WO2006004980A9

WO2006004980A9 - Method for pulse shape design for ofdm

Info

Publication number: WO2006004980A9
Application number: PCT/US2005/023424
Authority: WO
Inventors: Arogyaswami Paulraj; Thomas Strohmer
Original assignee: Univ The Board Of Trustees Of; Univ California; Arogyaswami Paulraj; Thomas Strohmer
Priority date: 2004-06-28
Filing date: 2005-06-28
Publication date: 2006-03-16
Also published as: US7656786B2; WO2006004980A1; US20060039270A1

Abstract

A computationally efficient pulse shaping method for OFDM that produces mutually orthogonal transmission pulses having fast spectral decay is provided. The pulse shaping method comprises an iterative method for designing OFDM transmission pulses that satisfy prescribed time-frequency localization conditions. The iterative method may be implemented in a computationally efficient way and can be used to adapt the transmission pulses to time-varying channel conditions in real-time, thereby maximizing the bit-error performance of an OFDM system while maintaining high data rates in wireless transmission.

Description

METHOD FOR PULSE SHAPE DESIGN FOR OFDM

CROSS REFERENCE TO RELATED APPLICATIONS

This application claims priority to U.S. Provisional Application No. 60/583,741, entitled "Method for Pulse Shape Design For OFDM" and filed on June 28, 2004, the entire disclosure of which is incorporated herein by reference.

FIELD OF THE INVENTION

This invention relates generally to the wireless transmission ana broadcasting of digital or analog data. More specifically, the invention provides a method of pulse shaping for data transmission in an OFDM-based communications system.

BACKGROUND OF THE INVENTION

Advances in communications and digital multimedia technologies have fueled the demand for faster and more efficient wireless transmission of voice, data, and video on a global basis. Wireless transmission of information is accomplished by sending an information signal from a source to a destination via a wireless channel, which is a band of frequencies in the electromagnetic spectrum capable of serving as a transmission medium and carrying an information signal.

Typically, the usable range of the electromagnetic spectrum allocated for wireless channels on any communication system is limited. For wireless phone systems, for example, the total bandwidth allocated is typically on the order of 50 MHz. As a result, multicarrier communication techniques and multiple access communication techniques capable of sharing and allocating the spectrum efficiently among many channels

(multicarrier techniques) and many users (multiple access techniques) are required.

Multicarrier techniques are used in wireless communication systems to improve the bandwidth efficiency and reduce the intersymbol interference of the system. These techniques divide the usable spectrum among many channels and transmit data with multiple carriers. Multiple access techniques are used to increase the number of users that may access the wireless services provided by the system at any given time. Multicarrier and multiple access techniques can also be combined to serve many users simultaneously and provide them with the bandwidth efficiency and wireless services they so desire.

An example of a multicarrier technique include Orthogonal Frequency Division Multiplexing ("OFDM") . Examples of multiple access techniques include Frequency Division Multiple Access ("FDMA"), Time Division Multiple Access ("TDMA"), and Code Division Multiple Access ("CDMA") and its many variations. Examples of combined multicarrier/multiple access techniques include Orthogonal Frequency Division Multiple Access ("OFDMA"), Multicarrier Code Division Multiple Access ("MC-CDMA")_/ and Multicarrier Direct Sequence CDMA ("Multicarrier DS-CDMA") ,

Among the multicarrier and multiple access techniques available today, OFDM and OFDMA have emerged as attractive and powerful choices due to their robustness against multipath fading and high spectral efficiency. OFDM is currently used in the European digital audio and video broadcasting standards, and in digital terrestrial TV broadcasting. OFDM-based hybrid multiple access systems such as MC-CDMA are being considered for third-generation wireless communications systems. The wireline digital subscriber line ("DSL") is based on OFDM. OFDM also outperforms FDMA and TDMA under many channel conditions, and is capable of outperforming CDMA in both single and multi-cellular- based systems.

OFDM achieves high spectral efficiency by dividing the available spectrum into multiple narrowband channels having carriers that are overlapping and orthogonal. Each carrier is at a different frequency and modulated by a given data symbol representing the information to be transmitted. The particular way in which the information is represented depends on the modulation scheme used, which may include phase shift keying

("PSK") and its common variations such as differential phase shift keying ("DPSK") and quadrature phase shift keying

("QPSK"), and quadrature amplitude modulation ("QAM"), among others.

