Embodiment
General hypothesis Guassian Minimum Shift Keying GMSK (Gaussian minimum shift keying, GMSK) linear approximation is in Quadrature Phase Shift Keying (quadrature phase shift keying, QPSK) system, therefore we can one comprise the Channel Models of the QPSK system of a Gaussian filter as gsm system, and its corresponding channel coefficients can get via the cross-correlation (cross-correlation) of the known training string symbol of the output of calculating channel and receiver.This type of utilizes known training string symbol to find out channel impulse response has a variety of with the method for carrying out channel estimating, different modes predicted channel impulse responses such as the peak value of utilization as search cross-correlation or calculating energy respectively, it is respectively delivered on patent document or academic journal.About the principle of utilizing known training string to find out channel impulse response is to be employed training string in the digital communication system, often be designed to have the autocorrelation and the approximate incoherent white noise (uncorrelated white noise) of height, therefore after the time calibration of received symbol, peak value can appear in its cross-correlation with the training string.See also Fig. 1.Fig. 1 is the calcspar of a wireless communication system receiver.10 is a wireless communication receiver, and it includes a cross-correlator (cross-correlator) 12 and a training string symbol (training sequence symbol) 13, and
training string symbol 13 is to be electrically connected to cross-correlator 12.In Fig. 1,
Be the quadruple sampling of received signal, meaning is to be
cycle 1/4th long of the bit received of receiver or symbol in its cycle, x
kBe the quadruple sampling of training string originally, k is the index (time step index) of time step, and the length that makes L representative training string.As shown in Figure 1, we are with x
kAnd
Send into and carry out computing cross-correlation in the
cross-correlator 12, the quadruple sampling of wherein received signal
Should comprise training string quadruple sampling x
kParagraph.So, then the output of
cross-correlator 12 can be expressed as:
In the formula (1)
Be x
L-iMany conjugation.
Train the position of going here and there in the signal that is received in order to want to detect,
Length should compare x
kLong.In the known technology, have directly so that
The absolute value that is the output of
cross-correlator 12 is the starting point of the pairing received sample of signal of maximum k value as the training string in the signal of being received, and the dopester of delivery channel impulse response.Yet because the Shi Bianyu of GSM channel decline characteristics such as (fading), for different bursts (burst), the position of the absolute value peak value of cross-correlation does not also mean that the position that training is gone here and there.Therefore, directly (promptly make with the time step index of the absolute value peak value of cross-correlation
Be peaked k) as the starting point of training string, be very may error take place and the demand that can't satisfy channel equalization.
See also Fig. 2.Fig. 2 is the calcspar of a wireless communication system receiver.20 is a wireless communication receiver, it includes a
cross-correlator 22, one
training string symbol 23 is electrically connected to
cross-correlator 22, and one
interval energy calculator 26 be used for calculating the energy in a predetermined length interval with the output of
computing cross-correlation device 22, and so that the energy in this predetermined length interval is the cross correlation value of maximum zero-time step index and follow-up time step index thereof, as channel impulse response and export it.Wherein
cross-correlator 22 is identical with the function of
training string symbol 13 with the
cross-correlator 12 among Fig. 1 with
training string symbol 23;
x
k, k, L and cross-correlator output
Definition and intension, also identical with the technology of the described Fig. 1 of epimere.
Interval energy calculator 26 can be calculated the energy in the predetermined length interval, makes the interval NT of being of this predetermined length, and then the energy in the predetermined length interval is:
Formula (2)
Wherein N is the number of the coefficient of the channel impulse response of prediction, and i then is initial time step index.
In receiver as Fig. 2, promptly so that
For peaked i value φ pairing
As the starting point of the training string in the signal of being received, and the prediction of delivery channel impulse response
Energy with the predetermined length interval described above is as the technology of predicted channel impulse response, and because of calculate its energy in addition after cross-correlation, amount of calculation is more only carried out cross-correlation for big; So it has improved the problem that channel impulse response skew will take place and can't directly train the starting point of string with the absolute value peak value prediction of cross-correlation when the channel of transmission signals is fading channel (fading channel).Yet if when the channel of transmission signals is non-when being fading channel but awgn channel, calculating energy will lower the usefulness of system greatly with predicted channel impulse response.Though, in fact will train string and the absolute value peak value that the signal of being received carries out the value of computing cross-correlation gained to change because of these noises because the noise in the awgn channel can influence whole energy.Therefore if the known technology that only uses Fig. 2 and lifted comes calculating energy with predicted channel impulse response, the system that makes is increased on time of cost and amount of calculation, thereby need design than complicated hardware, pay higher cost.
