HK1173868B - Channel estimation filtering - Google Patents

Description

Channel estimation filtering

Cross Reference to Related Applications

This patent document claims priority from us provisional patent application No. 61/428,222 entitled "OFDM Channel Estimation Filtering" filed on 12/29/2010.

The entire contents of the above-referenced provisional patent application are incorporated herein by reference as part of this patent document.

Technical Field

This patent document relates to apparatus, systems, and methods for Orthogonal Frequency Division Multiplexing (OFDM) or Orthogonal Frequency Division Multiple Access (OFDMA) based communications.

Background

Wireless communication systems use electromagnetic waves to communicate with wireless communication devices that are located within cells of the system's coverage area. The radio spectrum range or frequency band designated or allocated for a wireless communication device or a particular type of wireless service may be divided into different radio carrier frequencies for generating different communication frequency channels. For a given wireless system, communication capacity increases as the number of communication frequency channels increases. Two different frequency channels may interfere or cross-talk with each other when their frequencies are too close to each other, creating noise, thereby reducing the signal-to-noise ratio.

One technique to reduce the minimum frequency spacing between two adjacent channels is to generate different channels within a given frequency band by using Orthogonal Frequency Division Multiplexing (OFDM) to generate channel spectral profiles that are orthogonal to each other and free of interference when the different channels are centered at selected equally spaced frequencies. Under OFDM, the frequency spacing can be less than the minimum spacing in conventional frequency channels, and thus increase the number of channels within a given frequency band.

Existing and evolving specifications under the ieee802.16x standard and the Long Term Evolution (LTE) standard support wireless communication under OFDM and OFDMA. For example, the draft of ieee802.16d published in month 1 of 2004 provides technical specifications for OFDM and OFDMA wireless systems.

Summary of The Invention

This document provides devices, systems and methods for communication, including OFDM and OFDMA communication devices, systems and methods, that provide improved channel estimation as part of the process of decoding a transmission signal.

In one aspect, a method of wireless communication is disclosed. A plurality of reference signal transmissions are received on a plurality of subcarriers. A plurality of coarse channel estimates are calculated based on the plurality of received reference signals. The plurality of coarse channel estimates are processed to obtain an improved channel estimate.

In another aspect, a system for wireless communication is disclosed. The system includes a receiver for receiving a plurality of reference signal transmissions on a plurality of subcarriers; a coarse channel estimator for calculating a plurality of coarse channel estimates based on the received plurality of reference signals; and an improved channel estimator for processing the plurality of coarse channel estimates to obtain an improved channel estimate.

In another aspect, a computer program product comprising a non-transitory computer-readable medium having computer-executable code stored thereon is disclosed. The computer executable code, when executed, causes a computer to implement a channel estimation method comprising: receiving a plurality of reference signal transmissions on a plurality of subcarriers; calculating a plurality of coarse channel estimates based on the plurality of received reference signals; and processing the plurality of coarse channel estimates to obtain an improved channel estimate.

In another aspect, an Orthogonal Frequency Division Multiplexing (OFDM) -based communication system is disclosed that includes a network of base stations spatially distributed within a service area to form a radio access network for wireless communication devices. Each base station includes an antenna operable to receive a baseband signal having a cyclic prefix, the baseband signal corresponding to a channel, and a processor in communication with the antenna. The processor is configured to: receiving a baseband signal with a cyclic prefix, the baseband signal corresponding to a channel; removing the cyclic prefix of the baseband signal; transforming the baseband signal from a time domain to a frequency domain to produce a baseband signal in the frequency domain; compensating the baseband signal in the frequency domain by a first time offset of the received baseband signal to produce a compensated baseband signal in the frequency domain; extracting subcarriers used for transmitting the baseband signal from the compensated baseband signal in the frequency domain to generate a coarse channel estimate; determining neighboring subcarriers having channel responses highly correlated to the subcarriers of the channel; generating a compensated coarse channel estimate by compensating the coarse channel estimate in the frequency domain by a second time offset, wherein the second time offset corresponds to a minimum angle of an autocorrelation function of the subcarrier and neighboring subcarriers; and filtering the compensated coarse channel estimate with a filter to produce a filtered channel estimate. The filter length of the filter corresponds to the sum of the sub-carrier and the adjacent sub-carrier.

These and other aspects are described in more detail below.

Brief description of the drawings

The components in the figures are not necessarily to scale, emphasis instead being placed upon schematically illustrating various aspects thereof. Further, in the figures, like reference numerals designate corresponding parts throughout the different views.

Fig. 1 illustrates an example of a wireless communication system in which an OFDM receiver system may be implemented, according to one embodiment.

Fig. 2 is a schematic diagram of an exemplary OFDM receiver system.

Fig. 3 is a schematic diagram of an exemplary OFDM receiver system, according to one embodiment.

Fig. 4 is a diagram of an example autocorrelation function for a wireless channel.

Fig. 5 is a schematic diagram of an example autocorrelation function for a wireless channel.

FIG. 6 is a schematic diagram of an exemplary linear interpolation function implemented by an interpolation matrix multiplication operation.

Fig. 7 is a schematic diagram of an example wireless network architecture.

Fig. 8 is a schematic diagram of the structure of a wireless station.

Fig. 9 is a flow chart of a wireless communication process.

Fig. 10 is a schematic diagram of a structure of a portion of a wireless communication device.

Detailed Description

In one aspect, the disclosed techniques may be used to improve the performance of an OFDM receiver. On the other hand, the disclosed techniques preserve the performance and operation of the OFDM receiver under varying signal-to-noise conditions.

OFDM receivers typically require channel estimation as part of the process of decoding the received signal. To assist the receiver in estimating the channel response, the transmitter is typically configured to periodically transmit a known reference signal (sometimes referred to as a pilot signal) on certain subcarriers. By comparing the actual received signal with the known transmitted signal, the receiver is able to estimate the channel response.

