Method and apparatus for deriving a channel impulse response estimate for a wireless channel

Related technology

This application is the United States entitled "LMMSE Based Methods for Improvement of Channel Impulse Response Estimates in a WCDMA Downlink" filed on November 15, 2005, which has been transferred to the assignee of the present application and is referenced herein. Claim priority under provisional application No. 60/737454.

The present disclosure relates to communications in general, and more specifically to techniques for deriving channel estimates for radio channels.

In wireless communication systems, transmitters typically process traffic data (eg, encode, interleaves, and symbol maps) to generate data symbols, which are modulation symbols for the data. .. In a coherent system, the transmitter multiplexes the pilot symbol with a data symbol, processes the multiplexed data and pilot signal to generate a radio frequency (RF) signal, and transmits this RF signal over the radio channel. .. The radio channel distorts the transmitted RF signal by the channel response and further degrades the signal with noise and interference.

The receiver receives the transmitted RF signal and also processes the received RF signal to obtain multiple input samples. In coherent data detection, the receiver estimates the response of the radio channel based on the pilot received and guides the channel estimation. The receiver then uses channel estimation to perform data detection (eg, equalization) on the plurality of input samples in order to obtain a symbol estimation, which is an estimation of the data symbol transmitted by the transmitter. The receiver then processes the symbol estimation (eg, demodulates, deinterleaves, and decodes) to obtain the decoded data.

The quality of the channel estimation can have a significant impact on the detection performance of the data, and can affect the quality of the symbol estimation as well as the reliability of the decoded data. Therefore, techniques for deriving high quality channel estimation in wireless communication systems are sought in the art.

[Outline of Invention] Techniques for deriving channel impulse response estimates (CIRE) with improved quality are described herein. CIRE is an estimate of the time domain response of a communication channel. CIRE can be given by a set of channel taps (taps).

In one aspect, CIRE is derived by scaling multiple channel taps, for example, based on a point-wise linear least squares average error (LMMSE) technique. Early CIRE is derived in the time domain based on the received pilot, for example, by despreading the input sample using a pilot sequence. Early CIRE can be filtered to get the filtered CIRE. The first CIRE with multiple channel taps is derived based on the initial CIRE or the filtered CIRE. The channel tap in the first CIRE is scaling to get the second CIRE. Scaled by factor). LMMSE scaling with respect to points estimates the energy of each channel tap in the first CIRE. The noise energy for each channel tap in the first CIRE is also estimated, for example, based on the energy of a few channel taps at the left and / or right ends of the first CIRE. Each channel tap in the first CIRE is then scaled based on a scaling factor determined by the energy and noise energy of that channel tap. Threshold T_hEach channel tap with lower energy can be set to zero. Threshold T_hCan be set based on the noise energy or a predetermined value.

In another aspect, the second CIRE is obtained by zeroing the selected channel taps in the first CIRE. In one embodiment, the first subset of channel taps on the left edge of the first CIRE has a first threshold of aggregate energy for these channel taps._leftIf lower, it is set to zero. The second subset of channel taps on the far right of the first CIRE also has a second threshold of total energy for these channel taps._rightIf lower, it is set to zero. The first and second thresholds are the total energy E of the channel taps in the first CIRE._totalCan be set based on. In other embodiments, the threshold T_accAt least one channel tap with less total energy is set to zero. Threshold T_accIs E_totalCan be set based on. The channel taps to be zero can be selected in a predetermined order, alternating between the left and right edges of the first CIRE and traversing from both ends to the center. Instead, the channel taps in the first CIRE can be ranked from the weakest to the strongest, and the channel taps that should be zeroed out are their weakest. It can be selected in a sequential order starting from the channel tap. In yet another aspect, the threshold T_hEach channel tap in the first CIRE with less energy is set to zero. Threshold T_hCan be set based on the total energy, peak energy, or noise energy of the channel taps in the first CIRE.

Hereinafter, various viewpoints and embodiments of the present invention will be described in more detail.

[Detailed description] The features and properties of the present invention will become clearer from the detailed description described below when read in conjunction with the drawings. Similar reference symbols are correspondingly identified throughout the drawing. The term "exemplary" is used herein to mean "useful as an example, as an instance, as an illustration". Any embodiment or design referred to herein as "exemplary" should not necessarily be construed as favorable or advantageous over other embodiments or designs.

