive array antenna and operation principle
In order to remove from the signal of this user B by handle from output signal, adaptive array radio base station 1 can carry out following processing.Figure 17 is the structural representation sketch map of adaptive array radio base station 1.
At first the signal of user A output is defined as A (t), the signal of user B output is defined as B (t), and the following equation of received signal x1 (t) among the 1st antenna #1 of the array antenna 2 of formation Figure 16 is expressed:
x1(t)=a1*A(t)+b1*B(t)
In the equation, a1, b1 are the real-time change coefficients, and this will talk about below.
The expression equation of the received signal x2 of the 2nd antenna #2 (t) is as follows:
x2(t)=a2*A(t)+b2*B(t)
In the equation, a2, b2 also are the real-time change coefficients.
The expression equation of the received signal x3 of the 3rd antenna #3 (t) is as follows:
x3(t)=a3*A(t)+b3*B(t)
In the equation, a3, b3 also are the real-time change coefficients.
Equally, the expression equation of the received signal xn (t) of n antenna #n is as follows:
xn(t)=an*A(t)+bn*B(t)
In the equation, an, bn also are the real-time change coefficients.
Described coefficient a1, a2, a3 ..., an represents because with respect to the electric wave signal from user A, antenna #1, the #2 of forming array antenna 2, #3 ..., separately relative position difference of #n (such as, the arranged spaced of each antenna between mutually is 5 times of electric wave signal wavelength, promptly about 1 meter) and the difference at the signal receiving strength that each antenna produced that causes.
Equally, described coefficient b1, b2, b3 ..., bn represents because with respect to the electric wave signal from user B, antenna #1, #2, #3 ..., the #n institute difference of the signal receiving strength of generation respectively.Because each user presents mobile status, so these coefficients also are real-time change.
These signal x1 (t), x2 (t) that receive by antenna, x3 (t) ..., xn (t) by corresponding transducer 10-1,10-2,10-3 ..., 10-n enters the receiving part 1R that constitutes adaptive array radio base station 1, and be transported to weight vector control part 11 and corresponding multiplicative operator 12-1,12-2,12-3 ..., in some among the 12-n.
To other inputs of these multiplicative operators, carry out from weight vector control part 11 weight w1, w2 at the received signal of each antenna, w3 ..., the weighted of wn.These weights are calculated in real time by weight vector control part 11, and this also will introduce below.
Like this, the signal x1 (t) that antenna #1 is received is by the computing of multiplicative operator 12-1, become w1* (a1A (t)+b1B (t)), the signal x2 (t) that antenna #2 is received is by the computing of multiplicative operator 12-2, become w2* (a2A (t)+b2B (t)), the signal x3 (t) that antenna #3 is received is by the computing of multiplicative operator 12-3, become w3* (a3A (t)+b3B (t)), and then the signal xn (t) that received of antenna #n becomes wn* (anA (t)+bnB (t)) by the computing of multiplicative operator 12-n.
These multiplicative operators 12-1,12-2,12-3 ..., the output valve of 12-n adds in adder calculator 13, its output is as follows:
w1(a1A(t)+b1B(t))+w2(a2A(t)+b2B(t))+w3(a3A(t)+b3B(t))+......+wn(anA(t)+bnB(t))
These are divided into about the item of signal A (t) and the item of signal B (t), become following form:
(w1a1+w2a2+w3a3+......+wnan)A(t)+(w1b1+w2b2+w3b3+......+wnbn)B(t)
Here, as hereinafter will introducing, adaptive array radio base station 1 couple of user A, B discern, according to the principle of only extracting required user's output signal out, calculate described weight w1, w2, w3 ..., wn.Such as, in Figure 17 example, weight vector control part 11 is for only to release from the signal A of legal calling user A (t), with coefficient a1, a2, a3......, an, b1, b2, b3 ..., bn handles as fixed number, the coefficient of whole signal A (t) is 1, the coefficient of whole signal B (t) is 0, calculate weight w1, w2, w3 ..., wn.
That is to say that weight vector control part 11 is by once connecting finding the solution of upright equation group to following, is 1 according to the coefficient of signal A (t), and the coefficient of signal B (t) is 0 principle, calculate in real time weight w1, w2, w3 ..., wn:
w1a1+w2a2+w3a3+......+wnan=1
w1b1+w2b2+w3b3+......+wnbn=0
Though this paper has omitted the solution explanation that this is once connected upright equation group, can understand from previous cited document, they have obtained practical application at present in adaptive array radio base station.
So, weight w1, w2, w3 ..., wn by described policy setting good after, the output signal of adder calculator 13 becomes following form:
Output signal=1*A (t)+0*B (t)=A (t)
User Recognition, training signal
Described user A, the recognition methods of B is as follows.
Figure 18 is the electric wave signal frame structure schematic overview of mobile phone.The electric wave signal of mobile phone is roughly by being the preposition symbol of known burst to the wireless base station and being that the data (voice etc.) of unknown burst are formed to the wireless base station.
In preposition symbol sequence, comprise being used to confirm whether this user is the information signal sequence of the legal calling user of base station approval.The training signal corresponding with user A that the weight vector control part 11 (Figure 17) of adaptive array radio base station 1 will be extracted out from memory 14 and the burst that is received compare, and think that according to extraction the principle of signal of the burst that comprised respective user A carries out weight vector control (decision weight).The signal of the user A that is extracted out is like this sent out by adaptive array radio base station 1 as output signal SRX (t).
On the other hand, in Figure 17, enter from the input signal STX (t) of outside in the signal transport unit 1T of a part that constitutes adaptive array base station 1, become multiplicative operator 15-1,15-2,15-3 ..., a side of the input signal of 15-n.Other input side of these multiplicative operators duplicate respectively and weighting according to weight vector control part 11 in earlier the weight w1, the w2 that come out of the calculated signals of reception, w3 ..., wn.
By these multiplicative operator weightings input signal, by corresponding converter 10-1,10-2,10-3 ..., 10-n, be transported to corresponding antenna #1, #2, #3 ..., #n, be sent in the zone 3 of Figure 16.
Here, the signal that same array antenna 2 transmits during to use and received signal, because carry out with user A the same with received signal is the weighting of target, so the electric wave signal that is transferred is collected mail by the mobile telephone set 4 of user A like also having the directive property of relative user A just.Figure 19 is that the electric wave signal between user A and the adaptive array radio base station 1 transmits and the figure that receives reflectionization.As the contrast of the area schematic 16 that is arrived with actual electric wave, it is the directive property electric wave signal emission state map of target that the imaginary area 3a on Figure 19 is depicted as the mobile telephone set 4 with user A that sends from adaptive array radio base station 1.
As mentioned above, in the PDMA mode, this technology that can remove same channel interference effect is necessary.In this point, the adaptive array that is intended to make disturbing wave to lose efficacy on using also can suppress disturbing wave effectively even be higher than at the level of disturbing wave under the situation of required electric wave level, is a kind of effective and efficient manner.
But adopt in the base station under the situation of adaptive array, the interference in the time of not only will removing received signal, but also should be able to reduce unwanted radioactive wave when signal transmits.
Array mould equation when the array mould equation when signal transmits can receive with signal is identical, but people are for estimating that according to input direction this method also had consideration.Though the latter is suitable for FDD (Frequency Division Duplexing (FDD)) and TDD (time division duplex), need carry out complicated processing.On the other hand, the former is under the situation that adopts FDD, and signal receives with the array mould equation that transmits different, so be necessary to carry out the correction of array configurations and weight.For this reason, in general, under the preceding topic that guarantees the TDD applicability, can obtain good characteristic having under the successional environment of outside time slot.
