METHOD FOR MITIGATING INTERFERENCE IN OFDM COMMUNICATIONS SYSTEMS

Abstract

The method for mitigating interference in Orthogonal Frequency Division Multiplexing (OFDM) communications systems uses comb-type pilot signals in a doubly selective channel where the pilot signals are calculated to have zero correlation sequences in order to maintain orthogonality where there is a high Doppler effect to mitigate intercarrier interference. The pilot signals may be precomputed, stored in a look-up table, and selected based upon total power constraints of the pilot signals. The steps of the method may be performed by a Finite Impulse Response (FIR) equalizer.

Description

BACKGROUND OF THE INVENTION

1. Field of the Invention

The present invention relates to Orthogonal Frequency Division Multiplexing (OFDM) modulation methods for wireless communications systems, and more particularly to a method for mitigating interference in OFDM communications systems that provides a modified pilot signal design to mitigate interference in high Doppler effect (mobile) environments.

2. Description of the Related Art

An OFDM signal includes a number of independently modulated, mutually orthogonal subcarriers over which large number of signals can be transmitted over a similar time period. This allows for very effective use of the spectrum with high bandwidth efficiency. Many digital communication standards have selected OFDM as their modulation scheme of choice. Because OFDM is known to be one of the most efficient modulation methods available, it is desirable to use OFDM widely, including in a mobile environment. Such mobile environments often have high Doppler effect shifting of frequencies because of the relative speed differences between communicating transmitters and receivers. High data rate communication systems inherently suffer from frequency selectivity, which causes Inter Symbol Interference (ISI) in a high Doppler environment.

To mitigate the effect of ISI, OFDM systems use a Cyclic Prefix (CP) of length greater than or equal to the channel length. At the receiver, this CP is discarded to recover the ISI-free OFDM symbol. The CP decouples the subcarriers of the OFDM symbol, making it possible to use a single-tap equalizer at the receiver, thus simplifying the receiver design.

This ability of OFDM to allow high-speed data transmission has led to its adaptation in many broadband standards, including Digital Audio and Video Broadcasting (DAB, DVB), wireless local area network (WLAN) standards (e.g., IEEE 802.11a/b/g and HIPERLAN/2), high-speed transmission over digital subscriber lines (DSL) and the Digital Terrestrial/Television Multimedia Broadcasting (DTMB) standard. A number of emerging broadband wireless communication standards are using or planning to use OFDM modulation including 802.16 (WiMAX), 802.20 Mobile Wireless Broadband Access (MWBA) and other emerging cellular wireless communication systems like 3GPP evolution and 4G.

OFDM systems depend upon orthogonality of the subcarriers. If orthogonality is lost, the information on one subcarrier is leaked into other adjacent subcarriers, i.e., the subcarriers are no longer decoupled. This leakage is termed inter-carrier interference (ICI). There are three main contributing factors to ICI: namely, phase noise, frequency error, and Doppler shift. In practice, the effect of phase noise and frequency error can be minimized by proper receiver design, and thus these two factors do not amount for a large ICI component in well-designed systems. Doppler shift appears due to the relative motion of the transmitter and receiver and is the main cause of ICI, especially in mobile wireless environments where the channel is continuously changing with time. Under such conditions, maintaining the orthogonality of OFDM subcarriers is a challenge, particularly if the time variation is large.

A direct consequence of operating OFDM systems in a high-Doppler environment is the introduction of ICI. The effect of ICI is that the subcarriers are no longer orthogonal. This results in energy leakage from a subcarrier to its neighbors, i.e., the subcarriers are no longer decoupled. The presence of ICI makes the task of channel estimation more challenging, since now both the frequency response at the subcarrier and also the interference caused by its neighbors have to be estimated.

Recently, a frequency domain high-performance computationally efficient OFDM channel estimation algorithm in the presence of severe ICI was proposed by the inventors in U.S. patent application Ser. No. 12/588,585.

Many OFDM-based systems use pilot subcarriers for channel estimation and tracking. For fixed and slowly varying channels, the optimum pilot pattern consists of equally spaced individual pilot subcarriers. However, the same pilot patterns are no longer optimal when ICI is introduced in a high Doppler environment.