FIG. 1 shows an exemplary transmission in a wireless communication system. For simplicity, FIG. 1 shows only one base station 110 and one radio device 120. A base station is generally a fixed station that communicates with a radio, and may also be referred to by node B, an access point, or some other terminology. The radio may be fixed or mobile and may also be referred to by user equipment (UE), mobile station, user terminal, subscriber equipment, or some other terminology. The wireless device may be a cellular telephone, a personal digital assistant (PDA), a wireless modem card, or any other device or device.

The base station 110 transmits an RF signal to the radio device 120. This RF signal can reach radio device 120 via one or more signal paths, which may include direct and / or reflected paths. The reflected path is generated by the reflection of radio waves by obstacles (eg, buildings, trees, vehicles, and other structures) in the radio environment. The radio device 120 can receive multiple instances or copies of the transmitted RF signal. Each received signal instance is acquired via a different signal path and also has a particular time delay and a particular complex gain determined by that signal path. The RF signal received on the radio device 120 is a superposition of all signal instances received on the radio device 120. The radio device 120 may also receive interfering transmissions from other transmitting stations. Interfering transmissions are shown by the dashed lines in Figure 1.

The channel estimation techniques described herein include code division multiple access (CDMA) systems, time division multiple access (TDMA) systems, frequency division multiple access (FDMA) systems, orthogonal frequency division multiple access (OFDMA) systems, and single carriers. It can be used for various communication systems such as single-carrier) FDMA (SC-FDMA) systems and the like. CDMA systems can implement one or more radio access technologies (RATs), such as wideband CDMA (W-CDMA), cdma2000, and the like. cdma2000 covers IS-2000, IS-856, and IS-95 standards. The TDMA system can implement RAT such as the Global System for Mobile Communications (GSM) for mobile communications. These various RATs and standards are known in the art. W-CDMA and GSM are "3rd Generation Partnership Projects (3rd Generation)" It is described in a document from an organization named "Partnership Project" (3GPP). cdma2000 is described in a document from an organization named "3rd Generation Partnership Project 2" (3GPP2). The 3GPP and 3GPP2 documents are publicly available. The OFDMA system uses OFDM to transmit modulation symbols in the frequency domain on orthogonal frequency subcarriers. SC-FDMA systems transmit modulation symbols in the time domain on orthogonal frequency subcarriers.

The channel estimation techniques described herein can be used for radio devices as well as base stations. For clarity, these techniques are described below for radio devices in CDMA systems, which may be W-CDMA systems or cdma2000 systems. Some parts of the description relate to W-CDMA systems.

FIG. 2 shows a block diagram of the base station 110 and the wireless device 120. At base station 110, the transmit (TX) data processor 210 receives traffic data for the radio device being serviced and processes the traffic data (eg, encodes, interleaves, and symbolically maps). Generate data symbols. As used herein, the data symbol is the modulation symbol for the data, the pilot symbol is the modulation symbol for the pilot, and the modulation symbol is the signal constellation (eg, for M-PSK, M-QAM, etc.). It is a complex value with respect to a point in constellation), and the pilot is data known a priori in both the base station and the radio. The CDMA modulator 220 processes data and pilot symbols as described below and also provides the transmitter (TMTR) 230 with output chips. Transmitter 230 processes its output chip (eg, converts it to analog, amplifies it, filters it, and upconverts frequencies) to generate an RF signal, which is transmitted from antenna 232.

At the radio device 120, the antenna 252 receives the RF signal transmitted over the direct and / or reflected path and also provides the received RF signal to the receiver (RCVR) 254. The receiver 254 processes the received RF signal (eg, filters, amplifies, downconverts, and digitizes) and obtains the received sample. Receiver 254 can also perform pre-processing on the received sample, providing multiple input samples to the equalizer / Rake receiver 260. Preprocessing can include, for example, automatic gain control (AGC), frequency correction, digital filtering, sample rate conversion, and the like. The equalizer / rake receiver 260 processes the plurality of input samples (eg, using an equalizer or rake receiver) to provide a plurality of output samples. The CDMA demodulator (Demod) 270 processes the plurality of output samples in a manner complementary to the processing by the CDMA modulator 220 and provides symbol estimation, which are sent by base station 110 to radio device 120. Estimating the data symbol. The rake receiver and CDMA demodulator may also be combined. The receive (RX) data processor 280 processes symbol estimation (eg, symbol demaps, deinterleaves, and decodes) and provides the decoded data. In general, the processing by the CDMA demodulator 270 and the RX data processor 280 is complementary to the processing by the CDMA modulator 220 and the TX data processor 210 at base station 110, respectively.

Controllers / processors 240 and 290 direct the operation of various processors in base station 110 and radio device 120, respectively. The memories 242 and 292 store the data and the program code of the base station 110 and the radio device 120, respectively.