Further, as in the Long Term Evolution (LTE) OFDMA uplink, it often happens that the reference signal is configured to occupy several adjacent subcarriers. Because the channel responses between adjacent subcarriers are generally highly correlated, the channel responses of adjacent or nearby subcarriers may be combined and averaged to produce an improved channel estimate. This technique is sometimes referred to as channel estimation filtering in the frequency domain.

Filtering can be a computationally intensive operation. In some cases, the number of multiplication operations may be reduced by exploiting the correlation of adjacent subcarriers of the channel for which the estimate is calculated. As described further below, highly correlated adjacent subcarriers may be identified and the coarse channel estimate in the frequency domain may be compensated by a time offset corresponding to a minimum angle of the autocorrelation function of the channel subcarriers and the adjacent subcarriers. The generated compensated coarse channel estimate may be filtered by a filter having a length corresponding to the number of highly correlated adjacent subcarriers, thus reducing the number of multiplications required to produce the filtered channel estimate.

Fig. 1 illustrates an example wireless communication system 100 that provides wireless communication services using communication channels at different frequencies under OFDM or OFDMA, according to one embodiment. The system 100 may include a network of Base Stations (BSs) or Base transceiver stations (BSTs) 120 spatially distributed within a service area to form a wireless access network for wireless users or wireless Subscriber Stations (SSs) 110. In some implementations, the base station 120 may be designed with directional antennas and generate two or more directional beams to further divide each cell into different portions. A Base Station Controller (BSC) 130 is generally connected to the BS120 using a wire or cable and controls the connected BS 120. Each BSC130 typically connects to and controls two or more particular BSs 120.

Wireless system 100 may include a carrier packet Network 140, and carrier packet Network 140 may be connected to one or more Public Switched Telephone Networks (PSTN) 150 and one or more packet data networks 160 (e.g., IP networks). A Mobile Switching Center (MSC)152 may serve as an interface between BSC130 and PSTN 150. Packet data serving node 142 may be used to provide an interface between carrier packet network 140 and packet data network 160. In addition, a carrier packet network manager 144 may interface with carrier packet network 140 to provide various network management functions, such as authentication, authorization, and accounting (AAA) functions provided by an AAA server.

Each subscriber station 110 may be a stationary or mobile wireless communication device. Examples of stationary wireless devices may include desktop computers and computer servers. Examples of mobile wireless devices may include mobile wireless telephones, Personal Digital Assistants (PDAs), and laptop computers. Subscriber station 110 may be any communication device capable of wirelessly communicating with base station 120.

In one embodiment, the system of fig. 1 may be applied to communication bands of 2 to 11GHz under OFDM and OFDMA provided in the ieee802.16x standard, such as ieee802.16d (1 month of 2004). In OFDM and OFDMA systems, the available frequency band is divided into orthogonal subcarriers at different frequencies. In an OFDMA system, subchannels are formed from subsets of subcarriers. Within OFDMA, a total of 32 subchannels are allocated to each radio cell.

The Base Station (BS) or base transceiver station (BST)120 may include an OFDM receiver. OFDM receivers typically require channel estimation as part of the process of decoding the transmitted signal. To assist the receiver in calculating the response of the channel, the transmitter is typically configured to periodically transmit a known reference signal (also referred to as a pilot signal) on certain subcarriers. By comparing the actual received signal with the known transmitted signal, the receiver is able to estimate the channel response.

Further, as in the LTE OFDMA uplink, it often happens that the reference signal is configured to occupy several adjacent subcarriers. Because the channel responses between adjacent subcarriers are generally highly correlated, the channel responses of adjacent or nearby subcarriers may be combined and averaged to produce an improved channel estimate.

Fig. 2 shows a schematic diagram of an OFDM receiver that performs channel estimation filtering in the frequency domain. The process of the example shown in system 200 of producing filtered channel estimates may be performed using any combination of hardware and software suitable for implementing the required functions disclosed herein. In one embodiment, the hardware may include a digital signal processor, such as a wireless base station system-on-chip of TMS320TCI6616 by Texas instruments Incorporated of Dallas, Tex. The digital signal processor may be configured with processor-executable instructions. The processor-executable instructions may be stored in a Random Access Memory (RAM) within or in communication with the data signal processor or in a non-volatile Memory within or in communication with the data signal processor, such as a Read Only Memory (ROM), an Electrically Erasable Read Only Memory (EEPROM), or an E-flash (Embedded flash). The executable instructions that configure the digital signal processor may implement a number of software modules or applications that communicate with each other and with hardware and software within or outside the base station in order to implement the functions of the OFDM receiver 200.

The received baseband signal 202 is sent to a module 204, and the module 204 removes a Cyclic Prefix (CP) of the OFDM symbol. The output of this block 204 is sent to a systematic DFT block 206, which systematic DFT block 206 transforms the signal from the time domain to the frequency domain.

For filtering the channel estimate in the frequency domain, the basic requirements of an OFDM receiver are: the OFDM receiver must perform a time offset Estimation (TOE, time offset Estimation) as indicated by the TOE module 216 in fig. 2. There are several TOE algorithms available in the literature. One exemplary TOE algorithm operates in the time domain and performs a time domain correlation of the actual received signal to a known transmitted time domain signal. The peak position of the time domain correlation is taken as the time offset of the received signal.

Time Offset Compensation (TOC)208 is performed after the system DFT206 based on the Time offset estimated by the TOE module 216. Length N_DFTVector D of (2) represents the output of the system DFT, where N_DFTIs the size of the system DFT.

Further, let D (i) (i-th element of vector D) represent received time domain n signalA frequency component.

It is noted that in this document, all vectors are assumed to be column vectors unless otherwise stated. Further, the index of the vector or matrix starts with a value of 0. Thus, for the D (i) vector described above, i takes on values from 0 to N_DFT-1, inclusive.

