HomeChair: James H. McClellan, Georgia Institute of Technology, USA
HomeNaofal Al-Dhahir, GE Corporate R&D (U.S.A.)
Optimum FIR transmit filters for symbol-by-symbol transmission on linear dispersive additive-Gaussian-noise channels are derived by maximizing the channel throughput, subject to a fixed input energy constraint. This maximized throughput is compared with that achievable with water-pour and flat transmit filters. The effect of transmit filter optimization on the receiver performance is investigated by considering the popular MMSE-DFE receiver structure.
ic981012.pdf (From Postscript)
TOP
Ba-Ngu Vo, Chinese University of Hong Kong (Hong Kong)
Antonio Cantoni, ATRI Curtin University of Technology (Australia)
In an envelope-constrained filtering problem with uncertain input the set of possible inputs and the set of permissible outputs are each defined by envelopes or masks. This paper considers a continuous-time filter which in structure is comprised of an A/D converter, an FIR filter, a D/A converter and an analog post-filter. The object is to design the digital component of the filter structure so as to minimize the noise enhancement whilst satisfying the constraint that every signal in the input envelope evokes a response which stays in the output envelope.
ic981019.pdf (From Postscript) TOPScott T Acton, Oklahoma State University (U.S.A.)
This paper investigates the relationship between anisotropic diffusion and local monotonicity. A diffusion technique that has locally monotonic root signals is presented. The enhancement algorithm rapidly converges to a locally monotonic signal of the desired degree. It is shown that the diffusion coefficient used here is the only formation that guarantees idempotence for locally monotonic signals. The signals resulting from locally monotonic diffusion are closer to the original signals than the corresponding median root signals. Furthermore, the diffusion algorithm does not have a difficulty with alternating signals, as does the median filter. In contrast to other anisotropic diffusion techniques, the diffusion method given here does not preserve outliers and does not require a gradient magnitude threshold in the diffusion coefficient.
ic981231.pdf (From Postscript) TOP
Sven Nordebo, Curtin University of Technology (Australia)
Zhuquan Zang, Curtin University of Technology (Australia)
Previously the envelope-constrained filtering problem was formulated as designing an FIR filter such that the filter's L-2 norm is minimized subject to the constraint that its response to a specified input pulse lies within a prescribed envelope. In this paper, we recast this filter design problem as a frequency-domain L-infinity optimization problem with time-domain constraints. Motivations for solving this problem are given. Then recently developed infinite dimensional linear programming techniques are used for the design of the required FIR filter. For illustration, we apply the approach to a numerical example which deals with the design of an equalization filter for a digital transmission channel.
ic981635.pdf (From Postscript) TOP
Worayot Lertniphonphun, Georgia Institute of Technology (U.S.A.)
James H. McClellan, Georgia Institute of Technology (U.S.A.)
This paper provides an algorithm for designing FIR filters that approximate both magnitude and phase of the frequency response. The new algorithm produces a filter optimized under the weighted Chebyshev norm. The algorithm starts from a first stage unoptimized filter designed by a Remez-like algorithm and then uses shifted Dirichlet kernel functions to reduce large error peaks and converge to an equiripple set of peaks. The error function is modified at each iteration by subtracting a best-fit linear combination of kernel functions due to the large error peak(s). For one length-100 example, the computation of this algorithm was less than that of the complex Remez by two orders of magnitude.
ic982060.pdf (From Postscript) TOP
Uwe Meyer-Baese, University of Florida (U.S.A.)
Jon Mellott, University of Florida (U.S.A.)
Fred Taylor, University of Florida (U.S.A.)
Algebraic integers have been proven beneficial to DFT and non-recursive FIR filter designs since algebraicintegers can be dense in \BbbC, resulting in short word width, high speed designs. This paper uses another property of algebraic integers:algebraic integers can produce exact pole zero cancellation pairs that are used in recursive FIR, frequency sampling filter designs.
ic982077.pdf (From Postscript) TOP
Dietmar W Schill, University of Erlangen (Germany)
Andre Marguinaud, Alcatel Espace (France)
In approximate and real time processing, one encounters the problem of making efficient use of the limited processing power available. Furthermore it can be desirable to enable a system to react dynamically to a change in requirements. For digital filters this implies the possibility of calculating filter coefficients on the fly with low complexity algorithms. Such an algorithm is presented for the design of linear phase FIR lowpass filters. It has the additional property that subsets of coefficients of one filter constitute by themselves filters of reduced stop band attenuation and/or lower bandwidth reduction.
ic982121.pdf (From Postscript) TOPJeffrey O Coleman, Naval Research Laboratory (U.S.A.)
A systematic approach is proposed for the individual or joint design of FIR filters to meet specifications on either a single filter or an embedding system (possibly multirate). System power gains in response to particular input spectra are optimized using a generalized eigenvector method. Numerical integration is avoided through a time-domain formulation. Real or complex filters with linear or nonlinear phase or N-th band properties are easily handled.
ic982146.pdf (From Postscript) TOP
Tania Stathaki, Imperial College (U.K.)
Anthony Constantinides, Imperial College (U.K.)
Georgios Stathakis, Imperial College (U.K.)
In this contribution we propose a method for a minimum phase Finite Impulse Response (FIR) filter design from a given linear phase FIR function with the same amplitude response. We are concentrating on very high degree polynomials for which factorisation procedures for root extraction are unreliable. The approach taken involves the use Cauchy Residue Theorem applied to the logarithmic derivative of the transfer function. This leads into a set of parameters derivable directly from the polynomial coefficients which facilitate the factorisation problem. The concept is developed in a way that leads naturally to the celebrated Newton Identities. In addition to solving the above problem, the results of the proposed design scheme are very encouraging as far as robustness and computational complexity are concerned.
ic985088.pdf (Scanned) TOPPaulo J.S.G. Ferreira, Universidade de Aveiro (Portugal)
The importance of lattice structures in connection with filtering and prediction has been known for decades. The demand for faster processing has led to steadily increasing sampling rates, and as a result the behavior of the discrete filters as the sampling period tends to zero has become an important theoretical and practical issue. One way of solving the numerical problems that arise in the usual filter structures when the sampling period becomes small compared with the dynamics of the underlying physical processes is to resort to (delta) operators instead of delay operators. Although the interrelations between the continuous and discrete lattice structures have been rarely studied, it is known that the (delta) lattice naturally leads to a continuous form as the sampling rate increases. This paper addresses this point and establishes the rate of convergence of the discrete lattice filter to the continuous filter as a function of the sampling period or of the filter order.
ic982435.pdf (From Postscript) TOP