FIG. 3 shows a block diagram of the CDMA modulator 220 at the base station 110. For clarity, the following description relates to W-CDMA. The CDMA modulator 220 includes a physical channel processor 310 for each physical channel used for traffic data, and a pilot channel processor 320 for pilots. Within processor 310 for physical channel m, the spreader 312 has an orthogonal variable spreading factor (OVSF) sign o for physical channel m._mData symbols are spread using (n) to provide multiple data chips. The scrambler 314 is a scrambled sequence s for base station 110._pMultiply the data chip by (n). Multiplier 316 gain g_mScale the output of scrambler 314 with, and output chip x for physical channel m_mProvide (n).

Within the pilot channel processor 320, the diffuser 322 has an OVSF code for pilots that is a sequence of all zeros o_pUse (n) to spread the pilot symbol and provide multiple pilot chips. The scrambler 324 is the scrambled sequence s._pMultiply the plurality of pilot chips by (n). Multiplier 326 gain g_pScales the output of the scrambler 324 with to provide multiple output chips p (n) for the pilot channel. The adder 330 adds a plurality of output chips for all physical channels and provides a plurality of output chips z (n) for the base station 110.

In radio 120, multiple input samples from receiver 254 are represented as:<maths num="1"><img file="JP5113072B2_D0001.tif" /></maths>

However, x_m(n) is an important signal component for the wireless device 120 and p (n) is a pilot from base station 110, h (n) is the impulse response of the radio channel between the base station 110 and the radio device 120. w (n) is x_mThe total noise and interference observed by (n) and p (n), y (n) is a plurality of input samples in the radio device 120, and is also<maths num="2"><img file="JP5113072B2_D0002.tif" /></maths>

Indicates convolution.

Multiple input samples y (n) are the desired signal x_mIt can be processed with an equalizer to obtain the estimate of (n).

FIG. 4 shows a block diagram of the equalizer 260a, which is one embodiment of the equalizer / lake receiver 260 of FIG. In this embodiment, multiple input samples y (n) from receiver 254 are provided to the channel estimator 410 and the finite impulse response (FIR) filter 430. The channel estimator 410 provides channel impulse response estimation (CIRE) h for the radio channel between the base station 110 and the radio device 120._{o o}Lead (n). Computer 420 is CIRE h_{o o}Takes (n) and receives, for example, LMMSE, least mean squares (LMS), recursive least squares (RLS), direct matrix inversion (DMI), zero four. Using zero-forcing (ZF), or some other technique, this CIRE h_{o o}A plurality of equalizer coefficients c (n) are derived based on (n). The FIR filter 430 filters the plurality of input samples y (n) using the plurality of equalizer coefficients c (n), and a plurality of desired output chips x._mMultiple output samples that are estimates of (n)<maths num="3"><img file="JP5113072B2_D0003.tif" /></maths>

I will provide a.

The channel estimator 410 can guide the initial CIRE based on the pilot received from the base station 110. In one embodiment, its early CIRE can be derived as follows: For n = 1, ..., L,<maths num="4"><img file="JP5113072B2_D0004.tif" /></maths>

However,<maths num="5"><img file="JP5113072B2_D0005.tif" /></maths>

Is an initial estimate of the actual channel tap h (n) for the index n, K is the cumulative length, "^*"Indicates a complex conjugate.

In equation (2), the channel tap h (n) at index n is estimated by performing backdiffusion in the time domain using a pilot sequence. Inverse diffusion is a pilot chip p that combines multiple input samples y (n) with multiple complex conjugates.^*It can be achieved by multiplying by (n) and accumulating the result over K chips. K is an integral multiple of the length of the Walsh-Hadamard code used for the pilot. The OVSF code for pilots in W-CDMA has a length of 256 chips and the Walsh code for pilots in cdma2000 has a length of 128 chips. K can be set to one pilot symbol, multiple pilot symbols, one slot, multiple slots, and so on. One slot contains 2560 chips for W-CDMA and 768 chips for cdma2000.

FIG. 5 shows an exemplary early CIRE for the radio channel between base station 110 and radio device 120. Early CIRE included L channel taps. Each channel tap has a specific complex gain and a specific time delay, both of which are determined by the radio environment. Each channel tap is associated with a different signal path or multipath. The difference between the earliest channel tap and the slowest channel tap is called channel delay diffusion. "Long" channels with reflections from distant objects have long delayed diffusion. L can be chosen to be greater than or equal to the longest predicted delay spread for the system. In this case, the actual delayed spread can be much shorter than L in many cases. In addition, many channel taps are all or mostly noise and may contain little or no beneficial energy.