If T is₁To represent the vector of the output of the TOC module 208, then:

equation 1

Wherein, t_dIndicating the delay to be compensated. In general, t_dIs a floating point number but is proportional, so if t is_dThis is 1, which means that the time offset is estimated as one sample, sampled at the sample rate of the signal entering the system DFT 206.

After the time offset is compensated, the signal enters a Pilot Compensation (PC) module 210, which extracts the subcarriers used to transmit the reference sequence and also compensates for the transmitted known reference signal. Make H_rawRepresenting the complex-valued vector output by the pilot compensation module.

Equation 2

N_REFIs the number of sub-carriers occupied by the reference symbol. P₁Is the position of the first reference symbol in the output of the system DFT. P is a length N that accommodates known reference symbols sent by the transmitter_REFVector of (assuming no reference symbol has a value of 0). For example, assume that the transmitter is known to transmit 35 reference symbols (vector P) on subcarriers 123 through 157. In this case, P₁Is 123, N_REFIs 35.

Briefly, H_rawIs a vector that accommodates the coarse channel estimate. In an ideal, noise-free system, due to T₁(i+P₁)＝H_actual(i)P(i)，H_rawWill be represented perfectly at the transmitter and receiver (H)_actual) The channel response in between. However, in real world systems, noise corrupts the channel estimates.

The channel estimate filter 212 module attempts to exploit the known a priori autocorrelation properties of the channel response to produce a refined channel estimate with improved accuracy.

Make H_actualRepresenting the actual frequency domain channel response from the transmitter to the receiver, here viewed as a complex-valued random column vector. Further, let R_HAn autocorrelation matrix representing the channel:

equation 3

Wherein, E2]Is an expectation operator, the H superscript represents the Hermite conjugate transform operation, anIs the average received power of the channel.

Note H_actualIncluding the effects of the wireless channel and the effects of the receiver blocks through to the CEF block 212. Thus, for different TOE and TOC algorithms, R_HMay be different.

In estimation theory, the linear system model is written as:

equation 4x_e＝H_eθ_e+w_e

Wherein, theta_eThe column vector is the parameter to be estimated, H_eIs an observation matrix by which the parameters are observed, w_eIs a noisy column vector that corrupts the estimate, and x_eThe column vectors hold the actual observations.

For the Linear system model in equation 4, the Linear Minimum Mean square Error estimate (LMMSE, Linear Minimum Mean Squared Error Estimator) is given by:

equation 5

Wherein, C_XIs the covariance matrix of vector X (i.e., C)_X＝E[XX^H])。

Applied to this problem, the parameter θ to be estimated_eIs H_actual。H_actualIs passed through the identity matrix (H)_eI) and is observed by gaussian receiver white noise vector N (i.e., w)_eN) destruction. Autocovariance matrix of noise vectorsIs thatWhereinIs the average power of the noise introduced by the receiver. Thus, the linear system model of the current problem is given by:

equation 6H_raw＝IH_actual+N

Since the channel estimate can be assumed to have zero mean, one channelEstimate hasDistributing:and x_e＝H_rawA channel estimate has the following well-known results:

equation 7

Equation 8H_filt＝M_smoothH_raw

Thus, if the channel estimate is coarse (H)_raw) And N_REFxN_REFMatrix M_smoothMultiplication, the result will be a smoothed and filtered channel estimate H_filt214, channel estimation H_filt214 will have better noise properties than the initial coarse channel estimate. In fact, H_filtThere will be a minimum mean square error in all linear estimators.

One problem is to combine M_smoothThe application of the matrix to the coarse channel estimate is very expensive. For each filtered channel estimate, N is required_REFComplex multiplication.

Fig. 3 illustrates an example OFDM receiver system according to one embodiment. The received baseband signal 202 is sent to a module 204, and the module 204 removes the Cyclic Prefix (CP) of the OFDM symbol. The output of this block 204 is sent to a systematic DFT block 206, which systematic DFT block 206 transforms the signal from the time domain to the frequency domain.

The received baseband signal is also processed by a Time Offset Estimation (TOE) module as indicated by TOE module 212 in fig. 3. There are several TOE algorithms available in the literature. One exemplary TOE algorithm operates in the time domain and performs a time domain correlation of the actual received signal to a known transmitted time domain signal. The peak position of the time domain correlation is taken as the time offset of the received signal.

Time Offset Compensation (TOC)208 is performed after the system DFT206 based on the time offset estimated by the TOE module. Make length N_DFTVector D of (2) represents the output of the system DFT, where N_DFTIs the size of the system DFT. Further, let D (i) (i-th element of vector D) represent the received time-domain signalA frequency component. Note that i is taken from 0 to N_DFT-1, inclusive.

If T is₁To represent the vector of the output of the TOC module 208, then:

equation 9

Wherein, t_dIndicating the delay to be compensated. In general, t_dIs a floating point number but is proportional, so if t is_dThis is 1, which means that the time offset is estimated as one sample, sampled at the sampling rate of the signal entering the system DFT.

After the time offset is compensated, the signal enters a Pilot Compensation (PC) module 302, which extracts the subcarriers used to transmit the reference sequence and also compensates for the transmitted known reference signal. Make H_rawRepresenting the vector output by the pilot compensation module 210.

Equation 10

N_REFIs the number of sub-carriers occupied by the reference symbol. P₁Is the position of the first reference symbol in the output of the system DFT. P is a length N that accommodates known reference symbols sent by the transmitter_REFVector of (assuming no reference symbol has a value of 0). For example, assume that the transmitter is known to transmit 35 reference symbols (vector P) on subcarriers 123 through 157. In this case, P₁Is 123, N_REFIs 35.

Briefly, H_rawIs a vector that accommodates the coarse channel estimate. In an ideal, noise-free system, H_rawWill be represented perfectly at the transmitter and receiver (H)_actual) The channel response in between. However, in real world systems, noise corrupts the channel estimates.

