This action might not be possible to undo. Are you sure you want to continue?

Welcome to Scribd! Start your free trial and access books, documents and more.Find out more

5, MAY 1999

1335

**Discrete Fractional Fourier Transform Based on Orthogonal Projections
**

Soo-Chang Pei, Senior Member, IEEE, Min-Hung Yeh, and Chien-Cheng Tseng, Member, IEEE

Abstract—The continuous fractional Fourier transform (FRFT) performs a spectrum rotation of signal in the time–frequency plane, and it becomes an important tool for time-varying signal analysis. A discrete fractional Fourier transform has been recently developed by Santhanam and McClellan, but its results do not match those of the corresponding continuous fractional Fourier transforms. In this paper, we propose a new discrete fractional Fourier transform (DFRFT). The new DFRFT has DFT Hermite eigenvectors and retains the eigenvalue-eigenfunction relation as a continous FRFT. To obtain DFT Hermite eigenvectors, two orthogonal projection methods are introduced. Thus, the new DFRFT will provide similar transform and rotational properties as those of continuous fractional Fourier transforms. Moreover, the relationship between FRFT and the proposed DFRFT has been established in the same way as the conventional DFT-tocontinuous-Fourier transform. Index Terms— Discrete Fourier transform, discrete fractional Fourier transform, Fourier transform, fractional Fourier transform.

I. INTRODUCTION HE FOURIER transform (FT) is one of the most frequently used tools in signal analysis [1]. A generalization of the Fourier transform—the fractional Fourier transform (FRFT)—has been proposed in [2] and [3] and has become a powerful tool for time-varying signal analysis. In time-varying signal analysis, it is customary to use the time-frequency plane, with two orthogonal time and frequency axes [4]. Because the successive two forward Fourier transform operations will result in the reﬂected version of the original signal, the FT can in the be interpreted as a rotation of signal by the angle time–frequency plane and represented as an orthogonal signal representation for sinusoidal signal. The FRFT performs a rotation of signal in the continuous time–frequency plane to any angle and serves as an orthonormal signal representation for the chirp signal. The fractional Fourier transform is also called rotational Fourier transform or angular Fourier transform in some documents. Besides being a generalization of the FT, the

Manuscript received December 2, 1996; revised February 19, 1998. This work was supported by the National Science Council, R.O.C., under Contract NSC 85-2213-E002-025. The associate editor coordinating the review of this paper and approving it for publication was Dr. Akram Aldroubi. S.-C. Pei is with the Department of Electrical Engineering, National Taiwan University, Taipei, Taiwan, R.O.C. (e-mail: pei@cc.ee.ntu.edu.tw). M.-H. Yeh is with the Department of Computer Information Science, Tamsui Oxford University College, Tamsui, Taipei, Taiwan, R.O.C. C.-C. Tseng was with the Department of Electronics Engineering, Hwa Hsia College of Technology and Commerce, Taipei, Taiwan, R.O.C. He is now with the Department of Computer and Communication Engineering, National Kaohsiung First University of Science and Technology, Taipei, Taiwan, R.O.C. Publisher Item Identiﬁer S 1053-587X(99)03244-4.

T

FRFT has been proved to relate to other time-varying signal analysis tools, such as Wigner distribution [4], short-time Fourier transform [4], wavelet transform, and so on. The applications of the FRFT include solving differential equations [2], quantum mechanics [3], optical signal processing [5], timevariant ﬁltering and multiplexing [5]–[8], swept-frequency ﬁlters [9], pattern recognition [10], and time–frequency signal analysis [11]–[13]. Several properties of the FRFT in signal analysis have been summarized in [9]. Many methods for realizing the FRFT have been developed, but most of them are to utilize the optical implementation [14], [15] or numerical integration. Because the FRFT is a potentially useful tool for signal processing, the direct computation of the fractional Fourier transform in digital computers has become an important issue. The ideal discrete fractional Fourier transform (DFRFT) will be a generalization of the discrete Fourier transform (DFT) that obeys the rotation rules as the continuous FRFT and provides similar results as the FRFT. In [16], a method for a numerical integration FRFT has been proposed, but the method does not obey the rotation rules, and the signal cannot be recovered from its inverse transform. In [17], Santhanam and McClellan have developed a discrete FRFT, but their method does not provide the same transforms to match those of the continuous case. In this paper, we present a new discrete fractional Fourier transform (DFRFT). This DFRFT is a generalization of the DFT and will provide similar transforms as those of the continuous case. The relationship between the DFRFT and the FRFT can also be established and discussed in detail. Moreover, the proposed DFRFT has important unitary and rotation properties. This paper is organized as follows. In Section II, the previous development of continuous and discrete fractional Fourier transforms are reviewed. The concept for developing the DFRFT to have similar results as the continuous corresponding case are described in Section III. Two acceptable solutions for the DFRFT are considered and proposed in Section III. Then, the relationships between the FRFT and the DFRFT can be established in Section IV. Finally, conclusions and discussions are made in Section V. II. PRELIMINARY A. Continuous FRFT The Fourier transform of a signal can be interpreted as a angle rotation of the signal in the time–frequency plane. The FRFT is then developed and treated as a rotation of signal

