Chair: Maurice Bellanger, CNAM/Electronique (FRANCE)
Santiago Zazo, Universidad Alfonso
J. M. Paez- Borrallo, Universidad Politecnica de Madrid (SPAIN)
Ivan A. Perez Alvarez, Universidad Politecnica de Madrid (SPAIN)
Existing blind adaptive equalizers that use nonconvex cost functions (as Bussgang type algorithms) and stochastic gradient descent suffer from lack of global convergence to an equalizer tap set that removes sufficient ISI when an FIR equalizer is used. In this paper we propose a new algorithm including a tap anchoring and gain recovering into the classical schemes. The combined effect of these strategies is to establish the preservation of the transmitted symbol preventing for ill convergence, and therefore providing the ability of implementation the inverse filter regardless of the initial ISI. Under certain hypotheses, we suggest that a globally convex scheme can be proposed overcoming the existing structures. Several computer simulations support our theoretical results.
James P. LeBlanc, Cornell University (USA)
Inbar Fijalkow, ENSEA (FRANCE)
Birkett Huber, Cornell University (USA)
C. Richard Johnson Jr., Cornell University (USA)
CMA Fractionally Spaced Equalizers (CMA-FSEs) have been shown, under certain conditions, to be globally asymptotically convergent to a setting which provides perfect equalization. Such a result relies heavily on the assumptions of a white source and no channel noise (as is the case in much of the literature's analysis of CMA). Herein, we relax the white source assumption and examine the effect of source correlation on CMA. Analytic results are meshed with examples showing CMA-FSE source correlation effects. Techniques for finding all stationary and saddle points on the CMA-FSE error surface are presented using recent developments in the Algebraic-Geometry community.
P. Philippe, C.C.E.T.T. (FRANCE)
F. Moreau de Saint-Martin, C.C.E.T.T. (FRANCE)
L. Mainard, C.C.E.T.T. (FRANCE)
We address the issue of choosing an optimal wavelet packets transform for audio compression. We present a comparison method based on a perceptual approach which provides an entropic bit-rate for ``transparent'' coding of a given audio signal. The test with different wavelets leads to the conclusion that the most significant synthesis criterion for audio compression is the so-called ``coding gain'', while frequency selectivity, regularity and orthogonality seem less relevant.
T. Schirtzinger, University of Illinois (USA)
X. Li, University of Illinois (USA)
W.K. Jenkins, University of Illinois (USA)
Three constant modulus algorithms (CMA), the fast quasi-Newton CMA, the transform domain CMA, and the genetic search based CMA are proposed in this paper. The performances of these three algorithms are compared with each other via computer simulation. It is shown that the fast quasi-Newton CMA and the transform domain CMA achieve much faster convergence rates that the constant modulus algorithm based on the LMS algorithm. This fact shows that the whitening technique is not only useful, but also necessary for the CMA.
Kaywan H. Afkhamie, McMaster University (CANADA)
Zhi-Quan Luo, McMaster University (CANADA)
When a message signal is transmitted through a linear dispersive system, the system output may contain severe intersymbol interference (ISI). The removal of ISI without the aid of training signals is referred to as blind equalization. We present a new algorithm that achieves blind equalization of possibly nonminimum phase channels, based only on the second-order statistics of the source symbols. Source symbols may have an arbitrary distribution; specifically, they do not have to be independently identically distributed (i.i.d.). This is an extension to previous work done be Tong, Xu and Kailath [1]. Simulations show that the new algorithm compares favorably to the algorithm given in [1].
Michael L. Honig, Northwestern University (USA)
Minimum Mean Squared Error (MMSE) detection has been recently proposed for Direct Sequence-Code Division Multiple Access (DS-CDMA) systems. MMSE detectors are near-far resistant, and can be adapted with standard adaptive algorithms without knowledge of user parameters (i.e., spreading codes). These algorithms rely on a known training sequence for initial adaptation, and subsequently switch to decision-directed mode. After the switch, the performance of the adaptive algorithm may degrade substantially if a strong interferer suddenly appears (i.e., if power control is relaxed). We present a ``rescue'' algorithm that monitors for sudden changes in the signal space, which may be caused by the appearance of a strong interferer. If a new interferer is detected, decision-directed adaptation is suspended, and an estimate of the optimal filter coefficients is obtained without a training sequence. It is shown that in the presence of low-level background noise, a good estimate can be obtained within a few symbol intervals. A numerical example is given which illustrates the performance of the rescue algorithm in a synchronous DS-CDMA system.
Laura Ortiz-Balbuena, Universidad Autonoma Metropolitana Iztapalapa (MEXICO)
Alejandro Martinez-Gonzalez, Universidad Autonoma Metropolitana Iztapalapa (MEXICO)
Hector Perez-Meana, Universidad Autonoma Metropolitana Iztapalapa (MEXICO)
Luis Nino de Riveria, Universidad Autonoma Metropolitana Iztapalapa (MEXICO)
Ramirez-Angulo, New Mexico State University (USA)
Adaptive filters have been traditionally developed in a digital environment which involves large number of computations to get the coefficients that make the desired approximation. Most of the time, this calculations required a great capacity machines and that is not practical for some applications like channel equalization incellular systems. This paper proposes a continuos-time adaptive filter which is based on representing the impulse response of adaptive filter as a linear combination of a set of orthogonal exponentials. An important practical advantage of it is that if a satisfactory representation can be obtained by exponentials and simple filter structures can be synthesized.
Milos Doroslovacki, Compunetix Inc.
H. (Howard) Fan, University of Cincinnati (USA)
Using a small number of coefficients in Haar-wavelet- based models we can efficiently identify echo paths which have certain typical impulse response shapes. The obtained energy of modeling error is low (less than 2%). A simple wavelet-based LMS adaptive filter can be used for on-line estimation of coefficients. A low number of time-consuming computations is obtained per input sample due to the usage of Haar wavelets. This number is less than the ones obtained by FIR or DFT domain based modeling.
T.S. Castelein, Northern Telecom (THE NETHERLANDS)
Y. Bar-Ness, New Jersey Institute of Technology (USA)
R. Prasad, Delft University of Technology (THE NETHERLANDS)
High-speed communications suffer from ISI introduced by the channel. In order to combat ISI one commonly uses equalizers that need a training sequence to adjust the equalizer tap weights. When sending a training sequence is not appropriate, blind equalization has to be used. In this paper, a new blind linear equalizer is proposed. The equalizer has a recursive structure. To avoid stability problems, a soft limiter is used in the feedback loop. The weights are controlled by means of the decorrelation algorithm. For channels without precursors, the equalizer converges to the desired weights in around 500 iterations, depending on the amount of distortion. The novel blind equalization technique is globally convergent for minimum and non-minimum phase channels.
Sylvie Marcos, CNRS-ESE (FRANCE)
Sofiane Cherif, ENIT/ESPTT (TUNISIA)
Meriem Jaidane, ENIT/ESPTT (TUNISIA)
This paper presents the cancellation of intersymbols interferences= as a criterion for the blind adaptation of decision feedback equalizers= (DFE).=20 We show that this criterion is an alternative to the decision directed algorithm. This paper also proposes to replace the hard limiter in the decision= device of DFE by a soft decision implemented by a hyperbolic tangent function= in order to escape from local minima during the initialization of the blind algorithms. As the theoretical investigation of these criteria and= algorithms is difficult, we here analyse some simple examples to illustrate the= interest of the proposed solutions.