It should be noted that for simplicity, the discussion herein assumes E [ H ]_raw]Is zero. It will be easy for a person skilled in the art to attribute the formula set out herein to a non-zero average value.

In this regard, the autocorrelation function of the coarse channel estimateThe method comprises the following steps:

equation 11

Exemplary autocorrelation functions for a generic wireless channel are shown in fig. 4 (graph 400) and fig. 5 (graph 500)The magnitude of the correlation function is given in fig. 4 and it can be seen that for this particular channel, the subcarriers adjacent to a particular subcarrier have a very high correlation coefficient (approximately 0.997). In fig. 5, on the other hand, the angle of the same exemplary autocorrelation function is given, where it can be seen that in this particular case, the angle is approximately, but not completely, linear. General assemblyIt is not necessary that the autocorrelation function be approximately linear.

System 300 operates by exploiting the fact that the channel responses of adjacent subcarriers are highly correlated in many different types of propagation channels. The number of adjacent samples that can be considered "highly correlated" will vary based on the statistical properties of the channel and performance required by the receiver. For example, in typical implementations (where receiver performance is highest), it is not reasonable to assume that the adjacent carriers are highly correlated, and therefore an expensive receiver must be implemented that does not attempt to exploit the high correlation between adjacent subcarriers. However, in some receiver designs, some performance penalty may be acceptable when it comes to significant savings in implementation cost in view of the performance penalty. The specific trade-off between implementation cost and performance may be determined as appropriate.

In one embodiment, the number of subcarriers considered to be highly correlated is represented by the variable B. For example, if B is 3, the embodiment assumes that 3 subcarriers before a specific subcarrier, the specific subcarrier itself, and 3 subcarriers after the specific subcarrier are highly correlated. A total of 2B +1 subcarriers are considered highly correlated. Although the actual value of B will vary from receiver to receiver and from channel to channel, B is expected to be a very small value between 1 and 5.

Various methods may be used to determine a suitable value for B. For example, link level simulations may be performed to quantify the performance loss for each B value. Another estimation of the implementation savings may also be performed on the individual values of B. In one embodiment, the selected value of B has little impact on performance but a large impact on implementation cost.

It is difficult to provide an accurate definition of how it can be determined how many subcarriers are "highly correlated". Instead, the techniques described in the above figures can be used to trade off B values between performance and implementation cost.

To take full advantage of the high degree of correlation between close subcarriers,the system 300 introduces a second TOC module 302 in the receive chain that can smooth the phase response of the autocorrelation function. In one embodiment, t is independent of the second TOC module_RWhat the value of the parameter is, the amplitude of the autocorrelation function of the signal output by the second TOC module will be the same as the amplitude of the autocorrelation function of the signal output by the PC module. However, the phase will be affected.

Second TOC Module (T)₂) The signal output by 302 is:

equation 12

The autocorrelation function of the output of the second TOC module 302 may be defined as:

equation 13

Wherein for a subcarrier offset between-B and + B, t_RCan be selected to be H'_rawThe angle of the autocorrelation function of (a) will be close to zero. If it is notIs in fact a straight line, then for the sub-carrier offset between-B and + B, t is chosen_RValue of (1)It is possible that the angle of (c) is exactly 0. However, in general terms,is less likely to be a straight line, and therefore,is rarely exactly equal to zero.

One simple strategy is to ignoreIs not onlyAndall of the points of (a). t is t_RAnd then can be assigned as:

equation 14

Wherein the angle operator returns the angle of its parameter measured in radians.

Alternatively, only the use ofAndregardless of the value of B:

equation 15

Autocorrelation function of the output of the second TOC module 302Has the structure asBut has a phase closer to 0 in the range-B to + B. In particular, the method comprises the following steps of,

equation 16

The output of the second TOC module 302 is sent to a modified channel estimation filter (modified CEF) module 304.

The first step that may be taken within the improved CEF module 302 is that a filter, such as a simple all-one filter, is applied to the input data in order to reduce the total amount of data that needs to be processed. The filter can be expressed as a matrix operation:

equation 17H_D＝QH′_raw

In one embodiment, the Q matrix within the CEF module 302 is created by first creating N_REFxN_REFMatrix construction, N_REFxN_REFThe matrix will have the main diagonal divided by the 2 x B diagonal on the main diagonalWith the value 1, all other terms are zero. In summary, the 2 × B +1 diagonal contains the value 1 and all other diagonals contain the value 0. From this matrix, row 1, row B +1, row 2B +1, row 3B +1, etc. are retained up to a maximum of rows N_REF-2 × B. For example, if N_REFIs 9 and B is 2:

equation 18

In all cases, each row of Q contains exactly 2 × B +1 and the other values are 0. Other similar or equivalent Q matrix values or arrangements are possible.

Note that since only addition is involved, H'_rawVector multiplication by Q matrix is a very low cost operation.

In estimation theory, the linear system model is written as:

equation 19x_e＝H_eθ_e+w_e

Wherein, theta_eThe column vector is the parameter to be estimated, H_eIs an observation matrix by which the parameters are observed, w_eIs a noisy column vector that corrupts the estimate, and x_eThe column vectors hold the actual observations.

For the linear system model in equation 19, the linear minimum mean square error estimate (LMMSE) is given by:

equation 20

Wherein, C_XIs the covariance matrix of vector X (i.e., C)_X＝E[XX^H])。

Applied to this problem, the system formula is:

equation 21H_filt＝QH_actual+QN

The parameter theta to be estimated_eIs H_actual。H_actualIs passed through a Q matrix (H)_eQ) and is observed by a noise vector QN (i.e., w)_eQN) fails. Autocovariance matrix of noise vectorsIs thatWhereinIs the average power of the noise introduced by the receiver.