According to above introduction, adopt in the base station under the mode of adaptive array TDD/PDMA, the array mould equation that obtains in uplink (weight vector mould equation) is when being used for downgoing line, transmit in imagination under the situation of the bigger dynamic Rayleigh transmission degree of angle, because the time difference between the uplink and downlink circuit, the signal transmission direction in the downgoing line may become bad.
In a word, since from the user terminal of uplink (up link) to base station transmit signals to there is a time interval to user terminal transmits from the base station by downgoing line (down link), so do not allowing to ignore under the situation of user terminal translational speed, because the error between the orientation, place of signal of base station transmit direction and actual user's terminal, the directivity that signal is transmitted becomes bad.
This paper proposes a kind of method of the response signal vector of downgoing line being estimated by the extrapolation process of utilizing the resulting collection of letters response vector of uplink, to use the evaluation method of weight as the downgoing line of considering signal transmission road change.
But, because noise or sampling error in collecting mail if the collection of letters response vector of estimating in uplink has error, then also have error among the extrapolation process result, the estimation of the response vector of posting a letter of downgoing line just can not be correct, thereby be difficult to carry out good signal sender tropism's control.
Here, transmit the change of the transmission environment on road by the signal reception coefficient that transmits the road, promptly attenuation degree reflects.The physical quantity of the described degree of decay is represented with Doppler frequency (FD).
Promptly owing to according to the attenuation degree that transmits the road, can produce error in the result of extrapolation process, institute thinks the generation that prevents the extrapolation process error, is necessary to understand the attenuation degree on transmission road, i.e. Doppler frequency.
The evaluation method of obtaining the reference signal correlation estimation attenuation degree that in the collection of letters of surrounding time, is comprised in the past just once the someone propose, open introduction in announcing for flat 7-162360 number such as the spy.But, in this existing method,,, be difficult to carry out correct estimation so can contain more interfering components owing to be to utilize the reference signal that is contained in the received signal itself to calculate correlation.
In addition, owing to fix the opportunity of reference signal, can not lack the adaptability that calculation is handled calculating correlation arbitrarily opportunity.
On the other hand, also do not develop the evaluation method of handling the transmission road Doppler frequency of estimating each separated user terminal by adaptive array at present.
Summary of the invention
Even the objective of the invention is to release a kind of collection of letters response vector of estimating in uplink estimation error is arranged, also can correctly estimate, thereby can realize the unlimited device that good signal sender tropism controls the response vector of posting a letter of downgoing line.
Another object of the present invention is to release a kind of influence of the interfering components in not collected mail, and has bigger calculation to handle adaptability, the Doppler frequency estimation circuit that can estimate the Doppler frequency of each user terminal.
A further object of the present invention is, releases the influence that a kind of employing is not subjected to the interfering components in the received signal, and the adaptability that has bigger calculation to handle, the wireless device of the Doppler frequency estimation circuit that can estimate each user terminal Doppler frequency.
According to an aspect of the present invention, but provide a kind of real time altering antenna directivity, to and a plurality of terminal between signal receive and the wireless device that transmits divisional processing when carrying out, it has the antenna of a plurality of arranged discrete and receives the signal sending circuit and the receiving circuit of shared a plurality of antennas when transmitting at signal.When signal receiving circuit is included in received signal according to the collection of letters separated part of the signal from the particular terminal in a plurality of terminals being separated from the signal of a plurality of antennas and when the received signal according to the collection of letters of estimating from the collection of letters response vector on the transmission road of particular terminal being transmitted road estimation portion from the signal of a plurality of antennas.Signal sending circuit comprises the originator tropism control part that the antenna directivity when the estimation result who posting a letter of estimating of the response vector of posting a letter on the transmission road of letter signal when sending is transmitted road estimation portion and transmitted road estimation portion according to posting a letter according to the estimation result who collect mail to transmit road estimation portion sends signal upgrades.Post a letter and transmit road estimation portion and comprise according to the extrapolation process that transmits by collecting mail that road estimation portion estimates from a plurality of collection of letters response vector of the uplink time slot of particular terminal, calculate towards the extrapolation process portion of the described response vector of posting a letter of the forward link slot of particular terminal, with the memory of preserving a plurality of parameters that are used for extrapolation process of being scheduled to according to the transmission environment that transmits the road, with the transmission environment that transmits the road is estimated, the transmission environment corresponding parameter of from a plurality of parameters of being preserved, selecting and estimating, and be applied to be undertaken the selection portion of extrapolation process by extrapolation process portion.
So, even in the collection of letters response vector of in uplink, being estimated estimation error is arranged, by be used for the parameter of extrapolation process according to the transmission environmental selection that transmits the road, just can correctly estimate the response vector of posting a letter of downgoing line, thereby can realize the control of good transmission sense.
Comparatively ideal situation is, parameter is the extrapolation distance when carrying out extrapolation process by extrapolation process portion, memory can be preserved a plurality of extrapolation distances of being scheduled to according to the Doppler frequency of expression transmission environment, selection portion can be estimated the Doppler frequency that transmits the road, the corresponding extrapolation distance of Doppler frequency of from a plurality of extrapolation distances of being preserved, selecting and estimating, and be applied to the extrapolation process of being undertaken by extrapolation process portion.
Like this, if the collection of letters response vector of uplink has estimation error, the extrapolation distance during owing to extrapolation process is long more, and extrapolation error is just big more, so, can carry out the estimation of the correct response vector of posting a letter by selecting the corresponding extrapolation distance of Doppler frequency in expression transfer ring border.
Better is, the Doppler frequency of being estimated is low more, and the selected extrapolation distance of selection portion should be short more, and Doppler frequency is high more, and the extrapolation distance of selection should be long more.
So, because Doppler frequency is low more, it is just more little to transmit environmental variations, so by dwindling extrapolation distance, just can prevent to surpass the extrapolation of actual change amount, because Doppler frequency is high more, it is just big more to transmit environmental variations, so after extrapolation distance enlarges, can carry out outer push operation fully.
Better again way is, parameter becomes the extrapolation distance that reflects when extrapolation process portion carries out extrapolation process, memory can be preserved a plurality of extrapolation distances of being scheduled to according to the signal errors between separation signal and the desired signal that reflection transmits environment, selection portion can be estimated transmitting the road signal errors, from a plurality of extrapolation distance values of preserving, select the corresponding extrapolation distance of signal errors that estimates, and be applied to the extrapolation process of being undertaken by extrapolation process portion.
So if signal errors is bigger, then the estimation error of the collection of letters response vector of uplink also will increase, extrapolation error also can increase, like this, select extrapolation distance, just can carry out the estimation of the correct response vector of posting a letter by the signal errors that transmits environment according to reflection.
Comparatively ideally be, estimating signal error is big more, and the extrapolation distance that selection portion is selected should be more little, and signal errors is more little, and the extrapolation distance of selection is big more.
Like this, because the big more extrapolation error of signal errors is then big more, thus just can suppress extrapolation error after shortening extrapolation distance, because the more little extrapolation error of signal errors is then more little, so after prolonging extrapolation distance, just can carry out outer push operation fully.
More desirable again be, extrapolation distance when parameter is carried out extrapolation process for extrapolation process portion, memory can be preserved a plurality of extrapolation distances of being scheduled to according to error between Doppler frequency and separated signal and the required signal that reflection transmits environment, selection portion can estimate and Doppler frequency and the corresponding extrapolation distance of signal errors selecting to estimate from a plurality of extrapolation distances of preserving the Doppler frequency and the signal errors that transmit the road, and is applied to the extrapolation process of being undertaken by extrapolation process portion.
So, if Doppler frequency and signal errors are greater than extrapolation error, then by selecting to transmit the Doppler frequency of environment and the extrapolation distance of signal errors based on reflection, just can carry out the estimation of the correct response vector of posting a letter.