Thus, a method for mitigating interference in OFDM communications systems solving the aforementioned problems is desired.

SUMMARY OF THE INVENTION

The method for mitigating interference in OFDM communications systems provides Minimum Mean Square Error (MMSE) optimal pilot structures when a channel is fast varying, such as in high Doppler environments, in order to mitigate Inter-Carrier Interference (ICI) in Orthogonal Frequency Division Multiplexing (OFDM) systems (in the present application, OFDM is used generically to refer to both Orthogonal Frequency Division Multiplexing, a term typically used for wired systems, and to Orthogonal Frequency Division Multiple Access [OFDMA], a term typically applied to wireless systems). For a fixed number of pilot subcarriers and fixed total power constraints, optimum pilot patterns are identified.

The method provides MMSE-optimum pilot design for OFDM systems where the optimum design may include identical, equally spaced frequency domain pilot clusters. According to the method, Zero Correlation Zone (ZCZ) sequences are shown to be the MMSE optimal designs for frequency domain pilot clusters. The method provides an arrangement of pilot subcarriers of the OFDM symbol such that the receiver is able to detect the pilot subcarriers independent of data subcarriers. The method provides for MMSE-optimum pilot design for OFDM systems that can be used in conjunction with any frequency domain ICI cancellation method, and particularly with the method disclosed in the parent U.S. patent application Ser. No. 12/588,585.

These and other features of the present invention will become readily apparent upon further review of the following specification and drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a chart showing four cases (permutations) of R_E based on the circulant approximation of J depicting combinations of the off diagonal elements of R_E.

FIG. 2 is a diagram depicting the frequency domain minimum mean square error (MMSE) optimum pilot structure and its corresponding time domain representation.

FIG. 3 is a chart showing optimum pilot clusters of length 5 that are also Zero Correlation Zone (ZCZ) sequences.

FIG. 4 is a graph depicting a simulated bit error rate (BER) for a signal generated by the method of mitigating interference in OFDM communications systems according to the present invention compared with the perfect Channel State Information (CSI) case and impulsive pilot clusters for N=1024, M=3× and N_p=5.

FIG. 5 is a block diagram of an exemplary OFDM system implementing the method of mitigating interference in OFDM communications systems according to the present invention.

FIG. 6 shows a step by step flowchart of the operations performed at the transmitter to select the optimal ZCZ sequence based on system parameters in a method of mitigating interference in OFDM communications systems according to the present invention.

Similar reference characters denote corresponding features consistently throughout the attached drawings.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

The method presented implements a novel algorithm that mitigates the effects of ICI on signals transmitted in an OFDM system. The frequency domain representation of an OFDM system with N subcarriers is given as:

y QHQ ^H X+Z=GX+Z  (1)

where means “is defined as”, Q is the N-point Fast Fourier Transform (FFT) matrix and (.)^H is the Hermitian operator, X is a pilot data multiplexed OFDM symbol where certain subcarriers are allocated as pilots surrounded by data subcarriers. We refer to such a multiplexed OFDM symbol structure as a comb-type OFDM symbol. H is the N×N time domain channel matrix, which corresponds to convolution with the time-varying Channel Impulse Response (CIR) coefficients h_n(l) at lag l (for 0≦l≦L−1) and time instant n, and Z is the frequency domain noise vector.

The Channel Frequency Response (CFR) matrix is not diagonal over doubly selective channels. Rather, the energy of the main diagonal spills into adjacent diagonals. The extent of this spilling depends on the severity of the Doppler spread. We approximate G as a banded matrix and set all elements of G outside of M main diagonals as zero where M is an odd integer.