Distributing:and x_e＝H_DComprising the following steps:

equation 22

Equation 23H_filt＝M′_smoothH_D

Equation 24H_filt＝M′_smoothQH′_raw

Based on H_DIn all linear estimations, the above formula preferably estimates H_actual. Not directly filtering involving N_refH 'of one sample'_raw，H_DIs filtered and it contains about N_REFAnd the number of the samples is/B, which is already approximately one half of B, so that the implementation cost is reduced.

By recognition of M'_smoothMatrix simultaneous averaging H_DSamples and performing interpolation between them to produce a total of N_REFEstimation, implementation costs of the channel estimation filter 304 may be further improved. Since the user channel responses are highly correlated, further improvements are possible by separating the averaging operation from the interpolation operation.

A Q 'matrix may be created, which will apply to M'_smoothMatrices are used to reduce the number of operations that need to be performed. The matrix is created by detecting the Q matrix line by line. For each row, the Q matrix contains an odd number of successive ones1. The other elements are 0. The Q' matrix is created by holding the middle-most 1 of each row while setting all other values to zero. For example, the Q matrix previously shown in equation 18 is converted to a Q' matrix as:

equation 25

The final form of the channel estimation filter equation is:

equation 26H_filt＝X((Q′M′_smooth)(QH′_raw))

Where X is the interpolation matrix. Other similar or equivalent matrix values or arrangements for Q', Q, and X may also achieve implementation cost improvements.

The parenthesis section in the above equation is used to indicate how to expect implementation of the channel estimation filter 304 in order to minimize implementation cost. First, Q matrix is applied to H'_raw. This is a simple operation that requires only a few additions to be performed. QH'_rawWill be taken as comprising an approximation of N_REFVector of/B elements, where H'_rawContaining N_REFAnd (4) each element.

In addition, M 'will be calculated based on the received signal power and the received noise power'_smooth. This will result in a signal having N_REFLines and approximate N_REFA matrix of/B columns. Application of a Q 'matrix to this matrix will simply eliminate M'_smoothAnd a final mean square matrix is generated, where each side approximately has a size (N)_REF/B)x(N_REF/B).

Q′M′_smoothThe matrix is then applied to vector QH'_raw. This yields an approximation of N_REFVector of/B samples. For subcarriers B +1, 2 × B +1, 3 × B +1, etc., these samples may be interpreted as smoothed channel estimates. The interpolation matrix X will then interpolate these estimates and pair from 1 to N_REFProduces a final estimate 306.

Several possible interpolation matrices may be utilized. The simplest interpolation matrix performs linear interpolation based on the nearest two available smoothed channel estimates. A graphical example 600 of the operation of the interpolation matrix is shown in fig. 6. The three solid vertical arrows in FIG. 6 (Users 2, 4 and 6) represent the symbol by which Q 'M'_smoothMatrix applied to vector QH'_rawThe resulting estimate. These samples are used unmodified. The remaining samples are created by linear interpolation between the available subcarriers. For example, subcarrier 3 is created by averaging subcarriers 2 and 4. Subcarriers 0 and 1 are created by visually generating a line connecting estimates (represented by dashed lines in fig. 6) of subcarriers 2 and 4 and calculating the value of the line for subcarriers 0 and 1. Subcarriers 5, 7, and 8 are similarly created based on a line connecting the channel estimates of subcarriers 4 and 6.

For example, if N_REFBeing 9 and B being 2, the interpolation matrix X will be:

equation 27

Other interpolation matrices may be utilized. Linear interpolation using two samples may be the best matrix to use in many cases because of its ease of implementation. In particular, for linear interpolation, each filtered channel estimate requires at most two real-valued multiplications.

The advantage of this embodiment can be seen by correlating equation 26 with a filtering equation that does not exploit the high correlation between adjacent subcarriers.

Equation 28H_filt＝X((Q′M′_smooth)(QH′_raw))

Equation 29H_filt＝M_smoothH_raw

Apply Q matrix to H 'in equation 26'_rawIs a low cost operation involving the use of additions only. Further, the X matrix applied in equation 26 is also a low cost operation, and each filtered channel estimate includes only two simple multiplications. Note that when the interpolation factor is as simple as in equation 27, multiplication can often be performed as a shift and add operation.

Will be M in equation 29_smoothTo H_rawAnd Q ' M ' in formula 26 '_smoothTo QH'_rawAre compared between applications. M_smoothIs to requireApplication of a complex multiplication method to H_rawN of (A)_REFxN_REFAnd (4) matrix. Q 'M'_smoothIs approximately (N)_REF/B)x(N_REF/B) size and therefore approximation requirementsApplication of complex multiplication to QH'_raw. Clearly, even though B is only 2, there is a significant reduction in implementation cost within this embodiment over other implementations.

When Q ' M ' is recognized '_smoothThe coefficients along its main diagonal are usually the largest and further reductions in implementation are possible as coefficients within the matrix decrease generally more rapidly with increasing distance from the main diagonal. Instead of using the complete Q 'M'_smoothMatrices, simplified matrix M may be used instead_simpIt has a form similar to the following formula:

equation 30

Square matrix Q 'M'_smoothSize and square matrix M of_simpAre the same size. W represents Q 'M'_smoothIs measured in one of the dimensions of (a). Further, let L denote M_simpThe number of non-zero coefficients of each row of (a).

If L is an odd number, the leftmost L term of the upper (L +1)/2 line and the rightmost L term of the lower (L +1)/2 line will accommodate the coefficients. For lines (L +1)/2 through W- (L +1)/2-1, the coefficient position for any particular line will be at the same position as the non-zero entry for the preceding line, but shifted one position to the right.

If L is an even number, the leftmost L term of the upper L/2 row and the rightmost L term of the lower L/2+1 row will accommodate the coefficients. For the L/2 row to the W-L/2-2 row, the coefficient position for any particular row will be at the same position as the non-zero entry for the preceding row, but shifted one position to the right.