More desirable again be that selection portion is temporarily selected earlier and the corresponding extrapolation distance of estimating of Doppler frequency, according to the signal errors of being estimated the extrapolation distance of the temporary transient selection of institute is revised again.
Like this, carry out basic extrapolation distance according to the Doppler frequency that extrapolation error is had considerable influence and select,, just can carry out the estimation of the correct response vector of posting a letter again by the extrapolation distance based on signal errors is revised.
More better is that the relation that transmits between environment and a plurality of parameter should be set respectively according to each wireless device.
Like this, because the corresponding relation of the transmission environment of each wireless device and parameter is to measure respectively well in advance, so can estimate the higher response vector of posting a letter of precision.
More better is that the relation that transmits between environment and a plurality of parameter should decision jointly on a plurality of wireless devices.
Like this, under the less situation of the individual difference between the wireless device,, can simplify the manufacture process of wireless device by between a plurality of wireless devices the corresponding relation that transmits between environment and the parameter being changed jointly.
Based on another aspect of the present invention, but at this real time altering antenna directivity, to and each terminal between signal receive with the wireless device that transmits divisional processing when carrying out in, to and particular terminal between the Doppler frequency estimation circuit estimated of the Doppler frequency on transmission road, comprise the signal that a plurality of antenna received based on arranged discrete, the collection of letters separated part that signal from the particular terminal in a plurality of terminals is separated, based on the signal that a plurality of antenna received, the collection of letters of estimating from the collection of letters response vector on the transmission road of particular terminal is transmitted road estimation portion, collection of letters response vector based on the surrounding time of being estimated by the transmission road estimation portion that collects mail, the relevant calculation portion of compute vectors correlation, based on the vector correlation value that is predetermined by experience and the corresponding relation of Doppler frequency, the estimation portion that the pairing Doppler frequency of being calculated by relevant calculation portion of vector correlation value is estimated.
Like this, be not to be received signal itself by obtaining, but the correlation of collection of letters response vector, the influence of the composition that just can not be interfered, and then can carry out correct estimation to the Doppler frequency on the transmission road of each separated particular terminal.
It is desirable to, preferably include the moment correlation and calculating part that it is exported as the vector correlation value between the collection of letters response vector that can calculate surrounding time in the relevant calculation portion.
So, because can be at the correlation of obtaining opportunity arbitrarily between the collection of letters response vector, so can carry out correct estimation to the moment Doppler frequency that transmits the road.
Better is, preferably include in the relevant calculation portion between the collection of letters response vector that can calculate surrounding time moment correlation calculating part and by fixed weight coefficient the correlation in past of being calculated by calculating part and present correlation are weighted the equalization portion that average resulting mean value is exported as the vector correlation value.
Like this,, make the Doppler frequency of moment produce error, also can handle, do not influenced, and can carry out correct estimation Doppler frequency by this error by the equalization of correlation even rapid decay takes place.
Better is, in the fixed weight coefficient, be arranged at the weight of the correlation in past big, little at the weight of present correlation.
Like this, increase by correlation weighting in equalization is handled to the past, even thereby owing to the Doppler frequency error of moment takes place rapid relaxation phenomenon, can not influenced, correctly the estimation of Doppler frequency by this error yet.
Better is that relevant calculation portion can calculate the vector correlation value according to the collection of letters response vector in frame slot of collection of letters response vector in the present frame time slot and front.
Like this, collection of letters response vector is different with the reference signal that prior art is adopted, no matter can both obtain correlation values under the when machine in office, so the adaptability of correlation calculation has obtained increase.
Better is that relevant calculation portion preferably can calculate the vector correlation value according to the collection of letters response vector in the time slot of the error of not collecting mail nearest in collection of letters response vector in the present frame time slot and the past frame time slot.
Like this, collection of letters response vector, the reference signal that is adopted with prior art is different, no matter which can both obtain correlation values opportunity at, so the adaptability of correlation calculation has obtained increase, especially can eliminate the influence of collection of letters error.
Better is that relevant calculation portion can calculate the vector correlation value according to the collection of letters response vector of same time slot first half and latter half of collection of letters response vector.
Like this, collection of letters response vector is different with the reference signal that prior art is adopted, no matter which can both obtain correlation values opportunity at, and then the adaptability of correlation calculation has obtained increase.
According to a further aspect in the invention, but real time altering antenna directivity, to and each terminal between signal receive and the wireless device that transmits divisional processing when carrying out, comprise the antenna of a plurality of arranged discrete and receive the signal sending circuit and the receiving circuit of shared a plurality of antennas when transmitting at signal.When collection of letters circuit is included in received signal according to the collection of letters separated part of the signal from the particular terminal in a plurality of terminals being separated from the signal of a plurality of antennas and when the received signal according to the collection of letters of estimating from the collection of letters response vector on the transmission road of particular terminal being transmitted road estimation portion from the signal of a plurality of antennas.Transmitting circuit comprises the originator tropism control part that the antenna directivity when the estimation result who posting a letter of estimating of the response vector of posting a letter that transmits the transmission road of signal when transmitting is transmitted road estimation portion and transmitted road estimation portion according to posting a letter according to the estimation result who collect mail to transmit road estimation portion transmits sending signal upgrades.Post a letter and transmit road estimation portion and comprise according to the extrapolation process that transmits from collecting mail that road estimation portion estimates from a plurality of collection of letters response vector of the uplink time slot of particular terminal, calculate towards the extrapolation process portion of the response vector of posting a letter of the forward link slot of particular terminal, with the memory that is used to preserve a plurality of parameters that are used for extrapolation process of being scheduled to according to the Doppler frequency that transmits the road, with the Doppler frequency corresponding parameter of from a plurality of parameters of being preserved, selecting Yu estimate, and be applied to be undertaken the selection portion of extrapolation process by extrapolation process portion.Doppler frequency estimation portion is according to the collection of letters response vector that collect mail to transmit the surrounding time that road estimation portion estimated, the relevant calculation portion of compute vectors correlation, and according to by corresponding relation between predetermined vector correlation value of experience and the Doppler frequency, the estimation portion that the pairing Doppler frequency of vector correlation value that is calculated by relevant calculation portion is estimated.
Like this, because what obtained is not received signal itself, but the correlation of collection of letters response vector, the influence of composition so can not be interfered, thereby can carry out correct estimation to the transmission road Doppler frequency of each separated particular terminal, even, still can carry out correct estimation, thereby can realize good transmission sense control to the response vector of posting a letter of downgoing line so in the estimation collection of letters response vector of uplink estimation error is arranged.
It is desirable to, comprise the moment correlation between the collection of letters response vector that can calculate surrounding time in the relevant calculation portion, and its calculating part as the output of vector correlation value.
Like this, because can be at the correlation of obtaining opportunity arbitrarily between the collection of letters response vector, so can carry out correct estimation to the moment Doppler frequency that transmits the road, thus can carry out the estimation of the more accurate response vector of posting a letter.
Better is, comprise in the relevant calculation portion between the collection of letters response vector that can calculate surrounding time moment correlation calculating part and to the past correlation that calculates by calculating part and correlation at present by decide weight coefficient be weighted average after the equalization portion that exports as the vector correlation value of resulting mean value.
Like this, even rapid decay takes place, make moment Doppler frequency generation error, because correlation has carried out the equalization processing, so can not be subjected to the influence of this error, can carry out correct estimation to Doppler frequency, thereby can carry out the correct response vector estimation of posting a letter.
Better is, big to the weight of past correlation in determined weight coefficient, little to the weight of present correlation.