In the following, assuming Jakes's model with E[h_m(l)h*_n(l)]=J₀(2πf_d(m−n)T_s)J(m−n), where f_d is the Doppler frequency and J₀(.) is the zeroth order Bessel function of the first kind; define R_G=E[vec(G)vec(G)^H], so that the eigenvalue decomposition of R_G is given in terms of an N×N symmetric Toeplitz Bessel function matrix J whose (m, n)-th element is given by J(m, n)=J(|m−n|)=J₀(2πf_d|m−n|T_s); and let G_p denote the matrices formed by unvectorizing the NL eigenvectors of R_G. Then G_p can be expressed in terms of eigenvectors off as:

G _p =Qdiag(υ_n)B^lQ^H  (2)

where 0≦l≦(L−1), 1≦p≦NL, υ_n are the dominant eigenvectors of J for n=1, 2, . . . , N and B is a circulant shift matrix whose first column is [0 1 0 . . . 0]^T.

Considering the N_dL dominant eigenvectors of R_G, equation (1) can be approximated as:

y=GX+Z≈Σ _p=1 ^{N d L}α_pG_pX+Z=Σ_p=1^NdLα_pεpZ  (3)

where α_p's are the unknown independent variables. Considering only those output carriers that are free of interference from data carriers, we get T input-output equations in N_dL unknowns:

y=Σ _p=1 ^{N d L}α_pε_p+Z=E_pα+Z  (4)

where E_p=[ε₁ . . . ε_NdL] and α=[α₁ . . . α_NdL]^T with zero mean and covariance matrix R_α=diag([γ₁λ₁, . . . , γ_NdLλ_NdL]); γ₁ and λ₁ being the channel Power Delay Profile (PDP) path variances and the dominant eigenvalues of R_G, respectively.

Thus, α can be estimated by the following Linear Minimum Mean Square Error (LMMSE) estimator:

$\begin{array}{cc}\hat{\alpha }={\frac{1}{{\sigma }_Z^2}\left[R_{\alpha }^{-1}+\frac{1}{{\sigma }_Z^2}{\underset{\_}{E}}_p^H{\underset{\_}{E}}_p\right]}^{-1}{\underset{\_}{E}}_p^H\underset{\_}{y}=w\underset{\_}{y}& \left(5\right)\end{array}$

where σ_Z² is the noise variance. W can be pre-computed and stored in lookup tables to reduce real-time computational complexity significantly, given N, f_D, σ_Z² and PDP. The look up table will contain precomputed stored entries of W for various possible values of N, f_D, σ_Z². For a given system, the FFT size N is known a priori, while the Doppler frequency f_D and noise variance σ_Z² a can be estimated at run-time by the receiver using any of the various techniques available in literature. With the knowledge of these three parameters, the closest corresponding entry of W can be selected from the table based on a simple comparison rule. The error vector ε=α−{circumflex over (α)} has a zero mean and covariance matrix expressed by:

$\begin{array}{cc}{C}_{\epsilon }={\left[R_{\alpha }^{-1}+\frac{1}{{\sigma }_Z^2}{\underset{\_}{E}}_p^H{\underset{\_}{E}}_p\right]}^{-1}={\left[R_{\alpha }^{-1}+\frac{1}{{\sigma }_Z^2}R_E\right]}^{-1},& \left(6\right)\end{array}$

and is a measure of the performance of the LMMSE estimator.

Described herein is the design of a frequency domain pilot structure for an LMMSE channel estimator that minimizes the covariance matrix C_ε given in equation (6). This can be achieved by making the matrix R_E a diagonal matrix. From (3), (4) and (6), making R_E diagonal is equivalent to designing X such that:

X^HG_i^HG_jX=0  (7)

for i≠j and i, j=1, 2, . . . , N_dL and X^HX=c, where c is a constant that depends on the total pilot power. The (m, n)-th element of R_E can be written as follows:

$\begin{array}{cc}\begin{array}{c}{R}_E\left(m,n\right)=X^HG_{m}^HG_nX\\ =X^H{\mathrm{QB}}^{H\left(j1\right)}{\mathrm{diag}\left(v_{i1}\right)}^H\mathrm{diag}\left(v_{i2}\right)B^{j2}Q^HX\\ =x^HB^{H\left(j1\right)}{\Lambda }_{i_1i_2}B^{j2}x\\ =x^HI_c\left(i_1,i_2,j_1,j_2\right)x\end{array}& \left(8\right)\end{array}$