Early CIRE includes estimation error and noise. For a pure noise tap with h (n) = 0, its corresponding initial channel tap<maths num="6"><img file="JP5113072B2_D0006.tif" /></maths>

Is I_{o o}/ (K E_p) Can be shown to have noise energy, where I_{o o}Is the total received energy in radio device 120, E_pIs the energy per chip for the pilot transmitted by base station 110. Furthermore, it can be shown that weak channel taps are more noisy than stronger channel taps. Various techniques can be used to improve the initial CIRE. These techniques can be applied individually or in combination.

Early CIRE can be filtered across multiple update interpals to reduce noise. In one embodiment, filtering is performed separately for each channel tap based on a 1-tap infinite impulse response (IIR) filter, as follows: For n = 1, ..., L,<maths num="7"><img file="JP5113072B2_D0007.tif" /></maths>

However,<maths num="8"><img file="JP5113072B2_D0008.tif" /></maths>

Is a filtered channel tap at index n with respect to the update interval t, α_hIs the IIR filter coefficient that determines the amount of filtering.

Generally, 1&gt; α_h&gt; 0, greater than α_hCorresponds to less filtering and vice versa. For example, α_hCan be set in the range of 0.15 to 0.5. α_hCan be a fixed value or a configurable value and can be selected to provide good performance. Noise is reduced by filtering in equation (3). The filtering can also be performed in other ways and with other types of filters. For simplicity, the indicator t has been removed in the description below.<maths num="9"><img file="JP5113072B2_D0009.tif" /></maths>

Indicates a filtered channel tap for the current update interval.

In one embodiment, the initial channel tap<maths num="10"><img file="JP5113072B2_D0010.tif" /></maths>

Is the input channel tap for the first CIRE h_iAs (n), i.e.<maths num="11"><img file="JP5113072B2_D0011.tif" /></maths>

Output channel tap on the second CIRE used as and with improved quality h_{o o}(n) is derived. In other embodiments, filtered channel taps<maths num="12"><img file="JP5113072B2_D0012.tif" /></maths>

Is an input channel tap h_iAs (n), i.e.<maths num="13"><img file="JP5113072B2_D0013.tif" /></maths>

Used as an output channel tap h_{o o}(n) is derived. Output channel tap h for both embodiments_{o o}(n) may or may not be filtered. Output channel taps can be derived based on various post-processing schemes.

In one post-processing scheme, the output channel taps in the second CIRE are obtained by scaling the input channel taps in the first CIRE. Input channel tap h_i(n) can be expressed as: For n = 1, ..., L,<maths num="14"><img file="JP5113072B2_D0014.tif" /></maths>

However, w_i(n) is the input channel tap h_iEstimated error and noise (or simply noise) with respect to (n). Noise w for simplicity_i(n) is<maths num="15"><img file="JP5113072B2_D0015.tif" /></maths>

Can be assumed to be the zero mean with the variance of. Noise dispersion is L input channel taps h_iSimilar for all of (n) and therefore not a function of tap index n<maths num="16"><img file="JP5113072B2_D0016.tif" /></maths>

It can be shown that it can be represented by. This noise characteristic can be used to guide output channel taps with improved quality, as shown below.

An improved estimate of the actual channel tap h (n) can be derived based on the LMMSE technique or some other technique with respect to points. LMMSE in terms of points refers to applying the LMMSE technique on a tap-by-tap basis. The LMMSE estimate for the point h (n) can be expressed as: For n = 1, ..., L,<maths num="17"><img file="JP5113072B2_D0017.tif" /></maths>

However, E {} indicates the expected value operation. E {| h (n) |²} Is the expected energy of the actual channel tap h (n), E {| h_i(n) |²} Is the input channel tap h_iIt is the expected energy of (n), and h_{o o}(n) is an output tap that is an LMMSE estimate for point h (n).

Equation (5) is a noisy estimation h_iFrom (n), we provide the "best" linear estimation of h (n), where "best" is used to mean the least squares mean error in the estimation, and therefore h._{o o}(n) is E {| h_{o o}(n) -h_i(n) |²Generates a minimum value for. Equation (5) shows that the best linear estimation of h (n) is h._iIt is shown that it is given by the scaled version of (n), where the scaling factor is E {| h._i(n) |²}and<maths num="18"><img file="JP5113072B2_D0018.tif" /></maths>

Is determined by. E {| h_i(n) |²}and<maths num="19"><img file="JP5113072B2_D0019.tif" /></maths>

The amount of can be estimated as described below.