M_simpCan be calculated by calculating Q 'M'_smoothTo solve, like N_refEqual to L × (B +1) + B. Let in assume N_refCalculated Q ' M ' under L — (B +1) + B '_smoothThe matrix is called M_ref. The matrix size is LxL, and d is_a，bIs shown at M_refRow a, column b. For assigning coefficients to M shown above_simpAn exemplary matrix, the following arrangement will be used:

equation 31c_a，b＝d_a，b 0≤a≤1 0≤b≤4

Equation 32c_a，b＝d_a，b 0≤a≤1 0≤b≤4

Equation 33c_a，b＝d_a-7，b 10≤a≤11 0≤b≤4

Thus, for odd numbers L, M_refIs allocated to M_simpAnd M is in the upper (L +1)/2 rows of_refIs allocated to M_simpThe following (L +1)/2 lines. Intermediate rows are all assigned M_refThe coefficients of (L +1)/2-1 lines of (1). For even numbers L, M_refIs assigned to M_simpUpper L/2 lines of (1), and M_refIs assigned to M_simpThe lower L/2 line. Intermediate rows are all assigned M_refThe coefficients of the L/2-1 line of (1).

This additional reduction is alsoThe number of necessary multiplications is reduced. Wherein Q ' M ' is '_smoothMatrix applied to QH'_rawRequire aboutMultiple multiplication of M_simpTo QH'_rawRequiring about L N_refB multiplications.

The actual value chosen for L will depend on the simulation, which will trade off performance that may be lost using small values of L against simplified implementation costs.

It will be apparent to those skilled in the art that several modifications to the embodiments are possible without departing from the essence of the embodiments. For example, where the illustrated embodiment includes two TOC modules, it should be clear to one skilled in the art that the TOC and PC modules within the system 300 can be freely swapped and executed in any order. Further, performing a TOC operation with parameter a followed by a TOC operation with parameter b is the same as performing a TOC operation with parameter a + b. In the exemplary embodiment, the TOC operation is split in order to simplify the explanation of the description of the embodiment. Implementation in a real environment may simplify the execution of a single TOC operation by combining the two TOC operations described in the embodiments.

Further, while the TOE module 212 is located at the input of the system 300, it will be apparent to those skilled in the art that different TOE algorithms may be used, and that some of these algorithms operate by examining the output of the system DFT206 to determine the time offset.

Fig. 9 is a flow diagram of a process 900 for wireless communication. At 902, a plurality of reference signal transmissions are received over a plurality of subcarriers. The subcarriers carrying the reference signals may be, for example, adjacent to or separated from each other and may be known a priori by the receiver. At 904, a plurality of coarse channel estimates may be computed based on the received plurality of reference signals. The calculation may be, for example, including calculating H_rawAs previously described. At 904, the plurality of coarse channel estimates may be processed to obtain an improved channel estimate. In thatIn some implementations, this processing may include filtering the coarse channel estimate using a linear filter (e.g., represented as a matrix multiplication in the discussion above). The processing may also include interpolating the plurality of coarse channel estimates using an interpolation filter (e.g., averaging), as previously described. In some implementations, the processing may depend on the received signal power and the received noise power, as previously described. In some implementations, the improved channel estimate may be obtained by processing 2 × B +1 subcarrier results, where B is an integer.

Fig. 10 is a schematic structural diagram of a part of the wireless communication apparatus 1000. A module 1002 (e.g., a receiver) is configured to receive multiple reference signal transmissions on multiple subcarriers. A module 1004 (e.g., a coarse channel estimator) is configured to calculate a plurality of coarse channel estimates based on the plurality of received reference signals. Module 1006 (e.g., an improved channel estimator) is configured to process the plurality of coarse channel estimates to obtain an improved channel estimate. Device 1000 and modules 1002, 1004, 1006 may be configured to implement one or more of the techniques described in this patent document.

It should be appreciated that various techniques for performing channel estimation in OFDM receive operations are disclosed. In one aspect, a plurality of coarse channel estimates obtained on adjacent subcarriers are filtered to obtain an improved channel estimate to help improve receiver performance. On the other hand, the coarse channel estimate may be filtered with a smoothing filter, wherein the smoothing filter may be represented as a matrix multiplication operation, the number of matrix multiplications being smaller than the length of the channel estimate.

The embodiments disclosed herein, as well as other embodiments and functional operations described herein, may be implemented in digital electronic circuitry, or in computer software, firmware, or hardware, including the structures disclosed herein and their structural equivalents, or in combinations of one or more of them. The disclosed embodiments and other embodiments may be implemented as one or more computer program products, i.e., one or more modules of computer program instructions encoded on a computer readable medium for execution by, or to control the operation of, data processing apparatus. The computer-readable medium can be a machine-readable storage device, a machine-readable storage substrate, a memory device, a combination of things that affect a machine-readable propagated signal, or a combination of one or more of them. The term "data processing apparatus" encompasses all devices, apparatus, and machines for processing data, including by way of example a program processor, a computer, or multiple processors or computers. In addition to hardware, the apparatus can include code that creates an execution environment for the computer program in use, e.g., code that constitutes processor firmware, a protocol stack, a data management system, an operating system, or a combination of one or more of them. A propagated signal is an artificially generated signal, e.g., a mechanically generated electrical, optical, or electromagnetic signal, that is generated to encode information for transmission to suitable receiver apparatus.

A computer program (also known as a program, software application, script, or code) can be written in any form of programming language, including compiled or interpreted languages, and it can be deployed in any form, including as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment. A computer program does not necessarily correspond to a file of a file system. A program can be stored in a portion of a file that holds other programs or data (e.g., one or more scripts stored in a markup language file), in a separate file dedicated to the program in use, in multiple coordinated files (e.g., files that store one or more modules, sub programs, or portions of code). A computer program can be deployed to be executed on one computer or on multiple computers at one site or distributed across multiple sites and interconnected by a communication network.