See also Fig. 3.Fig. 3 is the calcspar of wireless communication system receiver of the present invention.30 is a wireless communication receiver, and it includes a
cross-correlator 32; One
training string symbol 33 is electrically connected to
cross-correlator 32; One
peak counter 37 is electrically connected to
cross-correlator 32, is used for calculating the peak value number n of
cross-correlator 32 operation results; One peak
value ratio comparator 34 is electrically connected to
peak counter 37, the peak value number n that is used for being calculated in
peak counter 37 is for for a long time, calculates the ratio A0 of square maximum of absolute value of
cross-correlator 32 operation results and second largest value and it is compared with one first predetermined value A1; One module 35 of calculating and export the value on correlation peak both sides is electrically connected to
peak counter 37 and peak
value ratio comparator 34, as A0 during greater than the first predetermined value A1 and when the peak value number is 1, with the cross correlation value of this peaked time step index and adjacent time step index thereof, as channel impulse response and export it; And one calculate and N correlation module 36 of output energy maximum, be electrically connected to peak
value ratio comparator 34, when A0 is not more than the first predetermined value A1, to calculate the energy in a predetermined length interval by the output of
computing cross-correlation device 32, so that the energy in this predetermined length interval is the cross correlation value of maximum initial time step index and follow-up time step index thereof, as channel impulse response and export it.Wherein, calculate and the function of N correlation module 36 of output energy maximum then identical with
interval energy calculator 26;
Be the sampling of received signal, in true (simulation) experiment of walking back and forth of the present invention, we find with overclocking sampling (over sample)
The usefulness of carrying out computing does not significantly strengthen, and therefore in the present embodiment, directly the bit of being received with receiver or the Cycle Length of symbol are taken a sample to received signal, and meaning promptly
Sampling frequency be bit or symbol frequency; x
kBe training string originally, k is the index (time step index) of time step, and the length that makes L representative training string; Similarly, train the position of going here and there in the signal that is received in order to want to detect,
Length should compare x
kLong and should comprise x
kParagraph.The output of
cross-correlator 32
Definition and intension, then identical with the described technology contents of Fig. 2.In of the present invention other used,
Still can be the integer multiple frequency sampling of received signal, x
kAlso relatively selling frequency sample for this integer of training string gets final product; Again
Or can be part in the integer multiple frequency sampling of received signal, for example for getting one in per two quadruples sampling, and x
kFor the relative sampling in the training string gets final product.
As previously mentioned, when signal when awgn channel transmits, the AWGN noise can influence received energy, but can not change
cross-correlator 32 operation results absolute value square the position that peak value occurred; Therefore utilize the position of this peak value to get final product predicted channel impulse response.If signal is via fading channel transmission, then computing cross-correlation result's absolute value square the position that occurs of peak value can be offset,, should determine by which cross correlation value as predicted channel impulse response with the energy in a regular length interval.Therefore use the method for the ratio of square peak-peak of computing cross-correlation result's absolute value and time big peak value relatively among the present invention, judge and export the computing cross-correlation result in which way as channel impulse response.Square ratio A0 of computing cross-correlation result's absolute value is during greater than one first predetermined value A1, that is maximum goes out greatly when many than second largest value, and our channel of decision signal institute warp is an awgn channel; This moment, the value module 35 of calculating and exporting the correlation peak both sides sent into the output of
cross-correlator 32 by peak
value ratio comparator 34, the computing cross-correlation result of square peaked time step index of the absolute value that cross-correlator 32 is exported and adjacent time step index thereof is as channel impulse response and export it.Otherwise when the ratio A0 of square maximum of computing cross-correlation result's absolute value and second largest value is not more than this first predetermined value A1, that is maximum is not when going out abundant multiple greatly than second largest value, and our channel of decision signal institute warp is a fading channel; This moment, N the correlation module module 36 of calculating and exporting the energy maximum sent into the output of
cross-correlator 32 by peak
value ratio comparator 34, N correlation module 36 calculating and export the energy maximum will be as the energy in the calculating one predetermined length interval as the
interval energy calculator 26 among Fig. 2, make the interval NT of being of this predetermined length equally, so that the energy in this predetermined length interval is the cross correlation value of maximum initial time step index and follow-up time step index thereof, as channel impulse response and export it.Wherein, rule of thumb, this first predetermined value A1 can be made as 8 with preferable, and so it also can be made as other value under varying environment and system requirements.Calculate and the value module 35 on output correlation peak both sides and the output of calculating and exporting N correlation module 36 of energy maximum
With
As follows respectively:
Formula (3)
Wherein γ is for making
Value be maximum k value, k is a time step index.
