TWI308443B - Channel estimation enhanced lms equalizer - Google Patents

Description

1308443 δ(ΤΤ2Γ^- 年月日日修正 replacement page IX, invention description: ~ - [Technical Field] The present invention relates to a wireless communication system. More particularly, the present invention relates to a channel evaluation enhanced LMS equalizer. One of the five-week adaptive filter filter coefficients is the LMS (Least Mean Square) algorithm. In the LMS filter, the filter coefficients are based on the error between the LMS filter and the reference value. The error is fed back to update the filter coefficients, and the updated filter coefficients are generated based on the step size and the error, which are repeatedly updated until convergence. Because the convergence speed cannot catch up with changing the channel quickly, With a small step size, the LMS equalizer (or normal LMS (NLMS) equalizer) in the fast change channel is degraded. Using a large step size increases the convergence speed and thus enhances the LMS equalizer performance. The use of large steps may result in excessive error adjustment errors. Therefore, there is a permutation relationship between the tracking ability and the error adjustment error. The large step tracking channel is better. However, the small step is long. Error adjustment error preferred. Thus, the step size is set to optimize the overall efficiency Shaw b 'but the shoulder nasal embodiment LMS method often produces suboptimal convergence time.

The Griffith algorithm is based on the LMS algorithm adaptation that does not require an error signal but requires a priori knowledge of the signal and data vector product prediction values. Therefore, it is predicted that channel estimation can be performed without limiting the prior art. SUMMARY OF THE INVENTION The present invention relates to the use of one of the channel evaluation enhancement intensifiers. The scaled version of the 'channel evaluation according to the present invention is taken as the predicted average behavior for the product of the transmitted signal and the received signal used to implement the Griffith calculus 1308443 method. The present invention also uses advance or reward channel evaluation to overcome the extension of the LMS algorithm secrets that exist in changing channels over time. Thus, the present invention can facilitate the use of small step sizes to achieve the same tracking capability in large steps. The channel evaluation at a certain point in the future is used to update the equalization denoising and decimation point coefficient. This is implemented by prediction and wave. Can be #代是, because of the riding filter tap point system

The number generator input data is delayed, so the delay can be introduced into the input data to the filter tap to generate H, which allows the channel to evaluate the prediction of the contact wave coefficient generator. "° Detailed Description of the Preferred Embodiments] The characteristics of the present invention can be incorporated into an integrated circuit (1C) or configured in a circuit including a plurality of interconnected components. The present invention provides a dimensionally good (10) hybrid Good tracking of high mobility channels - equalization | | (New is the New Zealand. The Griffflth algorithm is designed to allow error-free recording (axis antenna _ to reject the pulse =), but Lai Zhi is known by Wei domain The county accumulates the _(10) algorithm. Usually, the receiver does not know that this is pre-determined. According to the invention, the behavior is evaluated, and the evaluation is used in the fresh method. The channel evaluation unit can easily obtain the channel when the dumping signal is received and the received prediction of the county is known to lead the transmission into the transmission (for example, by associating the received signal with the receiving tiger). Evaluation. 2 Ming also uses pre- or pre-evaluation to overcome the delay in the lms algorithm variation in the channel of change, which leads to the 7 1308443 ^ 12 3 0 - year and day correction replacement page small The step size makes the rider the same as the big step. According to the invention, the frequency of the future point The evaluation system is used for the more normalizer filter tapping point, which can be implemented by pre-chopping II. Alternatively, since the input data of the chopping sprite coefficient generator is delayed, the delay can be Introduce the input data to the ship tap-off generator, which makes the channel evaluation similar to the prediction of the filter tap coefficient generator. According to the leakage 9NLMS method, the (10) green filter tap coefficient can be weighted. Write as follows:

<0kxk)H Equation (1) where the error signal 4 = (1 + force - shape, a is the leaking Lang, w is the equalizer chopper tap coefficient - especially in the equalizer filter The human data vector, y is the equalized thief wave output, · ν = is, e is garbled, and the subscript k means the kth repetition.