Like this, when averaging processing, because the correlation weighting to the past is bigger, even so owing to the Doppler frequency error takes place rapid relaxation phenomenon, can do not influenced by this error yet, estimation of Doppler frequency more correctly, thus can correctly post a letter the response vector estimation.
Better is, relevant calculation portion can calculate correlation according to the collection of letters response vector of the frame slot of collection of letters response vector and front of present frame time slot.
Like this, collection of letters response vector is different with the reference signal that prior art is adopted, no matter which can both obtain correlation values under the condition opportunity at, so the adaptability of correlation calculation has obtained increase.
Better is, relevant calculation portion can calculate the vector correlation value in the collection of letters response vector in the nearest time slot of the collection of letters response vector of present frame time slot and the error of not collecting mail in the past frame time slot.
Like this, collection of letters response vector is different with the reference signal that prior art is adopted, no matter which can both obtain correlation values under the condition opportunity at, so the adaptability of correlation calculation has obtained increase, especially can eliminate the influence of received signal error.
Better is that relevant calculation portion can calculate the vector correlation value according to the collection of letters response vector and the latter half of collection of letters response vector of same time slot first half.
In the present invention, collection of letters response vector is different with the reference signal that prior art is adopted, no matter which can both obtain correlation values under the condition opportunity at, so the adaptability of correlation calculation has obtained increase.
Embodiment
Figure 1 shows that the structured flowchart of embodiment of the present invention PDMA with base station radio device (wireless base station) 1000.
As shown in Figure 1, for identification user PS1 and PS2,4 antenna #1~#4 have been set up.The number of these antenna generally can be N root (N: natural number).
In the transmitting and receiving system 1000 shown in Figure 1 receiving part SR1 is arranged, it is used to receive the signal from antenna #1~#4, to corresponding user, separate such as the signal of user PS1, and the ST1 of portion that posts a letter in addition, it is used for sending signal to user PS1.Antenna #1~#4 with switch selectively the receiving part SR1 and being connected between the ST1 of portion of posting a letter by transducer 10-1~10-4.
I.e. collection of letters signal RX1 (t), RX2 (t), RX3 (t), the RX4 (t) that receives at each antenna, transducer 10-1,10-2,10-3,10-4 by correspondence enter receiving part SR1, be imported into collection of letters weight vector computer 20, collection of letters response vector computer 22, and the input of corresponding each multiplicative operator 12-1,12-2,12-3,12-4.
In other input of these multiplicative operators, have from the pairing weight coefficient wrx11 of the received signal at each antenna of collection of letters weight vector computer 20, wrx21, wrx31, wrx41.These weight coefficients are the same with the example of prior art, calculate in real time by collection of letters weight vector computer 20.
The portion ST1 of posting a letter comprises the collection of letters response vector of being calculated from 22 receptions of collection of letters response vector computer, then as hereinafter introducing, transmission road when signal is transmitted is estimated, hypothesis collection of letters response vector when promptly signal being transmitted is estimated, obtain the response vector estimating device 32 of posting a letter of the response vector of posting a letter, and the transceive data between the response vector estimating device 32 of posting a letter, and memory of data 34 is preserved in storage, estimation result based on the response vector estimating device 32 of posting a letter, calculate the weight vector computer 30 of posting a letter of the weight vector of posting a letter, accept letter signal in side input respectively, import the weight coefficient wtx11 that has from the weight vector computer 30 of posting a letter other side, wtx21, wtx31, the multiplicative operator 15-1 of wtx41,15-2,15-3,15-4.The output of multiplicative operator 15-1,15-2,15-3,15-4 is sent on antenna #1~#4 by transducer 10-1~10-4.
In addition, can carry out receiving part SR1 and the setting of the same structure of the ST1 of portion that posts a letter at each user, this does not mark in Fig. 1 yet.
The adaptive array operation principle
Below brief description is carried out in the work of receiving part SR1.
The received signal RX1 that receives on antenna (t), RX2 (t), RX3 (t), RX4 (t) represent with following equation.
Several 1
RX1(t)=h11Srx1(t)+h12Srx2(t)+n1(t) ......(1)
RX2(t)=h21Srx1(t)+h22Srx2(t)+n2(t) ......(2)
RX3(t)=h31Srx1(t)+h32Srx2(t)+n3(t) ......(3)
RX4(t)=h41Srx1(t)+h42Srx2(t)+n4(t) ......(4)
Here, the received signal of the antenna of signal RXj (t) expression j number (j=1,2,3,4), the signal that the user transmitted that signal Srxi (t) expression is i number.
In addition, the plural coefficient that coefficient h ji represents that the j antenna received, the noise that is contained in the received signal that nj (t) expression is j number from i user's signal.
The vector expression equation of last equation (1)~(4) is as follows.
Several 2
X(t)=H1Srx1(t)+H2Srx2(t)+N(t) ......(5)
X(t)=[RX1(t),RX2(t),......,RX4(t)]
T
......(6)
Hi=[h1i,h2i,......,h4I]
T,(i=1,2) ......(7)
N(t)=[n1(t),n2(t),......,n4(t)]
T ......(8)
In equation (6)~(8), [...]
TThe transposition of expression [...].
Here, X (t) expression input signal vector, Hi represents i number user's collection of letters response vector, N (t) expression noise vector.
Adaptive array antenna as shown in Figure 1, its to from the input signal of each antenna by coefficient wrx1i~~wrx4i is weighted and synthetic signal is exported as received signal SRX (t).
Under above preparation, such as, the work of the adaptive array when extracting the situation of No. 1 signal Srx1 (t) that the user sent out is as follows.
The output signal y1 of adaptive array 100 (t) takes advantage of calculation by the vector of input signal vector X (t) and weight vector W1, can use equation expression down.
Several 3
y1(t)=X(t)W1
T ......(9)
W1=[wrx11,wrx21,wrx31,wrx41]
T ......(10)
That is to say that weight vector W1 is to be the vector of element with the weight coefficient wrxj1 (j=1,2,3,4) that j number input signal RXj (t) is weighted usefulness.
Here, if replace y1 (t) in the equation (9) with the input signal vector X (t) of equation (5), then equation becomes following form.
Several 4
y1(t)=H1W1
TSrx1(t)+H2W1
TSrx2(t)+N(t)W1
T
......(11)
Here, when adaptive array 100 was worked ideally, by the method for generally acknowledging, weight vector W1 connected upright equation group below satisfying, and is controlled in order by weight vector control part 11.
Several 5
H1W1
T=1 ......(12)
H2W1
T=0 ......(13)
Satisfy equation (12) and equation (13), weight vector W1 is controlled fully, is finally represented by following equation from the output signal y1 (t) of adaptive array 100.
Several 6
y1(t)=Srx1(t)+N1(t) ......(14)
N1(t)=n1(t)w11+n2(t)w21+n3(t)w31+n4(t)w41
......(15)
That is to say, in output signal y1 (t), can obtain the 1st signal Srx1 (t) that the user sent among 2 users.
The work summary of wireless device 1000
Fig. 2 is the description of flow diagram as the groundwork summary of the wireless device of inscribing before the present invention 1000.
In wireless device 1000, the weight vector (weight coefficient vector) that is conceived to adaptive array is expressed by the collection of letters response vector one meaning ground of each antenna element, by the estimation to the time fluctuation of collection of letters response vector, indirectly weight is estimated.
At first, in receiving part SR1, estimate (step S100) according to received signal transmission road to received signal.In the estimation that transmits the road, be equivalent in equation (1)~(4), obtain impulse response from user's conveying signal.
In other words, in equation (1)~(4), if can estimate collection of letters response vector H1, the transmission road in the time of then can estimating reception from the signal of user PS1.