The processes and logic flows described herein can be performed by one or more programmable processors executing one or more computing programs to perform functions by operating on input data and generating output. The processes and logic flows can also be performed by, and apparatus can also be implemented as, special purpose logic circuitry, e.g., an FPGA (field programmable gate array) or an ASIC (application-specific integrated circuit).

Processors suitable for the execution of a computer program include, by way of example, both general and special purpose microprocessors, and any one or more processors of any kind of digital computer. Generally, a processor will receive instructions and data from a read-only memory or a random access memory or both. The essential elements of a computer are a processor for executing instructions and one or more memory devices for storing instructions and data. Generally, a computer will also include, or be selectively coupled to receive data from, or transmit data to, one or more mass storage devices, e.g., magnetic, magneto-optical disks, or optical disks. However, computers do not require these devices. Computer-readable media suitable for storing computer program instructions and data include all forms of non-volatile memory, media and memory devices, including by way of example semiconductor memory devices, e.g., EPROM, EEPROM, and flash memory devices; magnetic disks, such as internal hard disks or removable disks; magneto-optical disks; as well as CD ROM discs and DVD-ROM discs. The processor and the memory can be supplemented by, or included in, special purpose logic circuitry.

While this document contains many specifics, these should not be construed as limitations on the scope of the invention as claimed or as may be claimed, but rather as descriptions of features of particular embodiments. Certain features that are described herein in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of separate embodiments can also be implemented in separate embodiments or in any suitable subcombination. Further, although features may be described above as acting in certain combinations and even initially claimed as such, one or more features from a claimed combination can in some cases be excised from the combination, and the claimed combination may be directed to a subcombination or variation of a subcombination. Similarly, while operations are depicted in the drawings in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order, or that all illustrated operations be performed, to achieve desirable results.

Only a few examples and embodiments are disclosed herein. Variations, modifications, and improvements of the described examples and embodiments, as well as other embodiments, may be made based on the disclosure.

Claims (16)

1. A method of obtaining a channel estimate in a receiver, comprising:

receiving a baseband signal with a cyclic prefix, the baseband signal corresponding to a channel;

removing the cyclic prefix of the baseband signal;

transforming the baseband signal from a time domain to a frequency domain to produce a baseband signal in the frequency domain;

compensating the baseband signal in the frequency domain by a first time offset of the received baseband signal to produce a compensated baseband signal in the frequency domain;

extracting subcarriers used for transmitting the baseband signal from the compensated baseband signal in the frequency domain to generate a coarse channel estimate;

determining neighboring subcarriers having channel responses highly correlated to the subcarriers of the channel; wherein the B subcarriers preceding the subcarriers of the channel, and the B subcarriers following the subcarriers of the channel in total are considered a total of 2 × B +1 highly correlated subcarriers, where B is a positive integer;

generating a compensated coarse channel estimate by compensating the coarse channel estimate in the frequency domain by a second time offset, wherein the second time offset corresponds to a minimum angle of an autocorrelation function of the subcarrier and the neighboring subcarrier; and

filtering the compensated coarse channel estimate with a filter to produce a filtered channel estimate, wherein a filter length of the filter corresponds to a sum of the subcarrier and the neighboring subcarrier.

2. A method of obtaining filtered channel estimates in a receiver, comprising:

receiving a coarse channel estimate;

filtering the coarse channel estimate using a filter having a minimum determined filter length required to achieve a desired signal-to-noise ratio;

generating a compensated coarse channel estimate by time-shifting the coarse channel estimate in a frequency domain, wherein the time-shift corresponds to a minimum angle of an autocorrelation function of a subcarrier corresponding to the coarse channel estimate and a neighboring subcarrier having a channel response highly correlated to the subcarrier; the B subcarriers before, after, and considered as a total of 2 × B +1 highly correlated subcarriers, where B is a positive integer; and

filtering the compensated coarse channel estimate with a filter to produce the filtered channel estimate, wherein a filter length of the filter corresponds to a sum of the subcarrier and the neighboring subcarrier.

3. A receiver, comprising:

an antenna operable to receive a baseband signal having a cyclic prefix, the baseband signal corresponding to a channel; and

a processor in communication with the antenna, the processor operable to:

receiving a baseband signal with a cyclic prefix, the baseband signal corresponding to a channel;

removing the cyclic prefix of the baseband signal;

transforming the baseband signal from a time domain to a frequency domain to produce a baseband signal in the frequency domain;

compensating the baseband signal in the frequency domain by a first time offset of the received baseband signal to produce a compensated baseband signal in the frequency domain;

extracting subcarriers used for transmitting the baseband signal from the compensated baseband signal in the frequency domain to generate a coarse channel estimate;

determining neighboring subcarriers having channel responses highly correlated to the subcarriers of the channel; wherein the B subcarriers before the subcarriers of the channel, and the B subcarriers after the subcarriers of the channel are considered a total of 2 × B +1 highly correlated subcarriers, where B is a positive integer;

generating a compensated coarse channel estimate by compensating the coarse channel estimate in the frequency domain by a second time offset, wherein the second time offset corresponds to a minimum angle of an autocorrelation function of the subcarrier and the neighboring subcarrier; and

a filter operable to filter the compensated coarse channel estimate to produce a filtered channel estimate, wherein a filter length of the filter corresponds to an aggregate of the subcarrier and the neighboring subcarrier.