1053–587X/99$10.00 © 1999 IEEE

the (8) is the conventional DFT of signal . In [19]. The transform kernel of continuous FRFT is deﬁned as [2]. Fig. The equivalence of (1) and (2) has been proved in [3].1336 IEEE TRANSACTIONS ON SIGNAL PROCESSING. the two results . to any angles in the time–frequency plane [9]. The conventional methods to compute the DFRFT in [17] and [20] are mainly on numerical computing the fractional power of the DFT kernel matrix. VOL. otherwise here is deﬁned as [ ]. Furthermore. [3]. and a circular reﬂected version of its DFT . with unit variance. its DFT . The th-order normalized Hermite function with variance is deﬁned as (3) is the th-order Hermite polynomial [18]. (2) provides an eigendecomposition representation of the FRFT kernel. Using the kernel of the by angle is computed as FRFT. and the imaginary parts of the FRFT or DFRFT are indicated by dashed lines. . NO. the FRFT of the signal (4) (5) Equation (5) indicates that the FRFT can be interpreted as a weighting summation of Hermite functions. Fractional Fourier transform of a rectangular window. is the eigenfunction of the FRFT. 5. 2 are quite different from the results of Fig. it will have a different phase term in the weighting coefﬁcients. Thus. The discrete results shown in Fig. the Greek subscripts (for example. The weighting coefﬁcients are obtained from multiplying the phase term and the inner product of the input signal and the corresponding Hermite function. 1. Fig. x(t) = 1 when 02 t 2. [6]. the middle is not the intermediate state of transform result for and . [9] csc if if if is not a multiple of (1) is a multiple of is a multiple of (2) where indicates the rotation angle of transformed signal for is the th-order normalized Hermite function the FRFT. MAY 1999 Fig. the Hermite functions in [19] with a different scaling to those of the FRFT. B. The rectangular window used . Real part: solid line. The Old DFRFT The kernel matrix of the DFRFT is obtained by computing the fractional power for the DFT kernel matrix. . and the rotation output of the signal is also the weighting summation of Hermite functions. The fractional power of the DFT kernel matrix [20] is calculated by (6) (6) where (7) Applying the above deﬁned kernel to the signal is computed as DFRFT of the signal . and . elsewhere] for various angles. 1. and x(t) = 0 otherwise. a rotation operation using Hermite functions has also been proposed. The real parts of the FRFT or DFRFT in this paper are plotted by solid lines. However. In order to avoid ambiguity. 47. 1 shows the FRFT of the rectangular window function [ for . 2 shows the results of this DFRFT produced by a discrete rectangular window. Imaginary part: dashed line. and ) are used to denote the rotation angles in the time–frequency plane. Because where the normalized Hermite function with unitary variance. The fractional power of the DFT kernel matrix and the rotation angles in the DFRFT essentially mean the same thing. a circular reﬂected version of the signal ( ). Equation where is the linear (8) indicates that the DFRFT of the signal combination of the four major angular parts: the original signal . and the English superscripts are used to denote the fractional power values of the DFT kernel matrices in this paper.

In [20]. B. Matrix commutes with the DFT kernel . . and matrix is deﬁned as OF DFRFT The continuous FRFT has a Hermite function with unitary variance as its eigenfunction. are . Can the DFT have the eigenvectors with the similar shapes as the Hermite functions? These DFT eigenvectors are known as DFT Hermite eigenvectors in this paper.PEI et al. where is the spread variance of the Proof: It is assumed that DFT eigenvectors. Eigendecomposition of DFRFT The development of our DFRFT is based on the eigendecomposition of the DFT kernel. Proposition 1: The eigenvalues of and its multiplicities are shown in Table I. . In the discussion in the previous subsection. it satisﬁes the commutative property (10) will also be the eigenvectors The eigenvectors of matrix of the DFT kernel matrix . The eigenvalue multiplicities of DFT matrix indicate the ranks of the DFT eigensubspaces [21]. 1. . Proof: See [21]. where associated continuous spread variance is the sampling intervals of signal. a multiplicity function function is used to denote the DFT eigenvalue multiplicity for . we get (12) is the th-order Hermite polynomial. and many properties of the DFT matrix eigenvalues and eigenvectors have been discussed in [20] and [21]. The eigenvectors of the DFT matrix constitute four . a method for computing the DFT eigenvectors has been introduced. . Proposition 3: DFT Hermite eigenvectors should have the . we only summarize some results for our further development of DFRFT. Proof: See [20]. . . The parameter is the index for the DFT eigenvalue . DFT Hermite Eigenvectors III. from the continuous FRFT in Fig. In [22]. If the Hermite function are sampled in this way. —respectively. but it cannot obtain the real-valued DFT eigenvectors. .: DISCRETE FRACTIONAL FOURIER TRANSFORM BASED ON ORTHOGONAL PROJECTIONS 1337 Fig. . The output is quite different TABLE I MULTIPLICITIES OF THE EIGENVALUES OF DFT KERNEL MATRIX where matrix . and then. DEVELOPMENT A. . Old DFRFT of a rectangular window function. a novel matrix is introduced to compute the real and complete set of DFT eigenvectors very elegantly. x(n) = 1 when 06 n 6. we have known that the eigendecomposition of the DFT kernel matrix is not unique. is deﬁned. The corresponding eigenfunction property for the DFT would be like (11) is the eigenvector of DFT corresponding to the where th-order discrete Hermite function. . . and x (n ) = 0 otherwise. 2. The continuous approximate form can be written as (13) Sampling by . and —and each is major eigensubspaces— . . . [22]. .. . Here. 1. Proposition 2: A matrix can be used to compute the real eigenvectors of the DFT matrix . This In Table I. (9) The Fourier transform of (13) can be computed as (15) . (13) will become (14) . and corresponding to one of the four eigenvalues—1. but they correspond to different eigenvalues.