A schematic diagram of a wireless communication system employing OFDM is shown in FIG. 1. The transmitted information signal consists of a linear combination of translations in the time-frequency space of a prototype pulse shape ψ defining the carriers, i.e., x(t) = ∑c_kJψ(t - kT)e^2m!Ft ( 1) k,l

where c_kιl are the information-bearing data symbols, chosen from some finite alphabet constellation of a given modulation scheme, T is the symbol period and F is the separation between the multiple carriers. With W denoting the total bandwidth available for transmission, then N = W/F denotes the number of carriers used in the system.

The time-frequency translations of the prototype pulse shape ψ defining the carriers may be denoted by ψ_klχ as:

_Ψkj(t)=ψ(t-kT)e^2mlFT (2)

A function family of pulse shapes identifying the OFDM system may therefore be represented by the triple (ψ,F,T) .

A necessary condition for perfect reconstruction of the transmitted signal at the receiver is that the functions ψ_k,i are linearly independent (regardless of whether they are orthogonal or not), which implies that TF≥1. The OFDM system {ψ,F,T) is orthogonal if the following condition is satisfied:

_/ --Hv flo, Lifk=mandl=n ^<3) where <ψ,φ> denotes the inner product between two functions ψ and φ.

The orthogonality of the functions ψ_k)1 is not a requirement for perfect reconstruction at the receiver, but minimizes the error caused by additive white Gaussian noise ("AWGN") . The spectral efficiency p of the OFDM system in terms of data symbols transmitted per second per Hertz ("Hz") is approximately given by p = 1/(TF). Since TF≥l, the maximal spectral efficiency of an OFDM system is given by p = 1.

Ideally, one would like to construct an OFDM system (ψ,F,T) that satisfies the following three conditions simultaneously: (1) the functions ψ_k,i should be orthogonal; (2) the pulse shape ψ should be well localized in time and frequency; and (3) TF=I, that is, the OFDM should have maximal spectral efficiency. The first condition, as mentioned above, is desired because of the error minimization in the presence of AWGN. The second condition, that of good time-frequency localization, is important because it leads to the use of a simple equalizer and reduces timing errors effects, frequency offset error effects as well as out-of-band interference, i.e., the leakage of signal energy outside the assigned transmission bandwidth. And the third condition is important as it enables the OFDM system to utilize the allocated spectrum efficiently.

As is well known in the art, the three conditions cannot be satisfied simultaneously due to the Balian-Low theorem. This is the case even if the orthogonality condition is relaxed in favor of biorthogonality, which would increase the sensitivity of the OFDM system to AWGN. As a result, OFDM systems available today employ a number of techniques to achieve the conditions above while attempting to optimize other factors in the system, such as intersymbol interference ("ISI")/ interchannel interference ("ICI") Doppler effect, delay spread and overall system performance in the presence of fading. ISI is caused by time dispersion due to multipath propagation, and ICI results from frequency distortion due to the Doppler effect.

Standard OFDM systems use rectangular pulse shapes and employ a guard interval or cyclic prefix to combat ISI. The main problem with OFDM systems using a guard interval or cyclic- prefix is that the pulse shapes are poorly localized in the frequency domain, which severely limits their performance characteristics in wireless channels and leads to complicated and expensive equalizer design. Additionally, the use of guard intervals or cyclic-prefix does not reduce ICI. This can result in additional loss of spectral efficiency since the carrier signals cannot be placed across the entire available spectrum. If filtering is applied to reduce ICI, the carrier signals are no longer orthogonal, which increases interference and reduces performance. OFDM systems that use a guard interval or cyclic prefix have TF>1 and thus achieve a spectral efficiency of less than 1, typically p = 3/4 or 4/5.

To address the drawbacks of using a guard interval or cyclic-prefix in an OFDM system, pulse-shaping OFDM systems have been proposed. The idea is to construct pulse shapes ψ_kιχ that are well-localized in some sense in time and in frequency, in order to combat both ISI and ICI. For example, U.S. Patent No.