Input channel tap h_iThe energy of (n) can be estimated in various ways. In one embodiment, the energy of the channel tap is estimated by calculating the square of the input channel tap as follows: For n = 1, ..., L,<maths num="20"><img file="JP5113072B2_D0020.tif" /></maths>

However, E (n) is the input channel tap h_iIt is the estimated energy of (n).

In other embodiments, the energy of the channel taps is estimated by calculating the square of the input channel taps at each update interval and averaging them over multiple update intervals. The averaging can be performed using a 1-tap IIR filter as follows: For n = 1, ..., L,<maths num="21"><img file="JP5113072B2_D0021.tif" /></maths>

However, E_t(n) is the input channel tap h for the update interval t_i(n) is the estimated energy α_eIs the IIR filter coefficient for the energy of the channel tap.

Coefficient α_eCan be a fixed value or a configurable value and may be selected to achieve good performance. For example, α_eMay be set to 0.5 or some other value.

The energy of the channel tap can also be estimated in other ways. In the following description, E (n) is the input channel tap h_iThe estimated energy of (n) is shown. E (n) can be derived as shown in Eq. (6) or (7), or can be derived based on some other Eq.

Noise dispersion, also known as noise energy<maths num="22"><img file="JP5113072B2_D0022.tif" /></maths>

Can be estimated in various ways. The length of the first CIRE can be chosen to be long enough so that at least one end can be expected to contain a pure or mostly noisy tap. Non-zero channel taps can be centered so that both ends of the first CIRE can be expected to contain pure or mostly noisy taps. This centering is done by (1) calculating the weight center of the channel tap in the first CIRE, and (2) placing the weight center at or near the index L / 2. obtain. The noise energy can then be estimated based on the noise taps at one or both ends of the first CIRE.

In one embodiment, the noise energy is estimated by averaging the energies of the Q leftmost channel taps of the first CIRE and the Q rightmost channel taps as follows.<maths num="23"><img file="JP5113072B2_D0023.tif" /></maths>

However, E_lIs the average energy of the Q leftmost channel taps from indicators 1 to Q, E_rIs the average energy of the Q rightmost channel taps from the indicators L-Q + 1 to L, and E_{noise noise}Is the estimated noise energy. In general, Q and L can be any suitable value. As an example, L may be equal to 64 and Q may be equal to 4.

In other embodiments, the noise energy is E, as follows:_lAnd E_rEstimated by taking the minimum value of:<maths num="24"><img file="JP5113072B2_D0024.tif" /></maths>

Where η is E_lAnd E_rA scaling factor used to explain that taking the minimum value of is statistically resulting in a smaller estimate of the true noise energy. For example, η can be set to 1.3 or some other value. This embodiment is more robust when the non-zero channel tap is not centered on the first CIRE.

In addition, in other embodiments, the noise energy E_{noise noise}Is a programmable value and is not calculated based on a noisy estimate of h (n). In the following description, E_{noise noise}Is the output channel tap h_{o o}Input channel tap used to guide (n) h_iThe estimated noise energy for (n) is shown.

In one embodiment, each input channel tap is scaled based on the LMMSE technique with respect to points. In this embodiment, the output channel tap can be derived as follows: For n = 1, ..., L,<maths num="25"><img file="JP5113072B2_D0025.tif" /></maths>

In the embodiment shown in equation (12), each input channel tap h_i(n) is the energy E (n) and noise energy E of the channel tap._{noise noise}Scales based on the scaling factor determined by.

In other embodiments, each input channel tap is scaled based on the LMMSE technique with respect to points if the tap is strong enough, otherwise zero. The energy of the input channel tap is the threshold T_hIf it exceeds, it can be considered strong enough. In this embodiment, the output channel tap can be derived as follows.

For n = 1, ..., L,<maths num="26"><img file="JP5113072B2_D0026.tif" /></maths>

Threshold T_hIs the noise energy E_{noise noise}, Can be set based on total energy, etc. For example, T_hIs T_h= β_h E_{noise noise}Can be set as, however, β_hCan be set to 1.5 or some other value. T_hCan also be a programmable value.

Input channel taps can also be scaled in other ways. For example, input channel taps may be scaled based on some other technique instead of the LMMSE technique with respect to points. Input channel taps can also be scaled together. In this embodiment, the correlation across multiple channel taps is determined and can be used to scale that channel tap.