Note the product yC=eq_deSCram, and make the heart @ and the pilot signal P={l+j} 'The equation (1) can be rewritten as follows: ^ = cmk^ + fi[p{ckXk)H -ykCt{Ckxk^ \ Equation (2)

Sym — vec Equation (2) can be rewritten as follows: wk = aw^ + β(ρ. sym^vec11 - eq_descram sym vecHΊ – Equation (3) The NLMS algorithm enhancement according to the invention is replaced by the following expected equation ( 3) The left term of () is reached: wk = ocwk^ + β{Ε{ρ - sym_vecH} — eq descram sym

Equation (4) 1308443

__ is evaluated from the channel evaluation. As long as the citation is communicated, the prediction will be of interest, and the channel evaluation can replace the equation:; = the channel impulse response in the example. Replace the forecast with a = not only calculated channel estimate. If the frequency of the traitor is replaced by the day-to-day channel status assessment, additional performance improvements can be applied. This compensates for the existing delay in the NLMS algorithm. As described above, this prediction can be implemented by the delay in front of the disintegration.

Figure 1 is a block diagram of an equalizer i (8) in accordance with the present invention. The equalizer (10) includes an equalizer ship (10), a channel evaluator 112, a ferrite tap coefficient generator n4 and a multiplier 110. The equalizer just selects the step-by-step to include a delay buffer, and the signal combiner moves to the downsampler 108. The digitized samples 132, 134 are fed into the signal combiner 1〇2. The present invention can be extended to use multiple antennas to receive diversity. As many samples of the sample 132, 134 can be generated from the received signal via multiple antennas, the multi-sample stream 132' 134 is multiplexed by the signal combiner to produce a combined sample stream 136. It should be noted that Figure 1 depicts two sample streams 132, 134 (not shown) from two receive antennas, but only one or more sample flow depends on the antenna configuration. If only the same stream is generated, the signal combiner 102 is not needed and the sample stream is fed directly into the delay buffer 〇4 and the channel evaluator 112. Signal combiner 1 可 2 can alternatively interpolate samples 132, 134 to produce the same stream 136. The combined samples 136 are fed into the delay buffer 1〇4 and the channel evaluator 112. The delay buffer 104 may store the combined samples 130 for a predetermined time interval before the output is delayed by the combined samples 139 to 1308443. This allows the channel evaluation to be similar to the prediction of the filter and filter tap coefficient generator. Alternatively, sample 136 can be fed directly to the equalizer filter. The equalizer chopper 1〇6 can process the delayed combined sample 138 and output the filtered sample 14〇 using the filter & tap point 148 updated by the filter tap coefficient generator 1Η. If the sampling rate is greater than the wafer rate or a multi-sample stream is generated, then filtered, 140 can be downsampled by the downsampler (10), whereby the downsampler 108 generates the wafer rate data. Preferably, samples 132, 134 are produced at twice the wafer rate. For example, if the two sample streams are generated at twice the wafer rate' then the downsampler (10) downsamples the tear sample 140 by a factor of four (4). " The heart is downsampled by the sample 142 and then multiplied by the multiplier, 11 乘 the downsampled sample 142 by the garbled 157. The descrambled filtered samples 144 are output from the equalization _ 1 藉 by other downstream components _ S ' and are also fed back to the chopper tap coefficient generator ι 4 . ▲ As the input channel evaluation n 112 can receive the combined sample 136 and the car 引 引 引 i 52 and output a channel evaluation i5 〇. Channel evaluation can be generated by any prior art method. When the pilot signal is included in the received heart rate, the signal-free knowledge system can enhance the channel assessment. ~, the tap coefficient generation 11 114 can be generated by the equalization of the gift wave - the filter is the filter tap coefficient ^ of the combined sample 138 in 106. The filter tap coefficient generator 114 acts as an input, is unblocked by the Shimaji wave sample 144, taps the sample state vector in the delay line 146, and the channel evaluation 10 1308443 generates a channel evaluation 150, a one-step parameter μ154 And a leakage parameter α 156. Figure 2 is a detailed block diagram of the equalizer filter ι〇6. The equalizer filter 106 includes a tap delay line 2〇2 and an intra-vector product multiplier 204. The delayed combination sample 138 is shifted into the tap delay line 2〇2, and the vector multiplication multiplier 204 calculates the vector of the sample state inward 146 and the composite filter and the filter tap coefficient Mg that are shifted into the tap