5,790,516 describes pulse shaping methods that are based on a simple filtering of the transmission signal. The methods described therein, however, do not preserve the orthogonality of the pulses ψ_k,i and therefore result in interference between the pulses, leading to a loss of performance in the system.

Other pulse shaping methods described in the prior art, for example, those described in U.S. Patent No. 6,278,686 and U.S.

Patent No. 6,584,068, are very restrictive, as they are only designed for Offset-QAM OFDM ("OQAM/OFDM") and thus for a very limited choice of the parameters T and F. Although OQAM/OFDM maintains maximal spectral efficiency and allows for pulse shapes that are well localized in the time-frequency domain, it leads to increased complexity of the receiver. Furthermore, the pulse shaping methods described in these patents lead to pulses that are of infinite support in time and in frequency, which makes them of limited use in practice. In addition, those methods are not capable of constructing pulses that obey a prescribed spectral mask or a prescribed number of taps.

In view of the foregoing, there is a need in this art for a pulse shaping method for OFDM that combats both intersymbol and interchannel interference while maintaining high spectral efficiency.

There is a further need in this art for a pulse shaping method for OFDM that produces mutually orthogonal transmission pulses that have a prescribed number of taps and show fast spectral decay.

There is also a need in this art for a pulse shaping method for OFDM that is computationally efficient and produces orthogonal pulse shapes that are well localized in time and in frequency.

SUMMARY OF THE INVENTION

In view of the foregoing, one aspect of the present invention provides a pulse shaping method for OFDM that combats both intersymbol and interchannel interference while maintaining high spectral efficiency.

A further aspect of the present invention provides a pulse shaping method for OFDM that produces mutually orthogonal transmission pulses that have a prescribed number of taps and show fast spectral decay.

Another aspect of the present invention provides a pulse shaping method for OFDM that is computationally efficient and produces orthogonal pulse shapes that are well localized in time and in frequency.

These and other aspects of the present invention are accomplished by providing a computationally efficient pulse shaping method for OFDM that produces mutually orthogonal transmission pulses having a prescribed number of taps and fast spectral decay. The transmission pulses provide good protection against both intersymbol interference ("ISI") and interchannel interference ("ICI"), while maintaining high spectral efficiency and efficient bandwidth management.

The pulse shaping method comprises an iterative method for designing OFDM transmission pulses ψ_k.i such that the triple

(ψ,F,T) forms an orthogonal OFDM system and the transmission pulses ψ_kjl satisfy prescribed time-frequency localization conditions. These conditions can be, for example, a fixed number of taps or a prescribed spectral mask.

The iterative method may be implemented in a computationally efficient way by using Newton's method and Fast Fourier Transforms ("FFTs") . The computationally efficient implementation is based on a link between OFDM systems and Gabor frames and can be used to adapt the transmission pulses to time- varying channel conditions in real-time, thereby maximizing the bit-error performance of the OFDM system while maintaining high data rates in wireless transmission. The method may also be used to generate transmission pulses for a wide range of spectral efficiencies.

Advantageously, the pulse shaping method of the present invention may be used in many applications, including, but not limited to, wireless communications with moving transmitter and/or receiver at high bit rates or at low bit rates using a simple equalizer design, multicarrier systems in which the transmission pulses have to satisfy strict spectral constraints such as ultrawideband communication systems, multiuser communications systems such as TDMA and FDMA, digital broadcasting with moving bodies, digital wireline communications such as VDSL, underwater communications, and offset-QAM OFDM systems, among others. In addition, the method reduces equalizer complexity at the receiver in case of Doppler spread, reduces timing errors as well as frequency offset errors, and enables better spectrum management by reducing both ISI and ICI.

BRIEF DESCRIPTION OF THE DRAWINGS

The foregoing and other objects of the present invention will be apparent upon consideration of the following detailed description, taken in conjunction with the accompanying drawings, in which like reference characters refer to like parts throughout, and in which:

FIG. 1 is an exemplary diagram of a prior-art wireless communication system employing OFDM;

FIG. 2 is a flow chart illustrating exemplary method steps for designing mutually orthogonal pulse shapes according to the principles and embodiments of the present invention;

FIG. 3 is a flow chart illustrating exemplary method steps for generating an orthonormal system based on an initial pulse shape according to the principles and embodiments of the present invention;

FIG. 4 is a flow chart illustrating exemplary method steps for enforcing time-frequency localization conditions to the pulse shapes according to the principles and embodiments of the present invention; and FIG. 5 shows time and frequency plots of a pulse shape designed according to the principles and embodiments of the present invention.

