Search

This paper proposes efficient batch-based and online strategies for kernel regression over graphs (KRG). The proposed algorithms do not require the input signal to be a graph signal, whereas the target signal is defined over the graph. We first use random Fourier features (RFF) to tackle the complexity issues associated with kernel methods employed in the conventional KRG. For batch-based approaches, we also propose an implementation that reduces complexity by avoiding the inversion of large matrices. Then, we derive two distinct online strategies using RFF, namely, the mini-batch gradient KRG (MGKRG) and the recursive least squares KRG (RLSKRG). The stochastic-gradient KRG (SGKRG) is introduced as a particular case of the MGKRG. The MGKRG and the SGKRG are low-complexity algorithms that employ stochastic gradient approximations in the regression-parameter update. The RLSKRG is a recursive implementation of the RFF-based batch KRG. A detailed stability analysis is provided for the proposed online algorithms, including convergence conditions in both mean and mean-squared senses. A discussion on complexity is also provided. Numerical simulations include a synthesized-data experiment and real-data experiments on temperature prediction, brain activity estimation, and image reconstruction. Results show that the RFF-based batch implementation offers competitive performance with a reduced computational burden when compared to the conventional KRG. The MGKRG offers a convenient trade-off between performance and complexity by varying the number of mini-batch samples. The RLSKRG has a faster convergence than the MGKRG and matches the performance of the batch implementation.

This paper develops nonlinear kernel adaptive filtering algorithms based on the set-membership filtering (SMF) framework. The set-membership-based filtering approach is distinct from the conventional adaptive filtering approaches in that it aims for the filtering error being bounded in magnitude, as opposed to seeking to minimize the time average or ensemble average of the squared errors. The proposed kernel SMF algorithms feature selective updates of their parameter estimates by making discerning use of the input data, and selective increase of the dimension in the kernel expansion. These result in less computational cost and faster tracking without compromising the mean-squared error performance. We show, through convergence analysis, that the sequences of parameter estimates of our proposed algorithms are convergent, and the filtering error is asymptotically upper bounded in magnitude. Simulations are performed which show clearly the advantages of the proposed algorithms in terms of lower computational complexity, reduced dictionary size, and steady-state mean-squared errors comparable to existing algorithms.

Distributed estimation of a common unknown parameter vector can be realized efficiently and robustly over an adaptive network employing diffusion strategies. In the adapt-then-combine implementation of these strategies, each node combines the intermediate estimates of the nodes within its closed neighborhood. This requires the nodes to transmit their intermediate estimates to all their neighbors after each update. In this paper, we consider transmitting a subset of the entries of the intermediate estimate vectors and examine two different schemes for selecting the transmitted entries at each iteration. Accordingly, we propose a partial-diffusion least mean-square (PDLMS) algorithm that reduces the internode communications while retaining the benefits of cooperation and provides a convenient trade-off between communication cost and estimation performance. Through analysis, we show that the PDLMS algorithm is asymptotically unbiased and converges in the mean-square sense. We also calculate its theoretical transient and steady-state mean-square deviation. Our numerical studies corroborate the effectiveness of the PDLMS algorithm and show a good agreement between analytical performance predictions and experimental observations Refereed/Peer-reviewed

Full-duplex relaying is more spectrally efficient than half-duplex relaying as only one channel use is needed per two hops. However, it is crucial to minimize relay self-interference to render full duplex feasible. For this purpose, we analyze a broad range of multiple-input multiple-output (MIMO) mitigation schemes: natural isolation, time-domain cancellation, and spatial suppression. Cancellation subtracts replicated interference signal from the relay input while suppression reserves spatial dimensions for receive and transmit filtering. Spatial suppression can be achieved by antenna subset selection, null-space projection, i.e., receiving and transmitting in orthogonal subspaces, or joint transmit and receive beam selection to support more spatial streams by choosing the minimum eigenmodes for overlapping subspaces. In addition, minimum mean square error (MMSE) filtering can be employed to maintain the desired signal quality, which is inherent for cancellation, and the combination of time- and spatial-domain processing may be better than either alone. Targeting at minimal interference power, we solve optimal filters for each scheme in the cases of joint, separate and independent design. The performance of mitigation schemes is evaluated and compared by simulations. The results confirm that self-interference can be mitigated effectively also in the presence of imperfect side information.

Multichannel fast QR decomposition RLS (MC-FQRD-RLS) algorithms are well known for their good numerical properties and low computational complexity. The main limitation is that they lack an explicit weight vector term, limiting themselves to problems seeking an estimate of the output error signal. This paper presents techniques which allow us to use MC-FQRD-RLS algorithms with applications that previously have required explicit knowledge of the adaptive filter weights. We first consider a multichannel system identification setup and present how to obtain, at any time, the filter weights associated with the MC-FQRD-RLS algorithm. Thereafter, we turn to problems where the filter weights are periodically updated using training data, and then used for fixed filtering of a useful data sequence, e.g., burst-trained equalizers. Finally, we consider a particular control structure, indirect learning, where a copy of the coefficient vector is filtering a different input sequence than that of the adaptive filter. Simulations are carried out for Volterra system identification, decision feedback equalization, and adaptive predistortion of high-power amplifiers. The results verify our claims that the proposed techniques achieve the same performance as the inverse QRD-RLS algorithm at a much lower computational cost.