4. A method of channel estimation, comprising:

receiving a plurality of reference signal transmissions on a plurality of subcarriers;

calculating a plurality of coarse channel estimates based on the plurality of received reference signals; and

processing the plurality of coarse channel estimates to obtain an improved channel estimate;

wherein the plurality of subcarriers comprises 2 × B +1 subcarriers having highly correlated channel responses, where B is a positive integer;

wherein processing the plurality of coarse channel estimates to obtain an improved channel estimate comprises:

filtering the coarse channel estimate using a filter having a minimum determined filter length required to achieve a desired signal-to-noise ratio;

generating a compensated coarse channel estimate by time-shifting the coarse channel estimate in a frequency domain, wherein the time-shift corresponds to a minimum angle of an autocorrelation function of a subcarrier corresponding to the coarse channel estimate and a neighboring subcarrier having a channel response highly correlated to the subcarrier; the B subcarriers before, after, and considered as a total of 2 × B +1 highly correlated subcarriers, where B is a positive integer; and

filtering the compensated coarse channel estimate with a filter to produce the filtered channel estimate, wherein a filter length of the filter corresponds to a sum of the subcarrier and the neighboring subcarrier.

5. The method of claim 4, wherein the processing operation comprises:

filtering the plurality of coarse channel estimates using a linear filter; and

interpolating the plurality of coarse channel estimates using an interpolation filter.

6. The method of claim 5, wherein the interpolation filter implements linear interpolation.

7. The method of claim 4, wherein the processing operation is dependent on received signal power and received noise power.

8. A channel estimation system, comprising:

a receiver for receiving a plurality of reference signal transmissions on a plurality of subcarriers;

a coarse channel estimator for calculating a plurality of coarse channel estimates based on the received plurality of reference signals; and

an improved channel estimator for processing the plurality of coarse channel estimates to obtain an improved channel estimate;

wherein the plurality of subcarriers comprises 2 × B +1 subcarriers having highly correlated channel responses, where B is a positive integer;

wherein processing the plurality of coarse channel estimates to obtain an improved channel estimate comprises:

filtering the coarse channel estimate using a filter having a minimum determined filter length required to achieve a desired signal-to-noise ratio;

generating a compensated coarse channel estimate by time-shifting the coarse channel estimate in a frequency domain, wherein the time-shift corresponds to a minimum angle of an autocorrelation function of a subcarrier corresponding to the coarse channel estimate and a neighboring subcarrier having a channel response highly correlated to the subcarrier; the B subcarriers before, after, and considered as a total of 2 × B +1 highly correlated subcarriers, where B is a positive integer; and

filtering the compensated coarse channel estimate with a filter to produce the filtered channel estimate, wherein a filter length of the filter corresponds to a sum of the subcarrier and the neighboring subcarrier.

9. The system of claim 8, wherein the processing operation comprises:

a channel estimation filter for filtering the plurality of coarse channel estimates using a linear filter; and

an interpolator to interpolate the plurality of coarse channel estimates using an interpolation filter.

10. The system of claim 9, wherein the interpolation filter implements linear interpolation.

11. The system of claim 8 wherein the improved channel estimator is responsive to received signal power and received noise power.

12. A channel estimation device, comprising:

means for receiving, on a plurality of subcarriers, a plurality of reference signal transmissions;

means for calculating a plurality of coarse channel estimates based on the plurality of received reference signals; and

means for processing the plurality of coarse channel estimates to obtain an improved channel estimate;

wherein the plurality of subcarriers comprises 2 × B +1 subcarriers having highly correlated channel responses, where B is a positive integer;

wherein processing the plurality of coarse channel estimates to obtain an improved channel estimate comprises:

filtering the coarse channel estimate using a filter having a minimum determined filter length required to achieve a desired signal-to-noise ratio;

generating a compensated coarse channel estimate by time-shifting the coarse channel estimate in a frequency domain, wherein the time-shift corresponds to a minimum angle of an autocorrelation function of a subcarrier corresponding to the coarse channel estimate and a neighboring subcarrier having a channel response highly correlated to the subcarrier; the B subcarriers before, after, and considered as a total of 2 × B +1 highly correlated subcarriers, where B is a positive integer; and

filtering the compensated coarse channel estimate with a filter to produce the filtered channel estimate, wherein a filter length of the filter corresponds to a sum of the subcarrier and the neighboring subcarrier.

13. The apparatus of claim 12, wherein the means for processing comprises:

means for filtering, using a linear filter, the plurality of coarse channel estimates; and

means for interpolating the plurality of coarse channel estimates using an interpolation filter.

14. The apparatus of claim 13, wherein the means for interpolating comprises means for linear interpolation.

15. The apparatus of claim 12, wherein the means for processing is responsive to a received signal power and a received noise power.

16. An Orthogonal Frequency Division Multiplexing (OFDM) -based communication system, comprising:

a network of base stations spatially distributed within a service area to form a radio access network for wireless communication devices,

wherein each base station comprises a receiver comprising an antenna, a filter, and a processor in communication with the antenna, the antenna operable to receive a baseband signal having a cyclic prefix, the baseband signal corresponding to a channel; and

wherein the processor is configured to:

receiving a baseband signal with a cyclic prefix, the baseband signal corresponding to a channel;

removing the cyclic prefix of the baseband signal;

transforming the baseband signal from a time domain to a frequency domain to produce a baseband signal in the frequency domain;

compensating the baseband signal in the frequency domain by a first time offset of the received baseband signal to produce a compensated baseband signal in the frequency domain;

extracting subcarriers used for transmitting the baseband signal from the compensated baseband signal in the frequency domain to generate a coarse channel estimate;

determining neighboring subcarriers having channel responses highly correlated to the subcarriers of the channel; wherein the B subcarriers before the subcarriers of the channel, and the B subcarriers after the subcarriers of the channel are considered a total of 2 × B +1 highly correlated subcarriers, where B is a positive integer;

generating a compensated coarse channel estimate by compensating the coarse channel estimate in the frequency domain by a second time offset, wherein the second time offset corresponds to a minimum angle of an autocorrelation function of the subcarrier and the neighboring subcarrier; and

wherein the filter is operable to filter the compensated coarse channel estimate to produce a filtered channel estimate, wherein a filter length of the filter corresponds to a sum of the subcarrier and the neighboring subcarrier.