DETAILED DESCRIPTION OF EMBODIMENTS OF THE INVENTION

Generally, the present invention provides a pulse shaping method for OFDM and OFDMA. OFDM, as used herein, refers to orthogonal frequency division multiplexing for transmission of information signals. An information signal refers to a voice, data, audio, imagery, or video signal, or any other type of signal representing content. The information signal represents content by means of binary symbols, with each symbol having one or more bits.

Information signals according to the present invention are transmitted by modulating carriers at different frequencies. The carriers are represented by a pulse shape, also referred to herein as a transmission pulse, designed in the present invention to form an orthonormal system. A pulse shape, as used herein, comprises a time-frequency localized signal, such as a Gaussian-type pulse or a raised cosine pulse.

Aspects of the present invention provide for the iterative design of a pulse shape that has a prescribed number of taps and has a fast spectral decay. Fast spectral decay in this context may be the spectral decay of a raise cosine function (which has cubic decay) or even faster, such as exponential decay. The iterative design method generates the pulse shape from an initial pulse shape by applying a method described in T. Strohmer and S. Beaver, "Optimal OFDM system design for time- frequency dispersive channels," IEEE Trans. Comm. , 51(7) :1111- 1122, 2003, incorporated herein by reference. The iterative design method may be implemented in a computationally efficient way by using Newton's method and Fast Fourier Transforms ("FFTs") .

Referring now to FIG. 2, a flow chart illustrating exemplary method steps for designing mutually orthogonal pulse shapes according to the principles and embodiments of the present invention is described. At step 205, an initial pulse shape g₀ is chosen and a pulse shape h₀ is generated from g₀ by applying a method described in T. Strohmer and S. Beaver, "Optimal OFDM system design for time-frequency dispersive channels," IEEE Trans. Comm., 51 (7) :1111-1122, 2003, incorporated herein by reference. Pulse shape h₀ is optimally close to initial pulse shape g₀ and generates a set of mutually orthogonal transmission pulses. Step 205 is described in more detail hereinbelow with reference to FIG. 3.

At step 210, pulse shape h₀ is localized in time and frequency by applying a time-frequency localization operator. Since this operation will destroy the orthogonality of the OFDM transmission pulses, at step 215 the method iterates between steps 205-210 until a prescribed tolerance given by the required noise floor of the OFDM system is achieved. At this point the iterations are stopped and a pulse shape that satisfies the prescribed time-frequency localization properties and at the same time forms an orthonormal system is generated. Steps 210- 215 are described in more detail hereinbelow with reference to FIG. 4.

Referring now to FIG. 3, a flow chart illustrating exemplary method steps for generating an orthonormal system based on an initial pulse shape according to the principles and embodiments of the present invention is described. The pulse shape design method utilizes the following standard design parameters for OFDM: (1) symbol period T; (2) carrier separation F; (3) number of carriers N; and (4) required noise floor of ε dB. If estimates of the maximal expected delay spread τ and the maximal expected Doppler spread v are known a priori, which is usually the case in most OFDM-based communication systems, then T and F are chosen at step 305 such that:

At step 310, an initial pulse shape g₀ is chosen. The initial pulse shape go may be a Gaussian-type pulse

2 g₍₎(t)=e^~aπt where a = F/T₁ or a raised cosine pulse. The initial

OFDM system (g_o,T,F) is in general not an orthogonal system. At step 315, a counter is initialized at n = 0. At this point, g_n = go-

At step 320, for a given g_nι an orthonormal OFDM system (h_n, T,F) is constructed from {g_n, T₁F) by applying a method described T. Strohmer and S. Beaver, "Optimal OFDM system design for time-frequency dispersive channels," IEEE Trans. Comm. , 51 (7) :1111-1122, 2003, incorporated herein by reference, as follows:

where

The existence of R^~1/2 (g_n) follows from the positive-definiteness of R(g_n) , which in turn follows from the condition that (g_arT,F) be linear independent. In addition, pulse shape h_n computed in Equation (5) is optimally close to g_n in the sense that it minimizes the energy |g_n -A_n|₂.