delay line. Internal product. The intra-vector product is output from the equalizer filter 106 as a filtered sample 140. Figure 3 is a detailed block diagram of the filter tap coefficient generator 114. The filter tap coefficient generator 114 includes a first co-fortunate unit 302, a second conjugate unit 3〇4, a first multiplier 3〇6, an adder 308, and a second multiplication. The device 310, a vector norm square generator 32A, a divider 314 and a loop filter 311. The channel estimate 150 generated by the channel evaluator 112 is fed to the first conjugate unit 3〇2 to generate a common vehicle for the channel evaluation 332. The sample state vector 146 of the taper delay line 2 of the equalizer filter 1〇6 is fed to the second conjugate unit 3〇4 to generate a common vehicle of the state vector 334. The state vector 334 and the scrambled chop sample 144 are multiplied by the first-multiplier 3〇6. The first-multiplier is implemented as a scalar vector multiplier 3G5 that produces a vector signal. The output of the first-multiplier 挪 is taken from the conjugate of the channel evaluation 332 by the adder 308 to generate an uncorrected correction 338 corresponding to (E{p. sym-vecH) in equation (4). The eq-descram.sym-vecH) term state vector 146 is also fed to the vector norm square generator 32 to calculate the vector norm square of the state vector 340 with 1308443. The step size parameter is obtained by dividing the vector norm square of the state vector 34 by the divider 314 to generate a scaling factor 342 (that is, the magic in equation (4). The scaling factor is obtained by the second multiplier 310 is multiplied by the uncalibrated correction item 338 to generate the calibration correction item 3. The calibration correction item 344 is fed to the loop filter '3' to be added to the previous repeat chopper tap coefficient to produce an update. The filter tap coefficient 148. The loop filter 311 includes an adder 312, a delay unit and a multiplier 316. The chopper tap coefficient 148 is stored in the delay single training and is used for the lower- Repeat as the previous chopper tap coefficient. The delayed filter tap coefficient 346 is multiplied by the leakage parameter α 156 to generate the calibration first wave age contact 348, and the slave first _ waver The contact coefficient 348 is added with a scaling correction term 344 by the adder 312 to generate a filter tap coefficient 148. The performance improvement according to the present invention is shown in Fig. 4 as a simulation result. The simulation is for 10 dB. Signal Noise Ratio (SIR), Quadrature Phase Shift Keying (QPSK) Xia, Cai VA12G Ship - Receive Antenna, and Third Generation High Speed Downlink Packet Access (HSDPA) Fixed Reference Channel (FRC) test is configured. This simulation shows the advantages of the high mobility channel (12 〇 kph action speed) of the present invention. Figure 5 is a flow chart showing the process of equalizing the received signal according to the present invention. The sample is generated from the received signal (step 502). The sample is transferred to the equalizer filter. It has been temporarily stored in the delay buffer to delay the sample-segment predetermined period (step 504). The channel evaluation is generated based on the sample 12 1308443 97. 12.-30=-year month correction replacement page (step 506) The sample delayed by the buffer delay is processed by the 滤波 益 滤波 filter to generate the escaping sample (step garbled yoke yoke is multiplied by the filtered sample to generate a garbled code filtered sample (step 51 〇 The new filter tap coefficient is generated using the channel evaluation (step 512). Figure 6 is a flow chart for generating filter tap coefficient processing according to the present invention. The conjugate system for channel evaluation is generated. (step 602). The sample state vector conjugate in the tapped delay line of the equalizer filter is multiplied by the descrambled filtered sample (step 604). The multiplication result is subtracted from the conjugate of the channel estimate to produce an uncorrected correction term ( Step 606) The scaling factor is generated by dividing the step size by the vector norm square of the sample state vector moved into the tap delay line (step 608). The scaling vector is multiplied by the uncalibrated correction term to A scaling correction term is generated (step 610). The scaling correction term is added to the previously repeated filter tap coefficient to produce updated filter tap coefficient (step 612). The features and elements of the present invention are described in the preferred embodiments in a particular combination 'but the various features and elements do not require other features and elements of the preferred embodiments' or with or without other features and combinations of elements of the invention. Used separately. 13