Then, the transmission road the when response vector of posting a letter estimating device 32 can send signal predicts, that is to say the collection of letters response vector (step S102) when collection of letters response vector in the time of can receiving according to signal dopes signal and sends.Transmission response vector when predicted collection of letters response vector is equivalent to the signal transmission.
Then, the weight vector of posting a letter computer 30 is according to the predicted response vector of posting a letter, and the weight vector of posting a letter is calculated, and is transported to multiplying device 15-1~15-4 last (step S104).
The work of collection of letters response vector computer 22
Below the groundwork as prerequisite of the present invention of collection of letters response vector computer 22 shown in Figure 1 is described.
At first, be 4 at the antenna element number, Tong Xin number of users is under 2 people's the situation simultaneously, expresses with described equation (1)~(4) from the signal of signal receiving circuit output through each antenna.
At this moment, the antenna receiving signal that equation (1)~(4) are expressed is expressed with down routine (5)~(8) vector expression equation again.
Several 7
X(t)=H1Srx1(t)+H2Srx2(t)+N(t) ......(5)
X(t)=[RX1(t),RX2(t),......,RXn(t)]
T
......(6)
Hi=[h1i,h2i,......,hni]
T,(i=1,2) ......(7)
N(t)=[n1(t),n2(t),......,nn(t)]
T ......(8)
(0092)
Here, if adaptive array can work well owing to will each user's signal be separated and extract out, so described signal Srxi (t) (i=1,2) is just become given value.
At this moment, utilize signal Srxi (t) to be the condition of known signal, just can derive collection of letters response vector H1=[h11, h21, h31, h41 according to the method for following explanation] and H2=[h12, h22, h32, h42].
That is to say,, merge, calculate population mean (time average) such as signal Srx1 (t) from the 1st user to received signal with known subscriber signal.
Several 8
E[X(t)*Srx1
*(t)]=H1*E[Srx1(t)*Srx1
*(t)]
+H2*E[Srx2(t)*Srx1
*(t)]
+E[N(t)*Srx1
*(t)] ......(16)
In equation (16), E[......] express time is average, S
*(t) conjugate complex number of expression S (t).When the processing time of equalization is fully long, then this mean value will be following form.
Several 9
E[Srx1(t)*Srx1
*(t)]=1 ......(17)
E[Srx2(t)*Srx1
*(t)]=0 ......(18)
E[N(t)*Srx1
*(t)]=0 ......(19)
Here, the value of equation (18) is 0, is because signal Srx1 (t) does not have correlation each other with signal Srx2 (t).And the value of equation (19) is 0, is because signal Srx1 (t) does not have correlation each other with noise signals N (t).
Like this, the population mean of equation (16) the response vector H1 that equals to collect mail, as follows.
Several 10
E[X(t)*Srx1
*(t)]=H1 ......(20)
By above step, can estimate collection of letters response vector H1 from the transmission signal of the 1st user PSI.
Equally, by input signal vector X (t) is asked calculation with the overall average of signal Srx2 (t), can estimate collection of letters response vector H2 from the transmission signal of the 2nd user PS2.
Ask in the calculation carrying out described population mean, such as, can be listed as at the data symbol of the data symbol row of the stated number ahead in certain 1 time slot when signal receives and last stated number and carry out.
The estimation of the response vector of posting a letter
Fig. 3 is the groundwork key diagram as the response vector estimating device of inscribing before the present invention 32 of posting a letter.Under the PDMA burst mode, per 4 users distribute 8 time slots respectively in circuit up and down.The structure of this time slot is, is the 1st training character row such as 31 characters of front, and 68 characters of back are the data character row, and rearmost 31 characters are the 2nd training character row.
As mentioned above, be provided with the training character row,, can calculate both sides' collection of letters response vector by the algorithm in the described collection of letters response vector computer 22 at the front and the end of uplink time slot.
Then, can estimate the collection of letters response vector that downgoing line is used by extrapolation process (linear extrapolation).
That is to say, when if the value of any one moment t of collection of letters response vector element is f (t), then the last training character that is listed in the value f (t0) of t0 constantly and uplink time slot according to the front of uplink time slot training character is listed in the value f (t1) of moment t1, just can dope forward link slot by the following method at the value f of moment t (t).
f(t)=[f(t1)-f(t0)]/(t1-t0)×(t-t0)+f(t0)
Here, in described explanation, in the front of uplink time slot be provided with the training character row at last, though only carry out extrapolation process one time, if but the mid portion of uplink time slot also is provided with the training character row, then can with 2 extrapolations f (t) value of moment t be estimated from 3 point value the uplink time slot of collection of letters response vector.If or increase the position that is provided with of training character row in the uplink time slot, then can carry out the more extrapolation process of high order.
The present invention involves the improvement of the evaluation method of the collection of letters of the downgoing line in extrapolation process (posting a letter) response vector and the necessary Doppler frequency estimation circuit of this improvement, detailed content will be introduced below, and elder generation is explained the decision of the weight vector of posting a letter here.
The decision of the weight vector of posting a letter
As mentioned above, behind the estimated value of the collection of letters response vector when obtaining the signal transmission, just can obtain the weight vector of posting a letter according to any a kind in following 3 kinds of methods.
I) orthogonalization method
At the moment of user PS1 t=iT (i: natural number, T: under the situation unit interval), can think that weight vector W (1) (i)=[wtx11, wtx12, wtx13, wtx14].Because user PS2 trends towards zero, so can meet the following conditions.
For user PS2, the transmission road (collection of letters response vector) of prediction be made as V (2) (i)=[h1 ' (2) (i), h2 ' (2) (i), h3 ' (2) (i), h4 ' (2) is (i)].The hp ' here (q) (i) is q number user's the pairing predicted value of moment i of p number the corresponding collection of letters response vector of antenna.Equally, also can dope transmission road V (1) (i) for user PS1.
At this moment, (i) TV (2) is (i)=0 under the situation at W (1), and decision W (1) (i).Restrictive condition is as follows.
C1) W (1) (i) TV (1) (i)=g (certain value)
C2) ‖ W (1) (i) ‖ reach minimum.
Condition c2 is equivalent to the signal transmitting power minimum.
Ii) use the method for spurious correlation matrix
Here, as mentioned above, adaptive array is made up of several antenna elements and each unit weights control section.Generally speaking, the antenna input vector is represented with X (t), weight vector is represented with W, reach with output Y (t)=WTX (t) and reference signal d (t) mean square deviation between the two minimum be under principle the situation that weight vector is controlled (the MMSE benchmark: least squares error method benchmark), by descending equation calculating optimum weights W opt (Wiener separates).
Several 11
Wopt=Rxx
-1rxd ......(21)
(0117)
But should satisfy following condition,
Several 12
Rxx=E[x
*(t)x
T(t)] ......(22)
rxd=E[x
*(t)d(t)] ......(23)
Here, Y
TThe transposition of expression Y, Y
*The complex field of expression Y, E[Y] the expression overall average.According to this weighted value, can under the preceding topic that adaptive array suppresses unwanted disturbing wave, generate array mould equation.
Just, when adopting the spurious correlation matrix method, last equation (21) is calculated by the spurious correlation matrix of following explanation.
That is to say, utilize the n (i) of the plural received signal coefficient h estimated ' (k), the weight vector W (k) that calculates user K (i).K number user's array response vector is made as V (k) (i), calculates according to following method.
Several 13
V
(K)(i)=[h1′
(k)(i),h2
(k)(i),......,hN′
(K)(i)]
T
......(24)
At this moment, the autocorrelation matrix Rxx (i) of the hypothesis received signal in the t=iT condition utilize V (k) (i) following equation represent.