where m=(i₁−1)N_d+j₁, n=(i₂−1)N_d+j₂ for i₁, i₂=1,2, . . . , N_d and j₁, j₂=1,2, . . . , L. Based on the circulant approximation of J, there are four possible values of R_E, as shown in chart 100 of FIG. 1. These four values correspond to Case 1, designated as 102 in the drawing, Case 2, designated as 104, Case 3, designated as 106, and Case 4, designated as 108. So long as j₁=j₂, then I_c(i₁, i₂, j₁, j₂) is a diagonal matrix whose entries are real if i₁=i₂, or complex otherwise. If j₁≠j₂, then i_c(i₁, i₂, j₁, j₂) has zero diagonal elements and a non-zero d_j-th super-diagonal or sub-diagonal where j₁−j₂=+d_j or −d_j, respectively.

The frequency-domain pilot vector X has the periodic, clustered structure shown in diagram 200 of FIG. 2, with N_p adjacent subcarriers 202 in each pilot cluster 204, and where the number N_c and period L_c of the pilot clusters 204 satisfy the relation N=N_cL_c. Then, the time-domain pilot vector x will be sparse, as shown in the diagram 200 of FIG. 2. The term x^HI_c(i₁, i₂, j₁, j₂)x will be a weighted sum of the elements of) that correspond to the positions of ‘1’-s in the puncturing matrix P_Ic whose (m, n)-th element is given by:

$\begin{array}{cc}{P}_{I_c}\left(m,n\right)=\{\begin{array}{cc}1,& m=k_1N_c+1,n=k_2N_c+1\\ 0,& \mathrm{otherwise}\end{array}& \left(9\right)\end{array}$

where k₁, k₂=0, 1, . . . , (L_c−1) and (L−1) is the highest index of the super- or sub-diagonal of I_c(.) that is non zero. This means that the I_c(.)'s in Case 2 and Case 4 of FIG. 1 can be, at most, (L−1)-shifted upper or lower diagonal matrices. If the sparse x has at least L zeros between any two of its non-zero elements, all Case 2 and Case 4 off-diagonal elements of R_E will be zero. This requires the number of pilot clusters 204 in the frequency domain to be greater than the length of the channel impulse response vector, i.e.,

N_c>L  (10).

In addition, the pilot cluster size must satisfy:

M≦N _p≦2M−1;M=3,5,  (11)

where M is the number of main diagonals. The periodic structure of the clustered pilots 204 implies that the pilots must be equally spaced. The period of the pilot clusters 204 is given by:

$\begin{array}{cc}{L}_c=\frac{{\mathrm{NN}}_p}{N_T}& \left(12\right)\end{array}$

where N_T=N_cN_p is the total number of pilot subcarriers 202. The first and last

$\frac{M-1}{2}$

subcarriers 202 of the comb-type OFDM symbol are assigned zeros, as they cannot be assigned as pilots, so that all M main diagonals of G_p are included in the input output equations. Inserting N_p adjacent pilot subcarriers 202 following

$L_c-\left(N_p-\frac{M-1}{2}\right)$

zeros implies the following lower bound:

L _c≧(N_p+M+1)  (13),

and for N_c

$\begin{array}{cc}\max \left(\frac{N_T}{2M-1},L\right)\le N_c\le \frac{N_T}{M}& \left(14\right)\end{array}$

As all pilot clusters 204 are identical due to the periodic nature of X, only one pilot cluster 204 needs to be optimized.

Case 3 of FIG. 1, where i₁≠i₂ and j₁=j₂, I_c(i₁, i₂, j₁, j₂), becomes a diagonal matrix whose diagonal is a scaled, circularly shifted Fast Fourier Transform (FFT) vector. Then let the vector α_i contain these modified FFT vectors, where i=(−(N_d−1), . . . , −1, 1, . . . , (N_d−1))_N denotes the FFT column index and (.)_N is the modulo N operation. To reduce the computational complexity, N_d dominant eigenvectors of J are chosen. For each dominant eigenvector, we have (N_d−1) Case 3 off-diagonal elements, resulting in a total of (N_d−1)N_d non-diagonal elements in R_E to be forced to zero. However, due to the conjugate symmetry of the FFT vectors, we have only 2 (N_d−1) distinct non-diagonal elements in R_E to be forced to zero.