"V 1308443 - [Simplified illustration of the drawings] Fig. 1 is a block diagram of an equalizer according to the present invention. Figure 2 is a block diagram of the equalizer filter of Figure 1. Figure 3 is a block diagram of the filter tap coefficient generator of Figure 1. Figure 4 shows the results of performance improvement simulations compared to prior art NLMS equalizers. Figure 5 is a flow chart showing the process of equalizing the received signal in accordance with the present invention. 0 Figure 6 is a flow chart showing the processing of the filter tap point coefficients in accordance with the present invention. [Major component symbol description] 110, 204, 306, 310, 316 multiplier 106 equalizer filter 136, 138 combined sample 142 downsampled sample 146 sample state vector 148 filter tap coefficient φ 150 channel evaluation w, etc. 154 step parameter μ 156 leakage parameter 〇 c 308, 312 adder 311 loop filter 340 divider 342 calibration factor QPSK quadrature phase shift keying LMS (least mean square) algorithm 14

Claims (1)

1308443 197....12. 3 Q-- - Year Month Day Correction Replacement Page I . . _ X. Patent Application Scope: 1. An equalizer containing: a channel evaluator 'for generating a channel evaluation; a filter, the sample is processed by a filter tap coefficient to generate a filtered sample; a delay buffer delaying the sample for a predetermined time interval before the sample is transferred to the equalizer filter Φ; And a filter tap coefficient generator that can be used to update the filter tap coefficient. 2. If the towel meets the equivalent of H in the scope of the patent, wherein the equalizer tear waver comprises: u 3. 4. A tap delay line, the sample is sequentially moved into the tap delay line; A one-way multiplication product multiplier is used to calculate a vector product of the sample shifted into the tap delay line and the filter tap coefficient. For example, in the equalizer of claim 2, wherein the occupational wave point tap coefficient coefficient generator is based on the sample sampled by the descrambled filter, the sample state vector moved into the sub-k late line, and the channel evaluation. A new set of filter tap coefficients. For example, the equalizer of claim 3, wherein the occupational wave point tap coefficient generator comprises: a total scale element 'generating the channel-evaluated-common vehicle; the first conjugate unit is generated and moved into the point One of the sample state vectors connected to the delay line is light; a first multiplier multiplies the descrambled filtered sample by the state 15 • 1308443 ψ^.ΐ2Λ(}} 月 日 修正 替换 替换 替换 替换An adder deducting an output of the first multiplier from the conjugate of the channel evaluation to generate an uncorrected correction term; a vector norm square generator, calculating a square of the vector number of the state vector; The step size parameter is divided by the vector norm squared to generate a certain chirp factor; the second multiplier 'multiplies the scaling factor by the uncalibrated correction term to generate a certain target correction term; and the thin path filter 'borrows Adding the calibration correction term to a previous repetition = the data wave H tap coefficient to store the previously repeated the waveguide tap coefficient ' and outputting the new detector tap coefficient. 5. If the patent application scope Item 4 An equalizer, wherein the ferrite tap coefficient generator further comprises a third multiplier that multiplies the leak parameter by the previously repeated ferrite tap coefficient. The equalizer of the range i, further comprising - down sampling theft to downsample the filtered sample, whereby the downsampler outputs the post-ship sample at a wafer rate. The equalizer of the sixth item of the range further includes a signal combiner for combining the plurality of sample streams generated based on the signals received by the plurality of antennas. 8. The equalizer of claim 7, wherein The signal combiner can interpolate the complex sample stream to produce a combined sample stream. 9' - a method of equalizing the line of operations, the method comprising: 16 1308443 a positive replacement page produces a sample of the received signal; based on the sample Generating a channel evaluation; before transferring the sample-equalizer vessel, the _sample is stored in a delay buffer to delay the sample for a predetermined time interval; processing the recorded book with the equalized nm to produce a 'wave post sample; Take The channel is used to evaluate the new filter tap coefficient of the equalizer filter. H) The method of claim 9, wherein the new filter tap coefficient is after being descrambled (four) wave The sample, the shifting of the equalizer tear wave - the tap delay line - the sample state vector and the channel rating are generated. 11. The method of claim 9, further comprising down sampling the filtered sample to produce a Weibo post sample at a u rate. • 12. t. The method of claim 11 of the patent specification, the further step comprising combining the complex sample streams generated based on the signals received via the complex antennas. 