FIG. 6 shows a block diagram of an embodiment of the channel estimator 410 for the first post-processing scheme. The initial channel estimator 610 receives the input sample y (n) and derives the initial CIRE, for example, as shown in Eq. (2). The filter 612 is a channel tap in the initial CIRE, for example, as shown in equation (3).<maths num="27"><img file="JP5113072B2_D0027.tif" /></maths>

Filtered and filtered channel taps<maths num="28"><img file="JP5113072B2_D0028.tif" /></maths>

I will provide a. Selector 614 taps the input channel for the first CIRE h_iInitial channel tap as (n)<maths num="29"><img file="JP5113072B2_D0029.tif" /></maths>

Or filtered channel tap<maths num="30"><img file="JP5113072B2_D0030.tif" /></maths>

I will provide a. For example, selector 614 is an initial channel tap for fast changing channels.<maths num="31"><img file="JP5113072B2_D0031.tif" /></maths>

For channels that change slowly, filter channel taps<maths num="32"><img file="JP5113072B2_D0032.tif" /></maths>

Can be provided.

The channel tap energy estimator 616 estimates the energy E (n) of each input channel tap, for example, as shown in Eq. (6) or (7). The noise energy estimator 618 relates to an input channel tap based on the energy of a small number of input channel taps at one or both ends of the first CIRE, for example, as shown in equation (10) or (11). Noise energy E_{noise noise}To estimate. The final channel estimator 620 has an input channel tap h, for example, as shown in equation (12) or (13)._i(n), channel tap energy E (n), and noise energy E_{noise noise}Output channel tap for the second CIRE based on h_{o o}Lead (n). The final channel estimator 620 can also filter output channel taps based on, for example, IIR filters or some other type of filter.

Figure 7 shows process 700 for deriving CIRE by scaling channel taps. The initial CIRE is derived in the time domain based on the pilot received, for example by backdiffusing multiple input samples using a pilot sequence (block 712). The initial CIRE is filtered and the filtered CIRE is obtained (block 714). The first CIRE with multiple channel taps is derived based on the initial CIRE or filtered CIRE (block 716). Channel taps in the first CIRE are scaled with multiple scaling factors to get the second CIRE (block 720).

In one embodiment of block 720, the energy of each channel tap in the first CIRE is estimated, for example, as shown by equation (6) or (7) (block 722). The noise energy associated with the channel taps in the first CIRE is also estimated, for example, based on the energies of a few channel taps at the left and / or right edges of the first CIRE (block 724). Each channel tap in the first CIRE is scaled based on a scaling factor determined by the energy and noise energy of that channel tap, for example, as shown in equation (12) (block 726). Threshold T_hEach channel tap with lower energy can be set to zero, for example, as shown in equation (13) (block 730). Threshold T_hCan be set based on noise energy or a predetermined value.

In the second post-processing scheme, the output channel taps in the second CIRE are obtained by zeroing the selected input channel taps in the first CIRE. The second scheme can be implemented in various ways.

In the first embodiment of the second scheme, the second CIRE has an adaptively selected length based on noise energy. In this embodiment, the leftmost energy E of the first CIRE_left, Rightmost energy E_right, Total energy E_totalCan be estimated as:<maths num="33"><img file="JP5113072B2_D0033.tif" /></maths>

However, P_leftAnd P_rightIs the number of channel taps used to calculate the left and right energies, respectively. In general, P_leftIs P_rightIt may or may not be equal to. As an example, L may be equal to 64, P_leftAnd P_rightMay both be equal to 15. P_leftAnd P_rightMay be a fixed value or a configurable value and can be selected to achieve good performance. E (n) is the input channel tap h as shown in equation (6)._iInstantaneous energy in (n), or as shown in equation (7), h_iIt can be the average energy of (n). Leftmost energy E_leftCan also be determined based on the first subset of channel taps on the far left of CIRE, and also on the far right energy E_rightCan be determined based on a second subset of channel taps on the far right of the CIRE. Each subset can include any number and any of any number of channel taps at the relevant ends.

P on the far left_leftFor each channel tap, the energy at the left end is the threshold T as shown below._leftZero if lower, retained otherwise: n = 1, ..., P_leftAgainst<maths num="34"><img file="JP5113072B2_D0034.tif" /></maths>

Similarly, the P on the far right_rightFor each channel tap, the energy at the right end is the threshold T as shown below._rightZero if lower, retained otherwise: n = LP_rightFor +1, ..., L<maths num="35"><img file="JP5113072B2_D0035.tif" /></maths>

Threshold T_leftAnd T_rightIs the total energy E_totalCan be set based on. For example, T_leftAnd T_rightIs T_left= T_right= β₁ E_totalCan be set as, where β₁Can be set to 0.05 or some other value. T_leftAnd T_rightCan also be a programmable value.