then it can be shown that interval If is even (17) If is odd (18) for sufﬁciently large . we obtain and letting This expression is valid for any both sides.1338 IEEE TRANSACTIONS ON SIGNAL PROCESSING. Thus. The sequence obtained from (12) is also equal to the sampling of the Hermite function with the unit variance by . both errors approach zero. the sampling interval we will prove that even such samplings of Hermite functions can only have approximate Hermite eigenvectors of the DFT. 5. deﬁned in the range Proposition 5: If the sequence is obtained by shifting Hermite Gauss samples in the following way: If is even for (22) for and if is od. Thus. by replacing the continuous integral with the numerical integral. In Propositions 4 and 5. VOL. it is clear that there are two approximation errors in (17). because larger is . From the above proof. frequency spectrum that DFT can cover is the continuous spectrum spacing (or resolution) of the DFT is . and the other is the numerical error in (20). The th-order continuous Hermite function should have zeros [18]. the slower decay rate the Hermite function will have. we have the approximation expression (19) is wide This approximation is valid because is for large . we have (20) (24) for sufﬁciently large . is even. One is the truncation error in (19). but this will not inﬂuence the development of the DFRFT. Therefore. . 47. taking at (21) Equation (14) becomes The proof is completed. The sampled Hermite functions still can be used to construct the DFRFT kernel because they have the similar shapes and good approximations to the continuous Hermite functions. However. Next.e. the better approximation (17) is. the proof can be easily derived in the same way. the is. the larger order is. This implies that the truncation error in (19) is larger for high-order . The variable in (15) can be replaced by equal to to have its discrete form. NO. MAY 1999 The sampling theorem [23] tells us the continuous scope of . we can ﬁnd that the DFT Hermite eigenvectors should have associate continuous spread variance . . the approximation in (17) becomes worse. we only prove the case that odd. Next. For By truncating the integral interval of Fourier transform from to . the decay rate the degree of Hermite polynomial is proportional to for of the Hermite function sufﬁciently large . The small aliasing will occur while is closer to . The condition of for the two shapes of (14) and (16) being the same becomes This approximation is also valid because is is very large. and the decay rate of Gauss function very fast. i. when order becomes large. Proof: Here. Combining (19) and (20) very small when . we have . these functions are not bandlimited. is obtained by samProposition 4: If the sequence pling the unit variance Hermite function with the sampling . (16) adjusts the scaling of Hermite funcThe variance value to make the DFT tion. for (23) for then it can be shown that the DFT of the . and the samplings in Proposition 3 cannot guarantee that the numbers of the sign changes in this sampled th-order Hermite function are also . approximated by can be From Proposition 3. and our goal is to adjust the variance have the same eigenvector shape as itself.. When the value of approaches inﬁnity. Thus. Thus.

The idea for developing our DFRFT is to ﬁnd the discrete form for (2). (26) Substituting (26) into (25) and using Proposition 4. It is because the approximation error of the low DFT Hermite eigenvectors are small. we should not expect that a ﬁnite vector can express the oscillation behavior of the very high-order Hermite function very well. Because of the above ambiguity. In order to retain the eigenfunction property in (11). will be a DFT Hermite eigenvector. for the discrete case. In addition. In the case of continuous FRFT. This assignment rule matches the multiplicities of the eigenvalues of the DFT kernel matrix in Table I. In Propositions 4 and 5. The selections of the DFT Hermite eigenvectors are from low to high orders. Combining . Table II shows the eigenvalues assignment rules for the DFRFT. we also prove the case that is even. we can compute in its DFT eigensubspace to obtain a the projections of Hermite-like DFT eigenvector (32) mod . the terms of the Hermite functions are summed up from order zero to inﬁnity. is odd. therefore. This indicates that there exists root ambiguity in deciding the fractional power of eigenvalues. In (32). it has been proved that the samplings of Hermite functions can have approximate DFT eigenvectors. The normalized vectors for the samplings of Hermite functions are deﬁned as (31) . we obtain C. there exist two types of ambiguity in deciding the fractional power of the DFT kernel matrix. the unit variance Hermite functions are in the following sampled with a period of discussions. Unfortunately. The While is given by DFT of the sequence DFT TABLE II EIGENVALUES ASSIGNMENT RULE OF DFRFT KERNEL MATRIX (25) Using the equality second term on the right side of (25) becomes . (27) can be The fractional power of matrix can be calculated from its eigendecomposition and the powers of eigenvalues. eigenvectors for the However. we obtain DFT for for (30) The proof is completed.: DISCRETE FRACTIONAL FOURIER TRANSFORM BASED ON ORTHOGONAL PROJECTIONS 1339 Proof: Here. the DFT is computed from the eigenvectors of Hermite eigenvector matrix in the same DFT eigensubspace. . only DFT Hermite eigenvectors can be added. In addition. we know that there are several DFRFT kernel matrices that can obey the rotation properties. the eigenvectors can be used as bases for individual DFT eigensubspaces. • Ambiguity in Deciding the Eigenvectors of the DFT Kernel Matrix: The DFT eigenvectors constitute four major eigensubspaces. Newly Developed DFRFT DFT (27) (28) is limited in the range where equality rewritten as DFT (29) is limited in the range where (28) and (29). the choices for the DFT eigenvectors to construct the DFRFT kernel are multiple and not unique. and is the eigenvector of matrix where . the Because matrix can have complete real orthogonal DFT eigenvectors. the proof can also be easily derived. • Ambiguity in Deciding the Fractional Powers of Eigenvalues: We know that the square root of unity are 1 and 1 from elementary mathematics.PEI et al. Using the .