13. The method of claim 12, wherein the plurality of sample streams are interpolated to produce a combined sample stream. 14. An integrated circuit for equalization, comprising: a channel estimator for generating a channel estimate; a equalizer filter for processing samples with filter tap coefficients to generate filtered samples; a delay buffer that delays the sample for a predetermined time interval before the sample is transferred to the equalizer filter; and 17 • 1308443 — • Year Month 曰 Correction Replacement Page - Waves ϋ 分 分 分 分 分The road evaluation filter tap coefficient. μ 15. The integrated circuit of claim 14 of the patent scope, wherein the wave device comprises: a tapping extension 'the sample system 依 sequentially shifting the tap delay line; and the φ-vector product multiplier, It is used to calculate the -vector inner product of the ° Hai sample and the 3 Hai data filter tap coefficient that are moved into the tap delay line. 16. The integrated circuit of claim 15 wherein the chopper tap coefficient generator is filtered by the descrambled sample, the sample state vector moved into the tap delay line, and the The channel evaluation is based on a set of new filter tap coefficients. 17. The integrated circuit of claim 16 wherein the ferrocouple tap coefficient generator comprises: • a first conjugate unit for generating a conjugate of the channel estimate; a second conjugate a unit for generating a conjugate of the state vector; a first multiplier 'to multiply the descrambled filtered sample by a conjugate of the state vector; an adder 'to evaluate the conjugate from the channel Deducting the output of the first multiplier to generate an uncorrected correction term; a vanation norm square generator for calculating a vector norm square of the state vector; a divider for dividing the step length parameter by The vector norm is squared to produce a certain scaling factor; 18 -1308443 pull -1忿· 3-g~- - year, month, and day correction replacement page - a second multiplier for multiplying the scaling factor by the uncalibrated Correcting the term 'to generate a standard correction term; and a loop filter for storing the previously repeated filter tap point by adding the calibration correction term to a previously repeated filter tap point coefficient Coefficient and output new filter Sectional point coefficients. 18. The integrated circuit of claim 17, wherein the filter coupling coefficient generator further comprises a third multiplier for multiplying a leakage parameter by the previously repeated filter segment. Contact coefficient. 19. The integrated circuit of claim 14, further comprising a downsampler' to downsample the filtered sample, whereby the downsampler can output the filtered sample at a wafer rate. 2. The integrated circuit of claim 19, further comprising a signal combiner' for combining the plurality of sample streams generated based on signals received by the plurality of antennas. • The integrated circuit of claim 20, wherein the signal combiner interpolates the complex sample stream to produce a combined sample stream. 22--A filter-filter tap coefficient generator that generates a filter tap coefficient for an adaptive filter, the filter tap coefficient generator comprising: - a - conjugate unit, Conjugating one of the channel evaluations; a second conjugate unit' is used to generate one of the sample state vectors of the tapped delay line that is shifted into the adaptive filter; a first-multiplier is used to be descrambled The filtered sample is multiplied by the conjugate of the state vector; 19 -1308443 Corrects the replacement page. An adder for deducting the rounding of the first multiplier from the conjugate of the channel evaluation to generate an uncalibrated correction term; a vector norm square generator for calculating a vector norm square of the state vector a divider for dividing the step size parameter by the square of the vector norm to generate a certain scaling factor; a second multiplier for multiplying the scaling factor by the uncalibrated correction term to generate a scaling correction term; a loop filter for adding the scaling correction term to a previously repeated filtering The tap coefficient is stored to store the previously repeated filter tap coefficient and output a new filter tap coefficient. 23. A method of generating filter tap coefficient for an adaptive filter', the method comprising: generating a conjugate of a channel estimate; Multiplying the conjugate of the sample state vector shifted into an adaptive filter tapped delay line by the descrambled filtered sample; deducting the multiplicative result from the conjugate of the channel evaluation to produce an uncorrected correction term; The parameter is divided by the square of the vector norm of the vector state of the vector that is moved into the tap delay line to generate a certain scaling factor; multiplying the scaling factor by the uncalibrated correction term to generate a certain target correction term; and adding the The calibration correction term is scaled to a previously repeated filter tap coefficient to produce an updated filter tap coefficient. 20