In the second embodiment of the second scheme, the input channel taps are continuously zeroed, one channel tap at a time, in a predetermined order, until some fraction of the total energy is removed. To. In this embodiment, the index j indicates the number of input channel taps to be considered and can be from 1 to L, i.e. j = 1, 2, ..., L. Input channel taps can be considered one at a time, for example, starting from both ends and moving toward the center, and alternating between the two ends. The metric j'indicates which input channel tap to consider and can be defined based on the metric j as follows: For odd j<maths num="36"><img file="JP5113072B2_D0036.tif" /></maths>

For even j<maths num="37"><img file="JP5113072B2_D0037.tif" /></maths>

The index j'first selects the input channel taps at the two ends and alternates between the right and left ends. Therefore, as the index j progresses from 1 to L, j'= 1, L, 2, L-1, 3, L-2, ..., L / 2, L / 2 + 1.

Cumulative energy E for input channel taps of j_acc(j) can be expressed as: E_acc(j) = E_acc(j-1) + E (j'), Equation (21) However, E (j') is the input channel tap h_iThe estimated energy of (j'), which can be derived as shown in Eq. (6) or (7), and also E_acc(j) is E_accIt can be initialized with (0) = 0.

Input channel tap h for each value of j_i(j') is selected and the accumulated energy E_acc(j) is updated with the energy E (j') of the selected channel tap. Input channel tap h_i(j') is the updated accumulated energy E as follows:_acc(j) is the threshold T_accZero if lower, retained otherwise: For j = 1, ..., L<maths num="38"><img file="JP5113072B2_D0038.tif" /></maths>

Threshold T_accIs the total energy E_totalCan be set based on. For example, T_accIs T_acc= β_acc E_totalCan be set as, however, β_accCan be set to 0.05 or some other value. T_accCan also be a programmable value.

In a third embodiment of the second scheme, weak input channel taps are continuously zeroed, one channel tap at a time, in a predetermined order, until some of the total energy is removed. .. In this embodiment, the index j can be from 1 to L. Input channel tap h for each value of j_iFor (j'), its energy E (j') is the threshold T_hIf lower, it can be considered weak. Input channel tap h_iIf (j') is weak, (1) accumulated energy E_acc(j) is updated to include the energy of this channel tap, and (2) the updated accumulated energy is the threshold T._accIf lower, this channel tap is zeroed. Input channel tap h_i(j') is the threshold T if it is not weak or the updated accumulated energy_accRetained if not lower. Threshold T_hAnd T_accIs the noise energy E_{noise noise}, Total energy E_totalIt can be set based on, or it can be a programmable value.

In the fourth embodiment of the second scheme, the input channel taps are ordered from the weakest to the strongest, one channel tap at a time, sequentially until some of the total energy is removed. It is continuously zeroed in order. In this embodiment, the index j can be from 1 to L. For each value of j, the weakest channel tap of all unconsidered input channel taps is selected and the accumulated energy E_acc(j) is updated to include the energy of the selected channel tap. The selected channel tap has the updated accumulated energy threshold T_accIf it is lower, it is set to zero, otherwise it is retained. Threshold T_accIs the total energy E_totalIt may be set based on, or it may be a programmable value.

In the fifth embodiment of the second scheme, the weak initial channel tap is zeroed. The energy of the input channel tap is the threshold T_hIf it is lower, it can be considered weak. In one embodiment, the threshold T_hIs the total energy E_totalBased on, for example, T_h= β_ht E_totalSet as, however, β_htCan be set to 0.01, or some other value. In other embodiments, the threshold T_hIs the noise energy E_{noise noise}Based on, for example, T_h= β_hn E_{noise noise}Set as, however, β_hnCan be set to 1.5, or some other value. In yet another embodiment, the threshold T_hIs the energy of the largest input channel tap E_peakBased on, for example, T_h= β_hp E_peakSet as, however, β_hpCan be set to 0.01, or some other value. Threshold T_hCan also be set based on some other quantity, or it may be a programmable value.

Various embodiments have been described above for selectively zeroing input channel taps. The input channel tap can also be zeroed in other ways, which is within the technical scope of the present invention.