Several 14
But N is additional hypothesis noise item under the situation of integer.In calculating of the present invention, N=1.0 * 10
-5
The following equation of dependent vector rxd (i) between received signal and the reference signal is expressed.
Several 15
rxd(i)=V
(k)*(i) ......(26)
Therefore, just can obtain the weight of using at the downgoing line of moment t=iT according to equation (21) (25) (26).
In addition, when carrying out the inverse matrix computing of equation (25), can carry out best computing to user K according to the inverse matrix lemma.Particularly under 2 users' situation, can calculate weight according to following simple equation.
Several 16
W
(1)(i)=(p22+N)V
(1)*(i)-p12V
(2)*(i) ......(27)
W
(2)(i)=(p11+N)V
(2)*(i)-p21V
(1)*(i) ......(28)
pij=V
(i)H(i)V
(j)(i)
The weight vector computational methods of relevant autocorrelation matrix for example have introduction, document in following document: T.Ohgane, Y.Ogawa, and k.Itoh, Proc.VTC ' 97, vol.2, pp.725-729, May 1997, perhaps, document: Tanaka, great bell, coulee, her rattan, letter are learned skill newspaper, vol.RCS 98-117, pp.103-108, Oct, 1998.
Iii) with the method for wave beam user oriented PS1
Only be conceived to aim at the situation of user PS1, should satisfy following condition at wave beam.
w(1)(i)=v(1)(i)*
Under described any method, if send under the situation of the weight vector when decision is posted a letter, then transmit under the situation on road at dynamic Rayleighs such as hypothesis wide-angles, in the TDD/PDMA mode, also can suppress because the deterioration of the caused down link signal sender tropism of time difference between the uplink and downlink circuit.
Then, Fig. 4 A and Fig. 4 B are the explanation concept maps of the response vector estimation principle of posting a letter of the present invention." perfect condition " shown in Fig. 4 A is the further simplification to the concept map among Fig. 3 basically.
That is to say, 2 collection of letters response vector 1 and collection of letters response vector 2 in the same time slot of the uplink of in the step S100 of Fig. 2, calculating according to collection of letters response vector computer 22 by Fig. 1, pass through to the linear extrapolation that is carried out till the former opportunity of posting a letter of the pairing time slot of downgoing line, can be to the response vector estimation of correctly posting a letter of downgoing line.
Here, this Fig. 4 A " perfect condition " serves as preceding topic there not to be estimation error in collection of letters response vector 1 and 2.
But, shown in Fig. 4 B " under the situation that the response vector estimation error is arranged ", such as, when in collection of letters response vector 2, collecting mail the error of response vector 2 ' and so on owing to the estimation error that noise or sampling error produced, then when carrying out according to these collection of letters response vector 1 and 2 ' with " perfect condition " same (same extrapolation distance) linear extrapolation, the response vector error of posting a letter in signal transmission opportunity will enlarge, and what estimated will be the response vector of posting a letter of a mistake.
Therefore, the response vector of posting a letter based on this mistake, when (the step S104 of Fig. 2) handled in the weight decision of being posted a letter by the weight vector computer 30 of posting a letter of Fig. 1, the resulting weight of posting a letter also will be wrong, and this will cause the directivity mistake of downgoing line is transmit error signal.Especially because the distance between wireless base station and the terminal is longer, the minor error of directivity will become a main cause of big transmit error signal.
So, in the present invention, when there is estimation error in uplink collection of letters response vector, by carrying out the particularly adjusting of extrapolation distance of suitable parameters of extrapolation process according to transmitting the road environment, correctly estimate the response vector of posting a letter in the downgoing line, realize correct originator tropism.
Fig. 5 is the conceptual illustration figure of the extrapolation distance decision principle of embodiment of the present invention.
Transmit the road environment by receive the change of coefficient such as the signal that transmits the road, promptly attenuation degree is represented.The degree of decay is promptly represented (FD) with Doppler frequency as physical quantity.
Transmit the Doppler frequency FD in the environment, in Doppler frequency estimation principle of the present invention, roughly estimate with following method.Promptly calculate the correlation of 2 collection of letters response vector of surrounding time of each separated in adaptive array is handled user received signal.If there is not relaxation phenomenon, 2 collection of letters response vector unanimities then are just correlation is one.If it is bigger to decay, the difference of the response vector of then collecting mail will increase, and correlation will diminish.If the relation of such collection of letters response vector correlation and Doppler frequency FD is obtained in examination in advance, and it is deposited in memory, then by calculating the correlation of collection of letters response vector, just can estimate the Doppler frequency FD of this moment.The details of relevant Doppler frequency estimation circuit of the present invention are described below.
At first be conceived to represent the changeable frequency of reining in of attenuation degree, the relevant operation principle of the execution mode of extrapolation distance that determines of the present invention is explained.
As previously mentioned, influenced by estimation error in collection of letters response vector 2 and become under the situation of collection of letters response vector 2 ', extrapolation distance is long more, and extrapolation error is just big more, thereby increasing with the error of the original response vector of posting a letter.
In general, the degree of decay is more little, and promptly Doppler frequency FD is low more, and the change of the received signal coefficient on transmission road is just more little.So, in this case,, just can prevent to surpass the extrapolation of actual change amount by shortening extrapolation distance.More more specifically, be exactly under the lower situation of Doppler frequency FD, carry out the short distance extrapolation that the collection of letters response vector 2 ' from Fig. 5 (A) order to a of * sign, the response vector of posting a letter that a is ordered is estimated the response vector of correctly posting a letter that the b of conduct * sign is also ordered.
To this, it is big more to decay, and promptly Doppler frequency FD is high more, and the received signal coefficient change that transmits the road is just big more.So extrapolate fully in this case, by enlarging extrapolation distance.More more specifically, be exactly under Doppler frequency FD condition with higher, carry out collection of letters response vector 2 ' from Fig. 5 (B) to the extrapolation of longer distance of * sign c, estimate the correct response vector of posting a letter at the d point place of post a letter response vector and conduct * sign that c is ordered.
Described processing procedure is mainly carried out by the response vector estimating device 32 of posting a letter among Fig. 1.Fig. 6 is the flow chart that is conceived to the expression extrapolation process of this Doppler frequency FD.
With reference to Fig. 6, in step S1, at first transmit the estimation on road by collection of letters response vector computer 22, be exactly the collection of letters response vector 1 and 2 of estimating uplink specifically.
Next, in step S2, be that Doppler frequency FD estimates to the degree of decay according to the method that hereinafter will introduce in detail.
In step S3, utilize the response of posting a letter among Fig. 1 to vow reason estimating device 32, with the method for introducing among Fig. 5, determine the optimal extrapolation parameter of Doppler frequency FD, i.e. extrapolation distance.And under this target, the best extrapolation distance of measuring in advance and determining according to the height of Doppler frequency FD is kept in the memory 34 of Fig. 1 in advance.
Then, in step S4, utilize the extrapolation parameter (extrapolation distance) that in described step S3, is determined to carry out extrapolation process, carry out the estimation that downgoing line transmits the road again, the estimation of the response vector of promptly posting a letter.
At last, in step S5, the downgoing line that is determined in described step S4 according to the weight vector calculation element 30 of posting a letter among Fig. 1 response vector of posting a letter is to the weight estimation of posting a letter.
As mentioned above, in this embodiment, be to select optimal extrapolation distance, so, also can carry out the estimation of the correct response vector of posting a letter even in the collection of letters response vector of uplink estimation error is arranged according to the height of Doppler frequency FD.
Fig. 7 is the structural representation block diagram that is used for the Doppler frequency estimation circuit in embodiment of the present invention of described Doppler frequency FD estimation (the step S2 of Fig. 6).Below in conjunction with Fig. 7 the operation principle of the Doppler frequency estimation circuit in the embodiments of the present invention is explained.