1340 IEEE TRANSACTIONS ON SIGNAL PROCESSING. Therefore. . orthogonalized Hermite eigenvector. MAY 1999 The transform kernel of the DFRFT can be deﬁned as (33) for (odd) (34) for (even) where . . Such a method is called the method in this paper. is the bottom of the page. The Gram–Schmidt Algorithm (GSA): The Gram–Schmidt method [29] is a well-known orthogonalization approach for vectors. . whereas is odd. Unfortunately. . So the DFT Hermite eigenvectors can be orthogonalized for every eigensubspace individually to obtain orthogonal eigenvectors in the whole eigenspace of DFT. for and the eigenvalues of the DFRFT are . and for even . 47. Although the Mathieu functions can converge to a Hermite function [28]. the convergence for the eigenvectors obtained from matrix are not so fast for the high-order Hermite functions by the LMS error criterion.. nonorthogonalized Hermite eigenvector. Equation (32) provides a method for ﬁnding DFT Hermite eigenvectors. In [25]. and for is even. Two methods for vector orthogonalization are proposed in the following of this paper. where the eigenvector corresponding to the th continuous Hermite function. The symbol notation in developing the two algorithms are as follows: continuous Hermite samples vector. . It is easy to show that the eigenvectors located in different eigensubspaces will be orthogonal. the transform kernel for the rotation angle can be computed as in (34a). the eigenvectors obtained from matrix are directly considered to be the discrete Hermite functions. the DFT Hermite eigenvectors obtained from (32) cannot constitute an orthogonal basis for DFT eigenspace. The DFT eigenvectors are orthogonalized for each eigensubspace. . . . 5. vector orthogonalization is required for the DFT Hermite eigenvectors obtained from (32). Example 1— : The rotation angle is equal to . However. Therefore. . VOL. shown at . The role of matrix in (32) is just as a tool to ﬁnd a complete set of real and orthogonal DFT eigenvectors. An eigenbased method for computing the DFRFT to have similar continuous results has been proposed by us in [24]–[27]. This is because the eigenvectors located in different eigensubspaces will be always orthogonal. is deﬁned as follows: where For odd . NO. The DFT Hermite eigenvectors in each DFT eigensubspace can be orthogonalized by the Gram–Schmidt method. the eigenvalue–eigenfunction relation is retained in deﬁning the DFRFT. Calculate the continuous samples of Hermite functions: Compute the eigenvectors of matrix : Using equation (32) to compute Hermite eigenvectors by projections: for to for to end end . the eigenvectors obtained from the matrix are just discrete Mathieu functions [21]. It must be mentioned that the last eigenvalue and should be assigned to the is not assigned to to match the eigenvalue multiplicities value of the DFT matrix in Table I. (34a) . This means that is replaced by in (34). In addition. . is the normalized eigenvector corresponding to the th-order Hermite function.. It is easy to verify that the angle rotation property of the DFRFT can be preserved only while the DFT eigenvectors are orthogonal.

) are plotted Transform results of the impulse signal ( in Figs. The chirp signal used here is equal . The error vectors between samples of Hermite functions and DFT Hermite eigenvectors are deﬁned as (37) Example 2: In this example. Therefore. Figs. The norm of the error vectors between the DFRFT and samples of the FRFT are also shown in the titles of Figs. . Example 3: The DFRFT by the GSA for a chirp signal is computed in this example. where . 9 and 10. the norms of error vectors between the computed DFT Hermite eigenvectors and . [ which is the same as the signal used in Fig. Norms of error vectors between DFT Hermite eigenvectors and samples of Hermite functions. We can formulate our problem as the OPA to ﬁnd the least Frobenius norm between the samples of Hermite functions and orthogonal DFT Hermite eigenvectors. 7. 4–6 show the DFRFT of the rectangular window function for various angles for method. Frobenius norm of (35) samples of Hermite functions are plotted for . the solution will also . The is equal to 4/13. and the number of points sampling interval is equal to 73. we can ﬁnd that the proposed DFRFT provides a similar orthonormal signal decomposition for discrete chirplike signals.: DISCRETE FRACTIONAL FOURIER TRANSFORM BASED ON ORTHOGONAL PROJECTIONS 1341 Fig. and OPA. In Fig. respectively. E. 3. it can be to found that the transform results change from a chirp signal ) to an impulse-like function ( ). ( the DFRFT can be used for the chirp signal and chirp rate detection. A more detailed theory and algorithm can be found in [7]. ]. is the subject to (36) the . 2. 11. respectively. The corresponding samples of the continuous FRFT for the impulse signal are plotted in Fig. GSA. satisfy Algorithm Calculate the continuous samples of Hermite functions: Compute the eigenvectors of : Using equation (32) to compute Hermite eigenvectors by projections: for to Compute the SVD of . the last half of the indices in the DFRFT must also be treated as the negative end The GSA minimizes the errors between the samples of Hermite functions and orthogonal DFT Hermite eigenvectors from low to high orders. Based on the unitary property in the DFRFT and the transform results shown in Figs. 8–10 for the method. The continuous FRFT is an orthonormal signal decomposition for chirp signals [9]. . Properties of DFRFT The properties of the DFRFT are shown in Table III. and OPA. otherwise. The sampled discrete data then becomes . 8–10. 582] for the two given spaces. and the OPA minimizes the total errors between the samples of Hermite functions and orthogonal DFT Hermite eigenvectors. Implementation of the New DFRFT As in the case of DFT frequency domain. The minimizing calculating the singular value decomposition (SVD) of . 1. for minimize minimize where matrix. Orthogonal Procrustes Algorithm (OPA): A traditional mathematical problem known as the orthogonal Procrustes Algorithm can be used to ﬁnd the least Frobenius norm [30. Because the .PEI et al. we use the DFRFT to deal with the rectangular window function shown in Fig. D. p. 3. and will can be found by be our solution. GSA. In Fig. .

47. the following method for implementing the DFRFT can be applied: (38) (39) frequency. identity transform) and any angle transform domains. and x(n) = 0 otherwise. computing the DFRFT is . and the eigenvalues of DFRFT are also determined. 6. x(n) = 1 when 06 n 6. When = 3=7. and it is the same as in the DFT case. and x (n ) = 0 otherwise. DFRFT by OPA method of a rectangular window function. x (n ) = 1 when 06 n 6. The complexity of . ej 20:1141k .1342 IEEE TRANSACTIONS ON SIGNAL PROCESSING. the continuous FRFT in Fig. NO. closer output to the continuous FRFT than Fig. DFRFT by GSA of a chirp signal. This ﬁgure has a Fig. 7. VOL. x (n ) = 1 when 06 n 6. an impulse-like output is obtained. MAY 1999 Fig. 4. DFRFT by S method of a rectangular window function. k = 032 1 1 1 32. 4. output to the continuous FRFT than Fig. Then. 4. the DFT Hermite eigenvectors can be computed a priori. 1. ( and the rotation angle are When the number of points determined. the computation of the DFRFT can be implemented only by a transform kernel matrix multiplication. 5. If the rotational angles are adjusting. The output is close to Fig. This concept is also applied in the time domain . DFRFT by GSA method of a rectangular window function. and x (n ) = 0 otherwise. 5. This ﬁgure has a closer Fig.