Equation (5) above is computationally expensive but may be solved very efficiently by using Newton's method. First, set g⁽⁰⁾ =g_n and for j = 0, 1,..., compute iteratively:

where the operator U^(3> is defined by:

and R(gip) is given by Equation (6) above.

Computing g⁽³⁾ with the Newton's method above will converge to h_n for increasing j. A few iterations are usually sufficient for convergence. The inversion of R(g_k) can be done numerically very efficiently by a method described in T. Strohmer, "Numerical algorithms for discrete Gabor expansions," in H. G. Feichtinger and T. Strohmer, editors, Gabor Analysis and Algorithms: Theory and Applications, chapter 8, pages 267-294, Birkhauser, Boston, 1998, the entire content of which is incorporated herein by reference.

The method described therein may be adapted to pulse shaping design for OFDM by utilizing a link between OFDM systems and Gabor frames. An OFDM system may be interpreted as an undersampled or critically sampled orthogonal Gabor system. If

{g,a,b) is a tight Gabor frame which has excellent time- frequency localization, that is, if ab > 1, then {g,T,F) with T = l/b and F = l/a is an OFDM system with excellent time- frequency localization.

A tight Gabor frame can be constructed from a given Gabor frame (g,a,b) via a canonical tight window defined as S^~1/2g. As pointed out in T. Strohmer and S. Beaver, "Optimal OFDM system design for time-frequency dispersive channels," IEEE Trans. Comm. , 51 (7) :1111-1122, 2003, incorporated herein by reference, S^~1//2g may be equivalently expressed as:

where the gi_c,i are associated with the system {g,l/b,I/a) and R(f) is defined by Equation (6) above with T = l/b and F = I/a.

The proposed method steps expressed in Equation (5) is therefore equivalent to computing the left-hand side of Equation (9) . That is, computing h_n in Equation (5) is equivalent to computing S^'1/2g in Equation (9) .

It has been shown in T. Strohmer, "Rates of convergence for the approximation of dual shift-invariant systems in I²Z₁" J. Four. Anal. Appl. , 5 (6) :599-615, 2000, incorporated herein by reference, that an arbitrarily accurate approximation to Gabor frame operator S can be derived by a periodic finite-dimensional model in (f¹. By doing so, Gabor frame operator S may be factorized as:

S=I_p®diαg(C₀,...,C_±_) (10)

PQ

where S is a L x L matrix representing the Gabor frame operator S, —≡:— with p, q being relative prime integers, and the q x q αb q submatrices C₃- are given by: diag(C_±b ,...C_jbmJ = ψ__b ®I_q) *B_k(F__b ®I_q) (11 )

where F_h is the Discrete Fourier Transform ( "DFT" ) of size — x — - 1 1 and the b x b matrices B_k axe given by :

(βk ) urn ~ S_k+mL I _bMllL I j (12)

for m, n = 0,...,b-l, k = 0,...,L/b-l.

Hence, the Gabor frame operator S may be efficiently factorized into Fast Fourier Transforms (^wFFTs") and simple permutations into L/(pq) different submatrices of size q x q where each of these submatrices is repeated p times. Since permutations are performed simply by data addressing, the whole inversion of S can be done very efficiently and hence, as computation of h_n is equivalent to computation of S^'1/2g, the computation of h_n can therefore be done very efficiently.

The resulting pulse shapes {(h_n)_k/χ)} form an orthonormal system. However, h_n will not in general obey the desired time- frequency localization conditions. Referring now to FIG. 4, a flow chart illustrating exemplary method steps for enforcing time-frequency localization conditions to the pulse shapes according to the principles and embodiments of the present invention is described. At step 405, time-frequency localization conditions are applied to the computed pulse shape h_n by multiplying h_n by a time-frequency localization operator as follows:

_n(t)=(Ah_n)(t) (13)

where A is the time-frequency localization operator. Time-frequency localization operator A may be used to ensure that pulse shape h_n has a fixed number of taps or a specific spectral mask. To enforce a given number of taps in pulse shape h_n, A may defined as follows:

To comply with FCC regulations and other standards, it is often required that transmission signals obey a certain spectral decay. To enforce that pulse shape h_n will have a specific spectral mask, A may defined as follows:

where V_ω is the short-time Fourier transform defined in Equation

(16) below and F(t,(D) determines the temporal and spectral properties of h_n. For example, F(t,ω) may be given by F(t,ω) =

W₁ (t)W₂(Co) , where W₁ and W₂ are Hamming-type windows, or W₁ could be a Gaussian function and W₂ could be a function whose shape is determined by a specific spectral mask corresponding to FCC standards, such as those for transmission pulses in ultrawideband communications.