This paper proposes an efficient adaptive realization of the Wiener model for the identification of complex-valued nonlinear systems. Using a two-dimensional simplicial canonical piecewise linear filter for the complex-valued nonlinear mapping, we derive a realization of the Wiener model requiring fewer parameters than previous approaches. An adaptive implementation of the proposed Wiener model is derived, and local convergence analysis for the updating algorithm is presented. The tradeoff between computational complexity and modeling performance is discussed. Simulations of a system identification example show that the proposed algorithm can provide similar or better performance than other approaches in terms of computational complexity, convergence speed, and final mean-squared error (MSE). Fil: Cousseau, Juan Edmundo. Consejo Nacional de Investigaciones Científicas y Técnicas. Centro Científico Tecnológico Conicet - Bahía Blanca; Argentina. Universidad Nacional del Sur. Departamento de Ingeniería Eléctrica y de Computadoras; Argentina Fil: Figueroa, Jose Luis. Universidad Nacional del Sur. Departamento de Ingeniería Eléctrica y de Computadoras; Argentina. Consejo Nacional de Investigaciones Científicas y Técnicas. Centro Científico Tecnológico Conicet - Bahía Blanca; Argentina Fil: Werner, Stefan. University of Helsinki; Finlandia Fil: Laakso, Timo I.. University of Helsinki; Finlandia

This paper proposes low-complexity constrained affine-projection (CAP) algorithms. The algorithms are suitable for linearly constrained filtering problems often encountered in communications systems. The CAP algorithms derived in this paper trade convergence speed and computational complexity in the same way as the conventional affine-projection (AP) algorithm. In addition, data-selective versions of the CAP algorithm are derived based on the concept of set-membership filtering. The set-membership constrained affine-projection (SM-CAP) algorithms include several constraint sets in order to construct a space of feasible solutions for the coefficient updates. The SM-CAP algorithms include a data-dependent step size that provides fast convergence and low mean-squared error. The paper also discusses important aspects of convergence and stability of constrained normalized adaptation algorithms and shows that normalization may introduce bias in the final solution.

This paper presents and analyzes novel data selective normalized adaptive filtering algorithms with two data reuses. The algorithms [the set-membership binormalized LMS (SM-BN-DRLMS) algorithms] are derived using the concept of set-membership filtering (SMF). These algorithms can be regarded as generalizations of the previously proposed set-membership NLMS (SM-NLMS) algorithm. They include two constraint sets in order to construct a space of feasible solutions for the coefficient updates. The algorithms include data-dependent step sizes that provide fast convergence and low-excess mean-squared error (MSE). Convergence analyzes in the mean squared sense are presented, and closed-form expressions are given for both white and colored input signals. Simulation results show good performance of the algorithms in terms of convergence speed, final misadjustment, and reduced computational complexity.

This paper presents a tutorial-like detailed explanation of linearly constrained minimum-variance filtering in order to introduce an efficient implementation that utilizes Householder transformation (HT). Through a graphical description of the algorithms, further insight on linearly constrained adaptive filters was made possible, and the main differences among several algorithms were highlighted. The method proposed herein, based on the HT, allows direct application of any unconstrained adaptation algorithm as in a generalized sidelobe canceller (GSC), but unlike the GSC, the HT-based approach always renders efficient implementations. A complete and detailed comparison with the GSC model and a thorough discussion of the advantages of the HT-based approach are also given. Simulations were run in a beamforming application where a linear array of 12 sensors was used. It was verified that not only the HT approach yields efficient implementation of constrained adaptive filters, but in addition, the beampatterns achieved with this method were much closer to the optimal solution than the beampatterns obtained with GSC models with similar computational complexity.

We develop a recursive total least-squares (RTLS) algorithm for errors-in-variables system identification utilizing the inverse power method and the dichotomous coordinate-descent (DCD) iterations. The proposed algorithm, called DCD-RTLS, outperforms the previously proposed RTLS algorithms, which are based on the line-search method, with reduced computational complexity. We perform a comprehensive analysis of the DCD-RTLS algorithm and show that it is asymptotically unbiased as well as being stable in the mean. We also find a lower bound for the forgetting factor that ensures mean-square stability of the algorithm and calculate the theoretical steady-state mean-square deviation (MSD). We verify the effectiveness of the proposed algorithm and the accuracy of the predicted steady-state MSD via simulations Refereed/Peer-reviewed