It can be seen that chirp-like outputs are obtained while the angle 6= =2. Here. It can be seen that chirp-like outputs are obtained while the angle = =2. DFRFT by S method of an impulse function. but it cannot have similar results as in the continuous case. and x(n) = 0 otherwise. and they can be computed in advance. and the output becomes a constant. It can be seen that chirp-like outputs are obtained while the angle = =2. Since any eigenvector in Therefore. 9. Applying the transform eigensubspace kernel deﬁned in (6) to the eigenvector . DFRFT by OPA of an impulse function. Discussion The method in [17] obeys the rotation properties. 6 (40) The deﬁnitions of matrix and matrix are the same as those .: DISCRETE FRACTIONAL FOURIER TRANSFORM BASED ON ORTHOGONAL PROJECTIONS 1343 Fig. Proof: Let be any DFT eigenvector located in the . When = =2. 10. we can obtain is also a DFRFT eigenvector. the inner products of signal and eigenvectors. and the value is the eigenvalue for this eigenvector . When = =2. This is the major reason for the mismatches in [17]. Proposition 6: The method in (6) assigns all the eigenvectors of the DFRFT matrix to only four different eigenvalues. and x(n) = 0 otherwise. x(0) = 1.PEI et al. . the DFRFT reduces to DFT. DFRFT by GSA of an impulse function. Fig. 8. x(0) = 1. and the output becomes a constant. x(0) = 1. The coefﬁcients ’s are in (33). the diagonal matrix F. If the rotation angle is changed. . only should be recomputed. the DFRFT reduces to DFT. When = =2. we stress the major reason for this mismatch. and x(n) = 0 otherwise. the DFRFT reduces to DFT. and the output becomes a constant. A rigorous discussion for the mismatches of [17] has been presented in [16]. 6 Fig.

In the following discussion. In [31].1344 IEEE TRANSACTIONS ON SIGNAL PROCESSING. the DFRFT can be used to give the similar continuous transform results within the accuracy of the discrete ﬁnite vector approximation. and the rotation angle in [31] is valid for a certain discrete set of rotation angles. we will establish the relationship between the FRFT and the DFRFT. However. To begin with. The development of the new DFRFT is based on the same idea as (2). MAY 1999 Fig. VOL. If x is odd. Samples of continuous FRFT of an impulse function. x(0) = 1. sampling interval for the original continuous signal. In the DFRFT of [17]. 11. Therefore. the signal and transform results in [31] are discrete and periodic. it has been proved that the spread variance . In Proposition 3. Transform Range and Resolution of DFRFT In this subsection. While errors of the DFT Hermite eigenvectors will be reduced. which satisﬁes the eigenvalue and eigenfunction relationship as a continuous FRFT kernel. there will be four different eigenvalues for all the eigenvectors. and transform range in the time domain as well. b) is the (a. 5. we will discuss the transform range and resolution for the DFRFT (see Table IV). Moreover. a) where F (a. Then. TABLE III PROPERTIES OF DFRFT Unitary Angle additivity Time Inversion Periodicity Symmetry Eigenfunction Impulse Transform Parity F )3 = (F )01 = F0 F F =F F x(0n) = X (0n) F =F F (a. the proposed DFRFT has assigned a different eigenvalue for each eigenvector. b) = F (b. IV. and the frequency resolution is [23]. and there are some approximation errors in the DFT Hermite approaches inﬁnity. Here. X is even. the periods in the transform results change for different rotation angles. On the other hand. A. The the overall frequency range that the DFT can represent is equal to . we will review and understand the transform range and resolution for the is the conventional DFT. It is based on retaining the property of the DFT that a sampled periodic function transforms into a periodic function. it is also is the number of points of the discrete the time resolution. Thus. It should be noted that the number of eigenfunctions for the continuous FRFT in (2) is inﬁnite. NO. the approximation eigenvectors. The DFRFT developed in this paper is based on the mimicking of the eigenvalue–eigenfunction relationship that the continuous counterpart of this transform has with the normalized unit variance Hermite functions. In the conventional DFT analysis. and x(t) = 0 otherwise. X is odd. the number of DFT Hermite eigenvectors is only ﬁnite. the transform range and resolution of the DFT have been well discussed [23]. is the total recorded signal duration and is signal. .b)-element in DFRFT kernel matrix F [vn ] = e0jn vn where vn is Hermite-like function N F [(k)] 2 10j2cot ej cot If x is even. RELATIONSHIP BETWEEN FRFT AND DFRFT In this section. ( one of the four eigensubspace has the same eigenvalue. an alternative DFRFT has recently been proposed. and more DFT Hermite eigenvectors are used to compute the DFRFT. we will of the DFT Hermite eigenvector is compute the FRFT for a Hermite function with any variance . our DFRFT can be used for any rotation angle and can provide very similar results as continuous cases. 47. the method assigns the DFRFT eigenvalues to only four values. (41) is the eigenvector corresponding to the th-order where Hermite function.