In Fig. 7, the receiving part SR1 in response vector estimation circuit 101 corresponding diagram 1, the especially collection of letters response vector computer 22 in the corresponding diagram 1.
Here, 8 time slots of total be made up of 4 time slots respectively in the uplink and downlink circuit among Fig. 3 to Fig. 5 are called 1 frame.So such frame communication interaction continuous and the up-downgoing circuit on time series carries out.
Response vector estimation circuit 101 adopts previously described overall average method to received signal, the collection of letters response vector in the time slot of current time frame is estimated, and be transported on relevant calculation and Doppler frequency estimation circuit 103 and the memory 102.
Correlation in collection of letters response vector when relevant calculation and Doppler frequency estimation circuit 103 calculate current that response vector estimation circuit 101 estimated in the frame slot and the memory 102 between the interior collection of letters response vector of the pairing time slot of previous time frame of preservation.
And the collection of letters response vector correlation α in surrounding time closes 2 time frames fastening defines with following equation.
α=|h1h2
H|/|h1||h2|
Here, h2
HBe the complex conjugate of each element of h2, and then the expression transposition relation.
And, hi (i=1,2) expression with the phase amplitude information of each antenna element of time frame I be element collection of letters response vector (hi1, hi2, hi3, hi4).
So trying to achieve the correlation and the correct corresponding relation between the Doppler frequency that calculate is the comparison difficulty, by experiment, can rule of thumb obtain corresponding relation roughly.Such as, if correlation is in 1~0.95 the scope, then can estimate Doppler frequency FD is FD=0Hz.And, if correlation is in 0.95~0.80 scope, then can estimate FD=10Hz, or the like.
If the collection of letters response vector correlation of so rule of thumb obtaining pre-deposits in relevant calculation and the Doppler frequency estimation circuit 103 with roughly corresponding relation between the Doppler frequency FD, then can be by the correlation between the vector of obtaining by described equation, select this Doppler frequency estimated value, and transfer out by circuit 103.
In processing method shown in Figure 7, adopt such as digital signal processor (DSP) usually and implement software.Fig. 8 represents the process chart based on circuit structure shown in Figure 7.This processing shown in Figure 8 can be obtained the vector correlation value between the collection of letters response vector of the collection of letters response vector in the frame slot and the pairing time slot of previous time frame when current.
At first in step S31, estimate the collection of letters response vector in frame slot when current.
Next, in step S32, whether to the collection of letters response vector estimated among the step S31 is that the collection of letters response vector of initial estimation is judged, if the collection of letters response vector of initial estimation then stores it in the memory (memory 102 among Fig. 7) in step S35.
On the other hand, collection of letters response vector if not initial estimation, then in step S33, calculate collection of letters response vector in the pairing time slot of preserving in the memory of previous time frame and the correlation CORR between the collection of letters response vector in the pairing time slot of current time frame that step S31 estimates.
And, in step S34, according to by described experiment in advance vector correlation value and the corresponding relation of Doppler frequency FD that obtain and that storage is got up, estimation and the corresponding Doppler frequency FD of correlation CORR that exports and calculate.
On the other hand, in step S35, in step S31, estimate current the time collection of letters response vector in the frame slot store (memory 102 among Fig. 7) in the memory into.
Implementing above step S31~S35 repeatedly, is the moment vector correlation value between the collection of letters response vector in the time slot corresponding between continuous 2 time frames thereby can obtain surrounding time continuously.
Then, Fig. 9 represents the flow chart of the processing example of other form shown in Figure 8.
In Fig. 9, if do not detect the received signal mistake at step S41, then later work is identical with example among Fig. 8 basically, no longer repeat specification.On the other hand, if in step S41, detect the received signal mistake, then till not making mistake, detection can not enter step S31 then.
If do not detect the received signal mistake,, be the processing of step S33 ' under this situation with the difference of Fig. 8 though can implement work from step S31 to S35.That is to say, in the response vector estimation in the time slot that contains the received signal mistake, owing to be to get rid of according to step S41, so in step S33 ', to the collection of letters response vector in the nearest time slot that does not have the received signal mistake in the past tense frame slot of in memory, preserving and in step S31, estimate current the time correlation CORR between the collection of letters response vector in the frame slot estimate.Later processing is identical with example among Fig. 8.
In Fig. 9 example, can get rid of the influence of received signal mistake, thereby can carry out more correct Doppler frequency estimation.
Then, Figure 10 represents other process chart of circuit structure shown in Figure 7.In processing shown in Figure 10, can obtain the first half collection of letters response vector of same time slot and the vector correlation value between the latter half of collection of letters response vector.
At first to having or not the received signal mistake to judge (also can omit step S41) among the step S41, if do not detect the received signal mistake, then in step S42, the collection of letters response vector of the forward position part in the current time slots is estimated, next back in the current time slots in step S43 along the part response vector estimation of collecting mail.
Next in step S44, calculate described forward position part and back along the correlation CORR between the part collection of letters response vector separately.
Then, in step S45, vector correlation value of obtaining and storing according to described sample plot in advance and the corresponding relation between the Doppler frequency FD are estimated and are exported the pairing Doppler frequency FD of correlation CORR that calculates.
In the execution mode of Figure 10, the collection of letters response vector of surrounding time can obtain correlation, so can increase the adaptability of correlation value calculation because different with the reference signal that becomes the correlation value calculation object of the prior art an opportunity in office at described Fig. 8.
Then, Figure 11 represents slightly block diagram of this Doppler frequency estimation circuit structural representation of inventing other execution modes.Circuit structure shown in Figure 11 is different at following some and circuit structure shown in Figure 7.
Promptly in Fig. 7 example, relevant calculation and Doppler frequency estimation circuit 103 can be obtained the moment correlation between 2 collection of letters response vector of surrounding time, more by the mean value between the correlation that obtains to be calculated, can carry out the estimation of more correct Doppler frequency to the relevant calculation among this Figure 11 and Doppler frequency estimation circuit 105.
More more specifically say, in Figure 11, correlation between the collection of letters response vector in collection of letters response vector when relevant calculation and Doppler frequency estimation circuit 105 can be calculated out estimated current in response vector estimation circuit 101 in the frame slot and the pairing time slot of in memory 102, preserving of preceding time frame, and get the correlation that just calculates out and from memory 104, read before quilt calculate mean value between the average correlation in stored past also.
Particularly in this embodiment, can obtain the weighted average of regulation weight coefficient.Such as, in circuit 105, can calculate average correlation of past to storage in the memory 104 after by the 1st weight coefficient weighting numerical value and to the mean value between the current correlation that just the calculated numerical value after by the 2nd weight coefficient weighting.
The 1st coefficient here can be made as the big coefficient such as 0.97, and the 2nd coefficient can be made as 0.03 less coefficient.
Based on the resulting average correlation of the average result after the weighting, circuit 105 estimations and the corresponding Doppler frequency of output simultaneously, deposit resulting current average correlation in the memory 104 preservation.
Figure 12 represents the process chart of circuit structure shown in Figure 11.In processing shown in Figure 12, the collection of letters response vector when current in the frame slot and when previous vector correlation value between the collection of letters response vector in the frame slot ask calculation, again past correlation and current correlation are weighted on average.
At first in step S51, the collection of letters response vector when current in the frame slot is estimated.
Next, in step S52, whether to the collection of letters response vector estimated among the step S51 is that the collection of letters response vector of initial estimation is judged, if the collection of letters response vector of initial estimation, then in step S59 it is deposited in the memory (memory 102 among Figure 11) and preserves.