Looking back at Figure 6, the final channel estimator 620 is the initial channel tap.<maths num="39"><img file="JP5113072B2_D0039.tif" /></maths>

Or filtered channel tap<maths num="40"><img file="JP5113072B2_D0040.tif" /></maths>

Can be an input channel tap h_iReceive (n). The final channel estimator channel 620 also has channel tap energy E (n), end energy E_leftAnd / or E_right, Total energy E_total, Noise energy E_{noise noise}, And the following can be received as well. The final channel estimator 620 can derive one or more thresholds based on the energy received and / or the programmable value. The final channel estimator channel 620 can then zero out a selection of input channel taps based on any one, or any combination, of the embodiments described above. ..

Figure 8 shows process 800 for deriving CIRE by selectively zeroing channel taps. Early CIRE is derived in the time domain, based on the pilot received, for example by backdiffusing multiple input samples using a pilot sequence (block 812). The initial CIRE is filtered and the filtered CIRE can be obtained (block 814). The first CIRE with multiple channel taps is derived based on the initial CIRE or filtered CIRE (block 816). The selected channel tap in the first CIRE is set to zero and the second CIRE is acquired (block 820).

In the first embodiment described above, the first subset of channel taps on the left edge of the first CIRE is the total energy E for these channel taps._leftIs the first threshold T_leftIf lower, it is set to zero (block 822). The second subset of channel taps on the far right of the first CIRE also has a total energy E for these channel taps._rightIs the second threshold T_rightIf lower, it is set to zero (block 824). The first and second subsets can contain the same or different number of channel taps. Each subset can include a portion of the total number of channel taps in the first CIRE (eg, about a quarter). The first threshold may be equal to the second threshold, and both thresholds can be set based on the total energy of the channel taps in the first CIRE.

In the second, third, and fourth embodiments described above, the threshold T_accAt least one channel tap with less total energy is set to zero (block 832). Threshold T_accCan be set based on the total energy of the channel taps in the first CIRE. In the second and third embodiments, the channel taps to be zeroed alternate between the left and right ends of the first CIRE, and traverse from the two ends toward the center. , Selected in a predetermined order. In a fourth embodiment, the channel taps in the first CIRE are ranked from the weakest to the strongest, and the channel taps to be zeroed are in sequential order starting with the weakest channel tap. Is selected with. In a fifth embodiment, the threshold T_hEach channel tap with less energy is set to zero (block 842). Threshold T_hCan be set based on the total energy, peak energy, or noise energy of the channel taps in the first CIRE.

Input channel taps can also be processed based on both the first and second post-processing schemes. For example, input channel taps can be zeroed based on any of the embodiments described above, and non-zeroed input channel taps are based on the (tap-wise) LMMSE technique for taps. Can be scaled.

The second CIRE can be used to derive the coefficients for the equalizer, as shown in Figure 4. The second CIRE can also be used for rake receivers. The rake receiver may include multiple finger processors (or simply "fingers") and assign each finger to process an interest signal instance. it can. A second CIRE can be used to identify a strong signal instance assigned for processing by the finger. A second CIRE can also be used to weight the assigned finger outputs before combining these outputs.

The channel estimation techniques described herein can be implemented by a variety of means. For example, these techniques can be performed on hardware, firmware, software, or a combination thereof. In hardware implementations, the processors used to perform channel estimation are one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processors (DSPDs), and programs. Possible logic devices (PLDs), field programmable gate arrays (FPGAs), processors, controllers, microcontrollers, microprocessors, electronic devices, others designed to perform the functions described herein. Can be implemented within an electronic device, or a combination thereof.

For firmware and / or software implementations, the technique can be performed using modules that perform the functions described herein (eg, procedures, functions, etc.). Firmware and / or software code (codes) may be stored in memory (eg, memory 292 in FIG. 2) and executed by a processor (eg, processor 290). Memory can be implemented inside or outside the processor.

Previous descriptions of the disclosed embodiments are provided to allow any person skilled in the art to create or use the present invention. Various changes to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein will not deviate from the technical scope of the invention. Can be applied to. Therefore, the invention is not limited to the embodiments presented herein, but should be given the broadest scope consistent with the principles and novel features disclosed herein.

<figref num="1">The figure which shows the transmission transmission in a wireless communication system.</figref><figref num="2">Block diagram of base station and wireless device.</figref><figref num="3">Block diagram of a CDMA modulator in a base station.</figref><figref num="4">Block diagram of an equalizer in a wireless device.</figref><figref num="5">The schematic which shows the exemplary channel impulse response estimation.</figref><figref num="6">Block diagram of the channel estimator.</figref><figref num="7">Diagram showing the process for deriving CIRE by scaling channel taps.</figref><figref num="8">Diagram showing the process for deriving CIRE by selectively zeroing channel taps.</figref>