but the DFRFT performs elliptical rotation in the continuous time–frequency plane while . Elliptical Rotation versus Circular Rotation The FRFT performs circular rotation in the continuous time–frequency plane. It is easy to check some special cases for (43): and . we will deﬁne circular and elliptical rotations clearly and establish the relationship between circular and elliptical rotations. 12. only the last term can affect the envelope . ( • The length of the time range is longer than that of the ). but it actually performs elliptical rotation in the continuous time–frequency plane. as drawn in Fig. it has been shown that there are approximation errors in the DFT Hermite eigenvectors. Three cases can be realized for the signal rotation of the DFRFT in the continuous time–frequency plane. frequency ( • The length of the time range is shorter than that of the ). In (42). 12. From (45) and Fig. The sampled spread variance of the DFRFT is vector in (12) can also be an approximate eigenvector in the fractional Fourier domain (angle ). is used to denote the resolution of the FRFT. Proposition 8: The resolution of the DFRFT with angular is equal to parameter (44) is the sampling interval of the signal. The overall transform range of the DFRFT can cover is equal to (45) Proof: From Proposition 3.: DISCRETE FRACTIONAL FOURIER TRANSFORM BASED ON ORTHOGONAL PROJECTIONS 1345 TABLE IV TRANSFORM RANGE AND RESOLUTION OF DFRFT Proposition 7: The continuous FRFT of a Hermite function with variance is equal to Moreover. In this subsection. it can be found that the signal rotation of the DFRFT is an elliptical rotation in . whereas This means that the DFRFT is implemented in a circular rotation for the discrete case. the spread variance of the DFRFT for angle is equal to (47) The overall transform range the DFRFT can cover is equal to (50) In (34). the DFRFT performs independent circular rotation in the discrete notation. B. will be equal to 1 for any value of . therefore. of the FRFT of From Propositions 4 and 5. and is the where number of points for the discrete signal. the approximation errors of the DFT Hermite eigenvectors are ignored in the following discussion. Angle actual angle for the elliptical rotation. The new variance is com- (43) Proof: This proposition can be easily proved by computing the FRFT of the normalized Hermite functions. For simpliﬁcation of analysis. the continuous time–frequency plane. frequency ( The rotation concept of the DFRFT in the continuous is the time–frequency plane is plotted in Fig. the resolution of the FRFT can be obtained as (49) (42) where puted as . • The lengths of time and frequency ranges are equal ). If . it has been known that the . 12.PEI et al. we can substitute the variance in (43) by to get the spread variance of the DFRFT in the fractional Fourier domain (angle ) (48) Both (47) and (48) indicate the corresponding variance of DFT Hermite eigenvectors in the fractional Fourier domain. • Circular Rotation: (51) . (46) Thus. The symbol and are the two special cases.

12. NO. While . MAY 1999 Fig.1346 IEEE TRANSACTIONS ON SIGNAL PROCESSING. 13. is the circular rotation of signal . 5. If the Hermite eigenvectors in the circular rotation are from order 0 to inﬁnity. and the post-phase compensa- is the transform kernel of elliptical rotation. 47. is the elliptical rotation . (52) (53) is the transform kernel of circular rotation. Results of Example 4 showing that the error norms between DFRFT and continuous FRFT samples are greatly reduced by angle modiﬁcation and post-phase compensation. the circular rotation is just the FRFT of the signal. The critical point for which we care is the (58) can be written as the weighted Proof: The signal sum of the normalized Hermite functions. (57) where tion factor is equal to . The transform kernels of the circular and elliptical rotations deﬁned above are inﬁnite numbers of sums for Hermite eigenvectors. and where is deﬁned as in (43). Three different cases of rotation concept for DFRFT in continuous time-frequency plane. • Elliptical Rotation: where (54) (55) (56) spread variance changes in these two schemes. VOL. (a) (b) (c) Fig. (59) . the elliptical rotation will of signal become the circular rotation. Proposition 9: The circular rotation of signal can be implemented by an angle modiﬁcation of the elliptical rotation and post-multiplying a phase compensation factor.

In Example 2. The correis equal to 1. The results shown in Fig. Fig. whereas the OPA minimizes the total errors between the samples of the Hermite functions and the orthogonal DFT Hermite eigenvectors. it will become evaluated at the points (60) (61) can be Equation (61) indicates that the FRFT with angle implemented by a DFRFT with angle . CONCLUSIONS The development of this DFRFT is based on the eigendecomposition of the DFT kernel matrix . However. It must be noted that the postphase where compensation factors in (58) and (62) are different. With the help of the commutative matrix . The new transform retains the eigenvalue–eigenfunction relationship using the sampled version of the normalized unit-variance Hermite functions that the continuous FRFT has with the unit variance Hermite functions. While the variance is far from unity. 13(a) is quite different from Fig. 13(b) is the DFRFT with angular parameter modiﬁcation and postphase compensation. the DFRFT in Example 2 is almost a circular rotation. Compute the FRFT of (circular rotation) for the both sides of (59). the sampling . .PEI et al. we directly compute the DFRFT without angle modiﬁcation and postphase compensation. The GSA minimizes the errors between the samples of the Hermite functions and the orthogonal DFT Hermite eigenvectors from low to high orders. and the number of points in the interval is still signal becomes 37. 13(c) very well. we only compute the results of by the GSA. Fig. Therefore. The operation in the DFRFT performs elliptical rotation in the continuous time–frequency plane for the case . and the effects of angle modiﬁcation and postphase compensation will be very small. V. However. In this case. Fig. the the unitary property in the DFRFT. 13(c) due to the elliptical rotation and not in the circular rotation in the continuous time–frequency plane. Using Proposition 9. the modiﬁcation of the angle and postphase compensation discussed above are very critical.: DISCRETE FRACTIONAL FOURIER TRANSFORM BASED ON ORTHOGONAL PROJECTIONS 1347 where . Here. the complete real and orthonormal eigenvectors of the DFT kernel matrix can be computed. Step 1) Compute the modiﬁcation angle . the elliptical rotation can be implemented by circular rotation. The rotation angle . The length of the time range is almost equal to the length of the frequency range. such DFT Hermite eigenvectors cannot form an orthogonal basis for the DFT eigenspaces. and Fig. where . which are equal to continuous counterpart of the variance in this example is .0488 in Example sponding variance 2. Step 2) Calculate the -point DFRFT of the signal with the angular parameter . The in (58) is replaced by in (62) for preserving variable . Two vector orthogonalization processes for the DFT Hermite eigenvectors are accordingly proposed in this paper: One is GSA. the following steps must be used to make the results of discrete cases match those of the continuous case. 13(c) shows the sample values of the continuous FRFT . Step 3) Multiply the result obtained from Step 2 by the post-phase compensation factor shown in (62). and the other is OPA. therefore. 13(b) match the corresponding continuous FRFT cases shown in Fig. The discrete post-phase compensation resolution is still factor for the DFRFT is the length of time range ( ) will almost equal the length ). we again deal with the rectangular window shown in Example 2. The rotation will almost be of the frequency range ( a circular rotation. for the indices . The DFT Hermite eigenvectors can be calculated by the projection of samples of the unit variance Hermite functions in the DFT eigensubspaces through the help of the eigenvectors of the matrix . 13(a) shows the DFRFT with angular parameter for the original signal. If the values of (57) are . and the transform . If variance of the continuous-time counterpart of the eigenvector will approximate to unity. (62) . Example 4: In this example.