The short-time Fourier transform ("STFT") of a function h with respect to a window function w is defined by:

where ω is typically a Gaussian-type function.

While setting g_n(?)=(Ah_n)(t) ensures that pulse shape g_n will satisfy the time-frequency localization conditions imposed by time-frequency localization operator A, (g_n,T,F) may no longer generate an orthonormal system. Therefore, at step 410, the counter is incremented by 1, that is, n = n + 1, and h_n is recomputed at step 415 by using g_n as its initial pulse shape. Pulse shape h_n is recomputed from g_n as described hereinabove with reference to step 320 of FIG. 3.

The method then repeats steps 405-415 until a prescribed tolerance given by the required OFDM noise floor of ε is achieved at step 420 at a given iteration, say at iteration n = n_*. For example, the tolerance criterion could be specified as follows:

At this point, the iterations terminate and the desired pulse shape designed according to the principles and embodiments of the present invention is set at:

where the denominator denotes the 2-norm of g_n*:

By construction, the designed pulse shape ψ satisfies the prescribed time-frequency localization properties and at the same time forms an orthogonal OFDM system given by (ψ,T,F) .

The number of iterations required to reach step 425 depends on the choice of the time-frequency localization operator A. If the range of A is a convex set or a linear subspace, then convergence can be shown using methods known in the art for alternating projection algorithms. It should be understood by one skilled in the art that orthogonality of the pulse shapes ψ_k/1 is not a strict requirement for perfect reconstruction of the data even though it minimizes the error caused by AWGN. If the constructed pulse shapes ψ_k)1 are not orthogonal due to a too early termination of the iterations or other reasons, then the receiver pulse shapes may employ a biorthogonal set of pulse shapes φ_m,_n for a given pulse shape φ. Two function systems {ψ_k,i} and {<P_m,_n} are biorthogonal if the following condition is satisfied:

_{/ \} fθ if k = m and l = n, K^) = J_{1 d} ^J _se ⁽20⁾

In this case, receiver pulse shapes φ_k)1 may be computed in a number of ways, for example, as:

Computation of φ can be done very efficiently by using the link to Gabor frames as described hereinabove. Other possibilities for computing φ may be used, which are analogous to the computation of general dual windows for Gabor frames, as described in T. Strohmer, "Numerical algorithms for discrete Gabor expansions," in H. G. Feichtinger and T. Strohmer, editors, Gabor Analysis and Algorithms: Theory and Applications, chapter 8, pages 267-294, Birkhauser, Boston, 1998, incorporated herein by reference.

It should also be understood by one skilled in the art that carrier functions for OQAM/OFDM may be constructed from pairwise combining the carrier functions of a standard pulse shaping OFDM system with parameters T.F = 2. Referring now to FIG. 5, time and frequency plots of a pulse shape designed according to principles and embodiments of the present invention are described. Plots 500-515 were produced by generating an OFDM pulse shape for an OFDM system with the following specifications: (1) T = I nanosecond; (2) F = 1.33 GHz; (3) ε = 40 dB; (4) p = 3/4; and (5) number of taps = 55, which corresponds to a pulse duration of less than 7 nanoseconds. The time-frequency localization operator was chosen as in Equation (14) and the pulse shape was designed according to the method steps described hereinabove with reference to FIGS. 2-4.

As illustrated in FIG. 5, while the initial pulse shape plotted in plots 500-505 has fast temporal and spectral decay, it does not generate an orthogonal system nor does it have a finite number of taps. Using this initial pulse shape in an OFDM system would lead to severe ISI and ICI. However, the pulse shape designed by the present invention and plotted in plots 510-515 has a finite number of taps, is highly localized in the time and frequency domains and generates an orthogonal system.

