Chia-Chang Wen

National Taiwan University, T’ai-pei, Taipei, Taiwan

Are you Chia-Chang Wen?

Claim your profile

Publications (6)10.84 Total impact

  • Soo-Chang Pei · Chia-Chang Wen · Jian-Jiun Ding
    [Show abstract] [Hide abstract]
    ABSTRACT: A new conjugate symmetric discrete orthogonal transform (CS-DOT)-generating method is proposed. The spectra of the CS-DOT for real input signals are conjugate symmetric so that we only need half memory size to store data. Meanwhile, the proposed CS-DOT also has a radix-2 fast algorithm so that it is suitable for hardware implementation. The CS-DOT generalized the existing transforms such that the discrete Fourier transform (DFT) and the conjugate symmetric sequency-ordered complex Hadamard transform (CS-SCHT) are special cases of the CS-DOT. The CS-DOT-generating method is more systematic and generalized than that of the original CS-SCHT. We can use the same implementation structure but only adjust the twiddle factors to construct the CS-SCHT and DFT so that it is easy to switch the behaviors between these transforms.
    No preview · Article · Apr 2014 · Circuits and Systems II: Express Briefs, IEEE Transactions on
  • Soo-Chang Pei · Chia-Chang Wen · Jian-Jiun Ding
    [Show abstract] [Hide abstract]
    ABSTRACT: A new transform family, called the sequency-ordered generalized Walsh–Fourier transform (SGWFT), is proposed in this paper. Using the kernel matrix generation process and the controllable phase quantization parameter, the Walsh–Hadamard transform (WHT), the sequency-ordered Hadamard transform (SCHT), and the discrete Fourier transform (DFT) become special cases of the SGWFT. The SGWFT can be adjusted by a single parameter to become the WHT, the SCHT, and the DFT. In addition, the SGWFT also has the radix-2 and the split-radix fast algorithms. Compared with the WHT and the SCHT, the properties and the performance of the SGWFT are more similar to those of the DFT. On the other hand, compared with the DFT, the number of multiplications in the SGWFT is less. We also show that the proposed SGWFT has better performance in the applications of DS-CDMA sequence design and transform coding.
    No preview · Article · Apr 2013 · Signal Processing
  • Soo-Chang Pei · Chia-Chang Wen
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we propose a new family orthogonal transforms defined over finite field called the Finite Field Orthogonal Transforms (FFOT). Unlike the traditional Number Theoretic Transform (NTT) that the relationship between the transform length and the field moduli has specific constraint in order to hold the orthogonality property, the FFOT has no such constraint so that the signal word length need not be limited by the transform length. In addition, the fast algorithm implementation like radix-2 Cooley-Tukey algorithm is also realizable for the FFOT and is suitable for fast data encryption.
    No preview · Conference Paper · Jan 2012
  • Soo-Chang Pei · Pai-Wei Wang · Jian-Jiun Ding · Chia-Chang Wen
    [Show abstract] [Hide abstract]
    ABSTRACT: The S transform, which is one of the time-frequency transforms, has been shown to be useful in time frequency analysis and many signal processing applications. The discrete counterpart of the S transform (DST) can be implemented in both the time and the frequency domains, i.e., the Time DST and the Freq DST. However, previous studies found that the conventional Time DST and the conventional Freq DST are not consistent, which may result in unreliable time-frequency information. In this paper, a new DST that adopts the folded window is proposed to eliminate the side effects of discretizing. The consistence of the time and the frequency versions is an important property of the continuous S transform and the proposed DST inherits this property. The proposed folded window can also be applied to the DST whose window is not a Gaussian function and the short-time Fourier transform.
    No preview · Article · Jun 2011 · Signal Processing
  • Soo-Chang Pei · Chia-Chang Wen · Jian-Jiun Ding
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we propose a new method to find the closed-form solution of Number Theoretic Transform (NTT) eigenvectors. We construct the complete generalized Legendre sequence over the finite field (CGLSF) and use it to solve the NTT eigenvector problem. We derive the CGLSF-like NTT eigenvectors successfully, including the case where the operation field is defined over the Fermat and Mersenne numbers. The derived NTT eigenvector set is orthogonal and has a closed form. It is suitable for constructing sub-NTT building blocks for NTT implementation. In addition, with different eigenvalue assignment rule, we can construct the fractional number theoretic transform (FNTT), including the fractional Fermat number transform (FFNT), the fractional complex Mersenne number transform (FCMNT), and the fractional new Mersenne number transform (FNMNT). They are the generalizations of the original transforms and all have the complexities of O(Nlog2N).
    No preview · Article · May 2011 · IEEE Transactions on Signal Processing
  • Source
    Soo-Chang Pei · Chia-Chang Wen · Jian-Jiun Ding
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we propose a new method for deriving the closed-form orthogonal discrete Fourier transform (DFT) eigenvectors of arbitrary length using the complete generalized Legendre sequence (CGLS). From the eigenvectors, we then develop a novel method for computing the DFT. By taking a specific eigendecomposition to the DFT matrix, after proper arrangement, we can derive a new fast DFT algorithm with systematic construction of an arbitrary length that reduces the number of multiplications needed as compared with the existing fast algorithm. Moreover, we can also use the proposed CGLS-like DFT eigenvectors to define a new type of the discrete fractional Fourier transform, which is efficient in implementation and effective for encryption and OFDM.
    Full-text · Article · Jan 2009 · Circuits and Systems I: Regular Papers, IEEE Transactions on