” J. Lohmann. G.” Opt. IEEE Int. “Fractional Fourier transformations and their optical implementation—I.” IEEE Signal Processing Lett. Mendlovic and H. Lett. His research interests include digital signal processing. “On Namias’ fractional Fourier transforms. Kutay. Taiwan. 303–309. [3] V. and L. “Linear and quadratic timefrequency signal representations. Namias.. From 1995 to 1997. in 1988. Soc. IEEE Int. DFRFT can be established as follows.E. McBride and F.-C. “Chirp ﬁltering in the fractional Fourier domain. Feb. NJ: Prentice-Hall.O. 44. Soo-Chang Pei (SM’89) was born in Soo-Auo. 1987. pp. Santarsiero.” IEEE Trans. Spence.” J. H. [4] F.E. 25. 39. Taipei. Taiwan. Circuits Syst. pp. pp. Ozaktas. Mendlovic. 45. “Eigenvectors and functions of the discrete Fourier transform.” J. Hsinchu. vol. Taipei. [13] H. image processing. Hlawatsch and G. Apr.E. Signal Processing. Soc. degree in computer engineering from the National Chiao-Tung University. 1989... 3. A. Tamsui. ASSP-30. and R. Opt. 994–998. Van Loan. 205–207. pp. 10. and Ph. and J. Chien-Cheng Tseng (S’90–M’95) was born in Taipei. H. Taipei. . Opt. N. Ltd. 34. C.1348 IEEE TRANSACTIONS ON SIGNAL PROCESSING. His research interests include digital signal processing. [18] G. respectively. “Convolution. Opt. An angle modiﬁcation and a postphase compensation in the DFRFT for elliptical rotation are required to match the results that are similar for the continuous FRFT. respectively. M. Amer. and electronic commerce. 1047–1049. M. A. G. Kutay. Lohmann. 40–41. Linear Algebra.S. Golub and C. p.” IEEE Trans. pp. R. Orthogonal Functions. 159–175. From 1971 to 1975. he is Professor and Chairman of the Electrical Engineering Department. Bourdeaux-Bartels. both from the National Taiwan University. pp. McClellan. 547–559. pp. M. Tseng. M. E. Mendlovic. A. National Kaohsiung First University of Science and Technology. and G. 5. REFERENCES [1] R.O. vol. in 1972 and 1975. Rev. 1994. M. Sept. Shenoy. pp. Lohmann and B. Amer. 1980. in 1964. “Generalized self-Fourier functions.” J. Taipei. and M. Audio Electroacoust. M. “The discrete rotational Fourier transform. vol. From 1997 to 1998. Signal Processing. 25. pp.. [5] H. M. Interscience. Presently. Amer. pp. Feschbach. T. 1959. in 1949. 11. He is currently an Assistant Professor with the Department of Computer and Communication Engineering. vol. Taipei. pp. “Digital computation of the fractional Fourier transform. 11. 1972. 1994. which is the same as that implementing the DFRFT is of the DFT.” Appl. 15–16. “Convolution and ﬁltering in fractional Fourier domain.” IEEE Trans. K. Feb. New York.O. 2nd ed. degree in computer science and information engineering in 1992 and the Ph. 1875–1881. “Effect of fractional Fourier transformation on time-frequency distributions belonging to the Cohen class.S.. “Improved discrete fractional Fourier transform. W...” in Proc. H. 1986. optical information processing. Taoyuan. pp.. [6] H. 7599–7602. The DFRFT proposed in this paper not only supplies the similar transforms to match with those of the continuous case but also preserves the rotation properties. Erkaya. “The fractional order Fourier transform and its application to quantum mechanics. Signal Processing. [16] H. Arikan... Taipei. Aademir. 66–74. Nov. from the Tatung Institute of Technology. Soc. “Separation of transient signals. Acoust. Cincotti. Time-Scale Anal.D. and O. “Discrete fractional Fourier transform. A. 46. W. A.” IEEE Signal Processing Mag. “Eigenvalue and eigenvector decomposition of the discrete Fourier transform. 3084–3091. 199–203. H. Min-Hung Yeh was born in Taipei.” in Proc.” in Proc. Santhanam and J. Mendlovic. Opt. Appl. July 15. 13. Amer. B. [21] J. Opt. A. “Relationships between the Radon–Wigner and fractional Fourier transforms. time–frequency plane. D. [31] O. Jan.. .D. [25] 22. Ozaktas. 1995.C. H. pp. M. NJ: Prentice-Hall.S. vol. pp. Tatung Institute of Technology. pp. all in electrical engineering.” Signal [26] Process. Circuits Syst. “Fractional correlation. vol. [30] G. vol. [7] R. and D. 2522–2531. [2] A. MAY 1999 Furthermore. M. H.” Appl. vol. vol. If the DFRFT performs a circular rotation of the signal in the . vol. Barshan. 665–675. NTU.” J. MD: Johns Hopkins Univ. 25–31. Parks. Mar. vol.S. This DFRFT provides a method for implementing the discrete FRFT. 42. O. Kutay. NO. Pei is a Member of Eta Kappa Nu and the Optical Society of America. New York: McGraw-Hill. and D. Baltimore. A. Ozaktas. M. pp. Taiwan. J. vol. Y.C. 2141–2150. J. Oppenheim.. He received the B. Math. in 1970 and the M. A. Ozaktas. 9. Parks. vol. He received the B. and the M. “The discrete fractional Fourier transform. II. M. [12] D. 1989. “Two dimensional discrete fractional Fourier transform. Dickinson and K. and laser holography. . C.” IMA J. Math.E. vol. Ozaktas. His main research interests are in fractional Fourier transforms. McClellan and T. and M. VOL. pp. N. he was a Research Assistant at UCSB. 536–539. pp. Santa Barbara (UCSB). Soc. 1995.. 119–124. degree from the National Taiwan University (NTU). 1798–1801.D. Chunghwa Telecom Company. [27] S. Apr. V. he was an Assistant Professor of the Department of Electronics Engineering. 1994. Taipei. New York: McGraw-Hill. vol. Oct. Soffer. if the DFRFT becomes an elliptical rotation in the continuous time–frequency plane. [29] S. Symp. vol. in 1987. Dr.. Yeh. vol. pp. 474–478.” J. Ozaktas. C. He received the B. Tamsui Oxford University College.. “Fractional Fourier domains. 1993. B. On the other hand. Opt. Taipei. “Fractional Fourier transformations and their optical implementation—II. F. 10. Yeh. Speech. Friedberg. He is currently an Assistant Professor with the Department of Computer Information Science. H. degrees from the National Taiwan University. W. June 1996.. with honors. M. 1992. 1982. degrees from the University of California. pp. 1416.” Opt. and L. 1996. vol. W. in 1965. time–frequency analysis. A. degree. Onural. 1994. degree in electrical engineering in 1997. 42. Morse and H. [19] B. F. 1953.” IEEE Trans. Pei and M. Ozaktas. [20] B. B. Steiglitz. 1996. 67.” IEEE Trans. Inst.. Gori. 241–265. Amer. he was an Associate Research Engineer at Telecommunication Laboratories. 99–108. Phys. Barshan. Time–Freq. [9] L. Arikan. 6th IEEE DSP Workshop. Dorsch. A. W. June 1994. 1989.-H. May 1996. pp. 1996. Applicat. Signal Processing. [10] D. in 1990 and 1995. H. R. The complexity for . Almeida. and A. Opt. from 1981 to 1983. 1998. Englewood Cliffs. [22] G. [17] B. He was a Professor and Chairman with the Electrical Engineering Department. A. H. vol. [23] A. Bozdagi. the relationship between the FRFT and the . Bracewell. Bitran. F.S. Mendlovic. and wavelets. C. R. Symp.S. Mar. pattern recognition. Sansone. pp. and Ph. pp. Weisburn. The Fourier Transform and Its Applications. ﬁltering. 1997. pp. Shyu. Ozaktas. 33.. 1994. vol.. [14] D. 47. Kerr. 1992. Methods of Theoretical Physics. vol. [8] A. Soc. pp. 1191–1194. vol.” Signal Process. Insel. “The fractional Fourier transform and time-frequency representation. He was an Engineering Ofﬁcer in the Chinese Navy Shipyard from 1970 to 1971. and it is an important tool for signal processing. and multiplexing in fractional Fourier domains and their relationship to chirp and wavelet transforms. He then received the M. 1996. Taiwan. Pei. 21–67. Dragonman.. “Discrete fractional Hartley and Fourier transforms. 1993. [24] S. [28] P. [15] . 1998. “Fractional Wigner distribution function. Matrix Computations.” J. [11] H. Englewood Cliffs.” IEEE Trans.C. AU-20. Taiwan. 1. Press. Feb. Hwa Hsia College of Technology and Commerce. W. Discrete-Time Signal Processing.