The foregoing descriptions of specific embodiments and best mode of the present invention have been presented for purposes of illustration and description only. They are not intended to be exhaustive or to limit the invention to the precise forms disclosed. Specific features of the invention are shown in some drawings and not in others, for purposes of convenience only, and any feature may be combined with other features in accordance with the invention. Steps of the described processes may be reordered or combined, and other steps may be included. The embodiments were chosen and described in order to best explain the principles of the invention and its practical application, to thereby enable others skilled in the art to best utilize the invention and various embodiments with various modifications as are suited to the particular use contemplated. Further variations of the invention will be apparent to one skilled in the art in light of this disclosure and such variations are intended to fall within the scope of the appended claims and their equivalents.

Claims

WHAT IS CLAIMED IS :

1. A method for orthogonal frequency division multiplexing pulse shape design comprising:

(a) providing a first pulse shape; (b) generating a second pulse shape from the first pulse shape;

(c) localizing the second pulse shape in time and in frequency; and

(d) repeating steps (b) and (c) until a prescribed tolerance is achieved and the second pulse shape has fast spectral decay and generates a set of mutually orthogonal transmission pulses.

2. The method of claim 1, wherein the first pulse shape is one of a Gaussian pulse and a raised cosine pulse.

3. The method of claim 1, wherein the generating step comprises choosing a symbol period and a carrier separation based on a ratio between a maximal expected delay spread and a maximal expected Doppler spread.

4. The method of claim 1, wherein repeating steps (b) and (c) until a prescribed tolerance is achieved and the second pulse shape has fast spectral decay and generates a set of mutually orthogonal transmission pulses comprises generating mutually orthogonal transmission pulses that are optimally time-frequency localized.

5. The method of claim 3, wherein the second pulse shape is part of an orthogonal system having a symbol period and carrier separation and wherein the second pulse shape has minimal energy among all pulse shapes that form the orthogonal system with respect to the symbol period and the carrier separation.

6. The method of claim 1, wherein the generating step comprises applying the Newton's method.

7. The method of claim 1, wherein localizing the second pulse shape in time and in frequency comprises applying a time- frequency localization operator to the second pulse shape.

8. The method of claim 7, wherein the time-frequency localization operator comprises an operator for fixing the number of taps of the second pulse shape.

9. The method of claim 7, wherein the time-frequency localization operator comprises an operator for specifying the spectral mask of the second pulse shape.

10. The method of claim 1, wherein the repeating step comprises recomputing the second pulse shape from the first pulse shape at each iteration.

11. The method of claim 1, wherein the first pulse shape has a 2-norm, further comprising generating a third pulse shape from the first pulse shape by dividing the first pulse shape by the 2-norm of the first pulse shape.

12. The method of claim 11, further comprising generating a receiver pulse shape from the third pulse shape, wherein the receiver pulse shape and the third pulse shape are biorthogonal.

13. The method of claim 1, further comprising applying the second pulse shape in an orthogonal quadrature amplitude modulation/orthogonal frequency division multiplexing communications system.

14. A method for orthogonal frequency division multiplexing pulse shape design comprising iteratively generating mutually orthogonal transmission pulses that are well localized in time and frequency and have fast spectral decay.

15. The method of claim 14, wherein iteratively generating mutually orthogonal transmission pulses comprises providing a first pulse shape and iteratively generating a second pulse shape from the first pulse shape.

16. The method of claim 15, wherein the first pulse shape is one of a Gaussian pulse and a raised cosine pulse.

17. The method of claim 15, wherein iteratively generating a second pulse shape from a first pulse shape comprises choosing a symbol period and a carrier separation based on a ratio between a maximal expected delay spread and a maximal expected Doppler spread.

18. The method of claim 15, wherein iteratively generating a second pulse shape from a first pulse shape comprises applying the Newton's method.

19. The method of claim 15, wherein the first pulse shape has a 2-norm, further comprising generating a third pulse shape from the first pulse shape by dividing the first pulse shape by the 2-norm of the first pulse shape.

20. The method of claim 15, further comprising applying the second pulse shape in an orthogonal quadrature amplitude modulation/orthogonal frequency division multiplexing communications system.