Calculate and channel impulse response that the value module 35 on output correlation peak both sides is predicted for
Be the center, the both sides, front and back are symmetry respectively.Characteristic according to the GMSK modulation, its can be approximately pulse amplitude modulation (Pulse Amplitude Modulation, PAM), and the employed training string of gsm system just is being the GMSK modulation signal, therefore its channel impulse response is bell symmetry in low noise environment, thus can with
Be the center, it is exported as channel impulse response in both sides, the front and back output of the cross-correlation of symmetry respectively.In the present embodiment, for cooperating the relevant regulations of gsm system, can
Be the center, the both sides, front and back are respectively got the output of two cross-correlation respectively, promptly
Export it as channel impulse response.The present invention in other is used, but also the relevant regulations of coupled system and need and before and after getting the output of symmetry three or other quantity cross-correlation export it as channel impulse response.
Formula (4)
Wherein φ is for making
Be peaked k value, and N is the number of the coefficient of the channel impulse response of prediction, k then is initial time step index.
Calculate and the prediction of the channel impulse response that N correlation module 36 of output energy maximum exported, identical with the output of the
interval energy calculator 26 of prior art among Fig. 2; Promptly with
As the starting point of channel impulse response, get four points (because of the output of the prediction of establishing channel impulse response in the present embodiment needs totally five points) more backward, be
In like manner, the present invention is in other is used, but the also relevant regulations of coupled system and need and get
And thereafter six or the output of other quantity cross-correlation export it as channel impulse response.
The above is a content of the present invention.Be illustrated more clearly in the flow process that the present invention carries out the channel impulse response prediction for asking, see also Fig. 4.Fig. 4 is the flow chart of the receiver predicted channel impulse response of wireless communication system of the present invention.
Step 100: beginning channel impulse response prediction;
Step 110: the computing cross-correlation that the sampling of the sampling of the training string that stores in the internal memory and received signal is carried out pre-determined number;
Step 115: calculate the number n of peak value, if n is 1 execution in step 140; If n is greater than 1 execution in step 120;
Step 120: the ratio A0 of the maximum of squared absolute value and second largest value in the computing cross-correlation of this pre-determined number in the calculation procedure 110;
Step 130: compare an A0 and a predetermined value A1; As A0 greater than A1 execution in step 140 then; Be not more than then execution in step 150 of A1 as A0;
Step 140: be judged as awgn channel, as channel impulse response and export it, execution in step 160 again with square peaked index of the absolute value of computing cross-correlation result in the step 110 and each predetermined quantity y cross correlation value of front and back thereof;
Step 150: be judged as fading channel, with the computing cross-correlation result's of gained in the step 110 absolute value square, a predetermined quantity n adds up, calculating the energy in a predetermined length interval so that the energy in this predetermined length interval for maximum initial index and afterwards 2y cross correlation value as channel impulse response and export it;
Step 160: finish the channel impulse response prediction.
Combine it, the invention provides a device and method that is applicable to through the mobile communication system predicted channel impulse response of AWGN and fading channel transmission signals, and the present invention is specially adapted to use the receiver of the global system for mobile communications of maximum possible sequence balancing technique (MLSE).In the present invention, the characteristic of awgn channel and fading channel all is included into and considers, and judges how to export cross correlation value as channel impulse response with a simple method.The invention enables when judging that signal is when awgn channel transmits, directly with the peak value predicted channel impulse response of computing cross-correlation, receiver neither need expend huge amount of calculation calculating energy, the balanced usefulness that can obtain again; And when judging that signal is when fading channel is transmitted, in the mode of computation interval energy, accurately predicted channel impulse response.The present invention is with succinct framework and small amount of calculation, utilize known channel impulse response Predicting Technique, add relatively square maximum of computing cross-correlation result's absolute value and the step of a ratio between second largest value and a predetermined value, judge that as one channel is the mechanism of awgn channel or fading channel (fading channel), comply with the result of its judgement again, select the pairing preferable channel impulse response Forecasting Methodology of this kind channel, use receiver of the present invention in awgn channel or fading channel no matter make, all accurately predicted channel impulse response reaches best channel equalization usefulness.
The above only is preferred embodiment of the present invention, and all equalizations of being done according to the present patent application claim change and modify, and all should belong to the covering scope of patent of the present invention.