The time domain sparse vector x is given by:

x=Q^HĨX_p  (15)

where X_p is a single frequency-domain pilot cluster 204 of length N_p, and Ĩ=1_NcĨ_p where

${\tilde{I}}_p={\left[0_{N_p\times \frac{M-1}{2}}I_{N_{p}}0_{N_p\times \left(L_c-\frac{M-1}{2}N_p\right)}\right]}^T,$

1 _{N c} is the length-N_c all ones column vector, and denotes the Kronecker product.

Using the sparse structure of x, as shown in FIG. 1, and from equations (8) and (15), the pilot optimization objective can be stated as:

$\begin{array}{cc}{x}^HI_c\left(i_1,i_2,j_1,j_2\right)x=X_p^H{\tilde{I}}^H{\mathrm{QB}}^{H\left(j1\right)}{\Lambda }_{i_1i_2}B^{\left(j1\right)}Q^H\tilde{I}X_p=0->X_p^H{\tilde{I}}^HQ\mathrm{diag}\left(a_i\right)Q^H\tilde{I}X_p\stackrel{\Delta }{=}X_p^HR_iX_p=0& \left(16\right)\end{array}$

where i=(−(N_d−1), . . . , −1, 1, . . . , (N_d−1))_N. Since all α_i's are scaled, circularly shifted FFT vectors, Qdiag(α_i)Q^H can be written as c_iZ_cⁱ, where c_i is a complex scalar and Z_c is the N×N circular upper shift matrix whose first column is [0 . . . 0 1]^T. The (m, n)-th element of R_i is a weighted sum of the elements of c_iZ_cⁱ that correspond to the positions of ‘1’-s in the puncturing matrix P_Ri(m,n) given by:

$\begin{array}{cc}{P}_{R_i\left(m,n\right)}\left(q,r\right)=\{\begin{array}{cc}1,& q=k_3N_c+m,r=k+4N_{zc}+n\\ 0,& \mathrm{otherwise}\end{array}& \left(17\right)\end{array}$

where k₃, k₄=0, 1, . . . , (L_C−1).

Let d_i denote i, as in equation (16). When d_i is negative, it can be shown that R_i=c′_iZ_u^di, where c′_i is a complex scalar and Z_u is the N×N linear upper-shift matrix whose first row is [0 1 0 . . . 0]. On the other hand, if d_i is positive, R_i=c′_iZ_l^di, where Z_l is a linear lower-shift matrix whose first column is [0 1 0 . . . 0]^T. There are (N_d−1) distinct R_i's associated with Z_u, and (N_d−1) distinct R_i's associated with Z_l. Thus, equation (16) can be written as:

X _p ^H Z _u ^{d i} X _p=0:d_i=1,2, . . . ,(N_d−1)  (18)

X _p ^H Z _l ^{d i} X _p=0:d_i=1,2, . . . ,(N_d−1)  (19)

For each d_i, equations (18) and (19) can be rewritten using the frequency domain pilot cluster notation as:

Σ_n=1+τ^NpP_nP*_n−τ=0:τ=1,2, . . . ,(N_d−1)  (20)

(Σ_n=1τ^NpP_nP*_n−τ=0)*:τ=1,2, . . . ,(N_d−1)  (21)

The same solution satisfies both equations (20) and (21). It can be seen from equation (18) that at lag d_i, the aperiodic auto-correlation of the optimum pilot cluster sequence 204 must be zero, which is same as the design criterion of a Zero Correlation Zone (ZCZ) sequence with Z_CN_d−1 zero lags. For N_p=5 and N_d=3, the MMSE-optimal pilot cluster 204 is a ZCZ sequence of length 5 with Z_c=2. FIG. 3 shows a chart 300 with three such ZCZ sequences obtained through numerical search under total power constraint. The optimization process can be performed offline, and the optimum pilot sequences of different sizes can be stored in a look-up table for faster real-time implementation. The look up table is a pre-computed table listing various possible ZCZ sequences for various values of N_p, N_d and Z_c satisfying equations (19-22), For a given system, N_p is known at the transmitter, N_d is also known at the transmitter, either a priori or calculated based on the severity of Doppler shift estimated at the receiver and conveyed to the transmitter through a feedback mechanism, while Z_c is easily computable from N_d. Based on these values, the receiver can choose the corresponding set of ZCZ sequences.