- 3 Equality
- Venus & Adonis - Shakespeare
- Troilus & Cressida - Shakespeare
- Tempest - Shakespeare
- Hamlet by William Shakespeare
- Phoenix & Turtle - Shakespeare
- Passionate Pilgrim - Shakespeare
- Othello - Shakespeare
- Merry Wives of Wndsor - Shakespeare
- Merchant of Venice - Shakespeare
- Lucrece - Shakespeare
- Love's Labours Lost - Shakespeare
- Julius Caeser - Shakespeare
- Comedy of Errors - Shakespeare
- As You Like It - Shakespeare
- Antony & Cleopatra - Shakespeare
- All is Well That Ends Well - Shakespeare
- UBUNTU8.04TUT
- PC Magazine - Linux Solutions Malestrom
- Linux Sys Calls
- Linux Drivers
- Linux Unix Concepts
- Intro Linux Device Drivers
- Embedded Linux Kernel and Drivers
- Arcom Embedded Linux Technical Manual

- Symbolic Analysis of Time Series Signals Using Generalized Hilbert Transform
- lecture490_ch7
- 2.Continuous Wavelet Techniques
- Digital Signal Processing
- Symbolic Computation via Gröbner Basis
- Theory of FRAP D. E. Wolf
- Jun 2004 Hydraulic Res
- YeckelA&DerbyJJ 1999 PressureDatumForIncompFlows
- tmp4811.tmp
- The AdSCFT Correspondence and a New
- La Place Transforms
- 0504536v1
- Hofstadter Butterfly
- Rotations 2
- On the Channel Capacity of Multiantenna Systems With Nakagami Fading
- 4. Maths - Ijamss - Neimark Sacker Bifurcation in Delayed Logistic Map Hemanta
- Digital Resonators
- Relativistic Quantum Fields 1
- Modal Analysis of Continuous Rotor-Bearing Systems
- Add_maths_(Normal Track) Sow_year 11 (310807) v2 - 3 years
- Gravity in Curved Phase Spaces and Two Times Physics
- LIBRO ANALISIS SISMICO_170_175.pdf
- State Space
- art%3A10.1023%2FA%3A1022506312444 (1)
- Johannes Brodel and Bernhard Wurm-New Twistor String Theories revisited
- R. Loll, W. Westra and S. Zohren- Nonperturbative sum over topologies in 2D Lorentzian quantum gravity
- tmp7B55
- Born's Reciprocal Gravity in Curved Phase-Spaces and the Cosmological Constant
- 28
- Chirp z Transform Algorithm

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue listening from where you left off, or restart the preview.

scribd