On the other hand, collection of letters response vector if not initial estimation, then in step S53, calculate collection of letters response vector in the time slot of the previous time frame of preserving in the memory and the correlation CORRNOW between the collection of letters response vector in the time slot of the current time frame that step S51 estimates.
Next, in step S54, whether be that the correlation that first calculated goes out is judged to the correlation that is calculated among the step S53, if the correlation that first calculated goes out, then in step S55 to CORR AVE substitution CORR NOW, averaging computing.
On the other hand, the correlation if not first calculated goes out then in step S56, is weighted average treatment with weight coefficient α and 1-α respectively on the past of reading from memory average correlation CORR OLD and current correlation CORR NOW.
Then in step S57, according to by described experiment in advance average correlation and the corresponding relation of Doppler frequency FD that obtain and that storage is got up, estimation and the corresponding Doppler frequency FD of average correlation CORR AVE that exports and calculate.
Next in step S58, the current average correlation CORR AVE that will calculate in step S55 or 56 deposits (memory 104 among Figure 11) in the memory in as the average correlation CORR OLD in past.
In step S59, again in step S51, estimate current the time collection of letters response vector in the frame slot deposit in the memory (memory 102 among Figure 11) in and preserve.
By repeating said steps S51-S59 repeatedly, can estimate Doppler frequency based on the vector correlation value after the equalization.By so obtaining the average of vector correlation value,, correctly estimate Doppler frequency even, also can not influenced by it owing to rapid decay produces moment Doppler frequency error.Particularly,, also can not influenced by it, correctly estimate Doppler frequency even owing to rapid decay produces the Doppler frequency mistake by the correlation in past is set big weight coefficient.
In addition, in Figure 11 and example shown in Figure 12, though can be according to averaging of the correlation processing between the collection of letters response vector in the pairing time slot of continuous time frame, but as Fig. 9 and shown in Figure 10, can also be when current averaging of correlation processing between the collection of letters response vector in the nearest time slot of no received signal mistake in collection of letters response vector in the frame slot and the past tense frame slot, and, also can be according to averaging of the correlation processing between first half under the same time slot and the latter half of collection of letters response vector.
As mentioned above, utilize Doppler frequency estimation circuit of the present invention, can be according to the height of the Doppler frequency FD of each terminal that correctly estimates, select optimal extrapolation distance, even in the collection of letters response vector of uplink estimation error is arranged, also can carry out the correct response vector estimation of posting a letter.
On the other hand, the transmission environment that transmits the road also can be represented by resulting signal weight estimation error from adaptive array output.This error uses the mean square deviation (Mean Square Error: to call MSE in the following text) between the desired signal value of resulting signal value and hope from adaptive array output to represent, this MSE is more little, just expression estimates desirable weight vector in uplink, and the precision of the output signal of adaptive array is just good more.Otherwise MSE is big more, and just expression does not estimate optimal weight vector in uplink, and the precision of the output signal of adaptive array just will worsen.In addition, be as you know about the computational methods of MSE, here no longer introduce.
Like this, in the estimation of the uplink collection of letters response vector error taking place according to the size of MSE, further causes the generation of extrapolation error.
Below being conceived to described MSE is explained the operation principle of embodiments of the present invention.
Return Fig. 5 now, bigger at MSE, under the bigger situation of the estimation error of collection of letters response vector, because extrapolation distance is long more, extrapolation error is also big more, so extrapolation distance is shortened.More particularly, be exactly under the bigger situation of MSE, as in the situation of Fig. 5 A, carry out the short distance extrapolation of a from collection of letters response vector 2 ' to * sign, estimate the response vector of correctly posting a letter at the b point place of post a letter response vector and conduct * sign that a is ordered.
On the other hand, less at MSE shown in Fig. 5 B, under the less situation of the estimation error of collection of letters response vector, even because extrapolation distance is longer, and extrapolation error is also less, so elongate extrapolation distance.More particularly, under the less situation of MSE, under Fig. 5 B situation, carry out the extrapolation of longer distance of the c from collection of letters response vector 2 ' to * sign, estimate the response vector of correctly posting a letter at the d point place of post a letter response vector and conduct * sign that this c is ordered.
Described processing is mainly carried out by the response vector estimating device 32 of posting a letter among Fig. 1, and Figure 13 represents to be conceived to the extrapolation process flow chart of this MSE.
Flow chart shown in Figure 13 is except following some, and is identical with flow chart shown in Figure 6.Promptly in step S12, calculate MSE by well-known method, then in step S13, the response vector estimating device 32 of posting a letter by shown in Figure 1 utilizes the illustrated method of Fig. 5, determines the extrapolation parameter that adapts to most according to MSE, i.e. extrapolation distance.In addition, under this target, deposit in the memory 34 among Fig. 1 by the optimal extrapolation distance that prior mensuration determined according to the big wisp of MSE.
Other processing method is referring to the relevant explanation among Fig. 6, no longer repeat specification here.
As mentioned above, in this embodiment, owing to select optimal extrapolation distance according to the size of MSE, so, still can carry out the estimation of the correct response vector of posting a letter even in the collection of letters response vector of uplink estimation error is arranged.
Next, Figure 14 represents the extrapolation process flow chart of other execution mode of the present invention.Though in the execution mode of Fig. 6, determine optimal extrapolation distance by Doppler frequency FD, in the execution mode of Figure 13, determine optimal extrapolation distance by MSE, but in the execution mode of following Figure 14, considering the optimal extrapolation distance of decision on the both sides basis of Doppler frequency FD and MSE.
Promptly in this embodiment, be according to the Doppler frequency FD parameter (extrapolation distance) that temporarily determines to extrapolate basically, according to MSE the extrapolation parameter is revised then, make last decision.
Referring to Figure 14, because step S1 is identical with flow chart among S2 and Fig. 6, so no longer repeat specification.If estimate Doppler frequency FD in step S2, then Doppler frequency FD that can preserve in advance from memory 34 and the corresponding relation between the extrapolation distance are selected optimal extrapolation distance, and do temporary transient decision.
Next estimate MSE in step S12, the size according to this MSE in step S22 is revised extrapolation distance.Such as, under the bigger situation of MSE, be necessary to shorten extrapolation distance, revise with the coefficient of X<1.On the other hand, under the less situation of MSE, be necessary to prolong extrapolation distance, revise with the coefficient of X>1.These coefficients are obtained and are kept in the memory 34 according to experiment in advance.
Like this, in step S22, make the final decision of extrapolation distance, estimate the weight of posting a letter according to step S4 illustrated in fig. 6 and S5 then.
As mentioned above, in this embodiment, owing to temporarily select extrapolation distance according to the Doppler frequency FD that extrapolation error is had a significant impact, and then according to MSE extrapolation distance is revised, so can more correctly estimate the response vector of posting a letter.
In addition, because that expression transmits the corresponding relation of the Doppler frequency of environment or MSE and extrapolation distance is different with the individual difference of each wireless device framework, so generally will measure in advance and determine at each wireless device.But, under the less situation of individual difference, can adopt unified corresponding relation to a plurality of wireless devices in the whole system.
In addition, the extrapolation parameter is not limited to described extrapolation distance, for example also can be other parameter of extrapolation gradient etc.
As mentioned above, in the present invention, even the collection of letters response vector of being estimated in uplink has estimation error, also can be by the parameter of selecting the extrapolation process corresponding to use with transmitting environment, the response vector of posting a letter to downgoing line is correctly estimated, thereby can be realized good originator tropism.
And, according to the present invention, owing to be not according to received signal itself, but carry out the estimation of Doppler frequency according to the correlation between the collection of letters response vector, so can not be subjected to the influence of the disturbing factor in the received signal, and increase the adaptability that calculation is handled simultaneously, and then carry out the estimation of more accurate Doppler frequency.