FIG. 4 is a graph 400 showing results for a system operating in high Doppler frequency of 10% (normalized to the subcarrier spacing) with ½ rate convolution code, N=1024, and M=3. The channel model is taken as SUI-3. Assuming a pilot cluster size of N_p=2M−1=5, FIG. 4 shows the Bit Error Rate (BER) of the perfect CSI under full and banded G assumptions, along with the BER obtained by the algorithm proposed in our prior U.S. patent application Ser. No. 12/588,585, filed Oct. 9, 2009, using the presented optimum pilot clusters 204. All three optimized pilot clusters 204 satisfy the MMSE criterion. However, their performance differs at high signal-to-noise ratio (SNR), where ICI dominates noise. It can be seen that sequence ‘a’, given in FIG. 3, has a higher aperiodic auto-correlation at lags larger than Z_c, and performs worse in a high SNR scenario as compared to sequences ‘b’ and ‘c’. As a benchmark, the BER of the impulsive pilot cluster, as proposed by A. Kannu et. al. (“Design and analysis of MMSE pilot-aided cyclic-prefixed block transmissions for doubly selective channels,” IEEE Transactions on Signal Processing, vol. 56, no. 3, pp. 1148-1160, March 2008), suffers from an irreducible error floor.

FIG. 5 is a block diagram of an exemplary OFDM communications system 500 according to the present invention. System parameters 502 are input into a look-up table 504 The look up table 504 will contain precomputed stored entries of W for various possible values of N, f_D, σ_Z². The best fitting Zero Correlation Zone (ZCZ) sequence 506 is selected as described herein and transmitted to a Pilot Insertion Module 508. Data 510 for wireless transmission is modulated by a modulator 512 and is also transmitted to the Pilot Insertion Module 508. The Pilot Insertion Module 508 combines the selected ZCZ pilot signals 506 with modulated data from the modulator 508 and transmits the resultant signals to an Inverse Fast Fourier Transform (IFFT) module 514 to convert them from frequency domain to time domain. The IFFT 514 output is transmitted to a Cyclic Prefix (CP) Insertion Module 516 for wireless transmission through a transmission antenna arrangement 518.

The CP is received from through a receiver antenna arrangement 520 by a CP Removal Module 522 and forwarded to a Fast Fourier Transform (FFT) Module 524 to convert from time domain to frequency domain. The FFT Module 524 performs fast Fourier transforms on the incoming signals and transmits the resultant signals to an Inter-Channel Interference (ICI) Estimation Unit 526 and a Doppler Estimation Unit 528. Output from the Doppler Estimation Unit 528 is used to select an entry in Lookup Table 530. Output from the Lookup Table 530 is also sent to the ICI Estimation Unit 526. Output from the ICI Estimation Unit 526 is sent to an Equalizer 532 and then to a Pilot Removal Module 534 that separates the pilot signals from the modulated data. The modulated data is sent to a demodulator 536 to reproduce the demodulated data for the receiver.

FIG. 6 is a flowchart showing steps associated with FIG. 5, in particular, those steps associated operations performed at the transmitter to select an optimal ZCZ sequence based on system parameters 600. At step 602 system parameters are obtained either through known design values or are updated by a feedback mechanism. In step 604, the system parameters are input into a look-up table to select the best fitting Zero Correlation Zone (ZCZ) sequence. In step 606, ZCZ sequence pilots are inserted on the pilot tones and modulated data on the data tones. The resultant signals from step 606 are converted to time domain symbol(s) by an IFFT operation in step 608 and inserted as a cyclic prefix in step 610. The signals are then transmitted wirelessly in step 612.

The method may be implemented in modulation or multiplexer circuits in a discrete transmitter, a discrete receiver, or a transceiver. The circuits may utilize one or more microprocessors, digital signal processors, application specific integrated circuits (ASICs), or other components programmed or configured to implement the steps of the method according to conventional construction techniques. In each of the embodiments, the various actions could be performed by program instruction running on one or more processors, by specialized circuitry or by a combination of both. Moreover, the method can additionally be considered to be embodied, entirely or partially, within any form of computer readable medium containing instructions that will cause the executing device to carry out the technique disclosed herein.

It is to be understood that the present invention is not limited to the embodiment described above, but encompasses any and all embodiments within the scope of the following claims.

Claims

1. A method for mitigating interference in an Orthogonal Frequency Division Multiplexing (OFDM) communications system, comprising the steps of: transmitting a plurality of data subcarriers, at least some of said plurality of data subcarriers representing an OFDM symbol, the OFDM symbol having data bits; andtransmitting a plurality of pilot subcarriers, the plurality of pilot subcarriers including pilot signals, the pilot signals being selected by a minimum mean square error (MMSE) estimator to form a pilot signal pattern.

2. The method of claim 1, wherein said MMSE estimator is a linear minimum mean square error (LMMSE) estimator.

3. The method of claim 1, further comprising the step of transmitting equally spaced frequency domain pilot clusters, the equally spaced frequency domain pilot clusters comprising said pilot signals.

4. The method of claim 3, wherein said pilot clusters further comprise adjacent subcarriers.

5. The method of claim 3, wherein said pilot clusters further comprise comb-type clusters, said comb-type clusters comprise zero correlation zone (ZCZ) sequences.

6. The method of claim 5, further comprising the steps of: precomputing said pilot clusters;storing said precomputed pilot clusters in a look-up table; andselecting one of said pilot clusters for the OFDM symbol to be transmitted based on total power constraints for the pilot signals,

7. The method of claim 1, wherein said pilot signals are inserted in the frequency domain.

8. The method of claim 1, wherein inherent constraints of the OFDM communications system are utilized to find an optimum pilot signal pattern.

9. The method of claim 1, wherein total power of the pilot signals are constrained.

10. The method of claim 1, wherein interference from adjacent data subcarriers is minimized in corresponding pilot subcarriers and available to a receiver for inter-channel interference (ICI) reduction.

11. The method of claim 1, wherein said pilot signals are inserted in the frequency domain.

12. A method for mitigating interference in an Orthogonal Frequency Division Multiplexing (OFDM) communications system, comprising the steps of: transmitting a plurality of data subcarriers, at least one of the plurality of data subcarriers representing an OFDM symbol, the OFDM symbol comprising data bits; andtransmitting a plurality of pilot subcarriers, the plurality of pilot subcarriers including pilot signals, the pilot signals being selected by a minimum mean square error (MMSE) estimator to form a pilot signal pattern.

13. The method of claim 12, wherein said MMSE estimator is a linear minimum mean square error (LMMSE) estimator.

14. The method of claim 12, further comprising the step of transmitting equally spaced frequency-domain pilot clusters, the equally spaced frequency-domain pilot clusters comprising said pilot signals.

15. The method of claim 14, wherein said pilot clusters further comprise adjacent subcarriers.

16. The pilot signal pattern of claim 14, wherein said pilot clusters further comprise comb-type clusters, at least one of said comb-type clusters being a zero correlation zone (ZCZ) sequence.

17. A method for mitigating interference in an Orthogonal Frequency Division Multiplexing (OFDM) communications system, comprising the steps of: receiving a plurality of data subcarriers, at least one of the plurality of data subcarriers representing an OFDM symbol, the OFDM symbol comprising data bits; andreceiving a plurality of pilot subcarriers, the plurality of pilot subcarriers including pilot signals, the pilot signals being selected by a minimum mean square error (MMSE) estimator to form a pilot signal pattern.

18. The method of claim 17, wherein said MMSE estimator is a linear minimum mean square error (LMMSE) estimator.

19. The method of claim 17, further comprising the step of receiving equally spaced frequency-domain pilot clusters, the equally spaced frequency-domain pilot clusters comprising said pilot signals.

20. The method of claim 19, wherein said pilot clusters further comprise comb-type clusters, at least one of the comb-type clusters being a zero correlation zone (ZCZ) sequence.