Rodrigo Pinto Lemos

A new DOA estimation algorithm based on differential spectrum

This work proposes a new DOA estimation algorithm based upon the difference between the singu- lar values of the estimated spatial covariance matrix. The proposed method presented competitive RMSE performance when compared to MODEX and Root-MUSIC. Specifically for low SNR values, near angles and correlated signals, this new method had a better performed. Index Terms— DOA estimation, Array signal processing, Adaptive arrays.

A branch-and-bound inspired technique to improve the computational efficiency of DOA estimation

The former SEAD method was recently proposed by the authors and showed maximum likelihood performance in resolving closely located DOA (direction of arrival) angles at low signal-to-noise ratios (SNR). However, it required a prohibitively high computational effort. We propose a new estimate selection strategy inspired on branch-and-bound in order to reduce the computational effort of SEAD. Equations of computational cost were derived and validated for both SEAD versions. Also, we introduced an adaptive strategy that detects the signal peaks and estimates the amount of signal sources grouped in each peak. Experimental results demonstrated that the proposed strategy dramatically reduced the computational effort and yielded an improved SEAD. Compared to MODEX, improved SEAD showed competitive runtime and its root mean square error (RMSE) performance attained the Cramer-Rao lower bound (CRLB) up to SNR=-12dB, 8dB lower than that of MODEX. Also, computational effort grew slower for improved SEAD than for MODEX, as the number of sources was increased.

A new concept of assistive virtual keyboards based on a systematic review of text entry optimization techniques

Abstract Introduction: Due to the increasing popularization of computers and the internet expansion, Alternative and Augmentative Communication technologies have been employed to restore the ability to communicate of people with aphasia and tetraplegia. Virtual keyboards are one of the most primitive mechanisms for alternatively entering text and play a very important role in accomplishing this task. However, the text entry for this kind of keyboard is much slower than entering information through their physical counterparts. Many techniques and layouts have been proposed to improve the typing performance of virtual keyboards, each one concerning a different issue or solving a specific problem. However, not all of them are suitable to assist seriously people with motor impairment. Methods: In order to develop an assistive virtual keyboard with improved typing performance, we performed a systematic review on scientific databases. Results: We found 250 related papers and 52 of them were selected to compose. After that, we identified eight essentials virtual keyboard features, five methods to optimize data entry performance and five metrics to assess typing performance. Conclusion: Based on this review, we introduce a concept of an assistive, optimized, compact and adaptive virtual keyboard that gathers a set of suitable techniques such as: a new ambiguous keyboard layout, disambiguation algorithms, dynamic scan techniques, static text prediction of letters and words and, finally, the use of phonetic and similarity algorithms to reduce the users typing error rate.

A new DOA estimation algorithm based on angle search through the difference between the principal singular values

This work proposes a new DOA estimation algorithm based upon the difference between the singular values of the estimated spatial covariance matrix. The proposed method presented competitive RMSE performance when compared to MODEX and Root-MUSIC. Specifically under low SNR values, for near angles and correlated signals, the new method performed better.

Further investigation on Frobenius Spectrum for DOA estimation

The SEAD method relies on the Differential Spectrum to provide accurate DOA estimation even at very low values of SNR. However, calculating the Differential Spectrum requires performing successive SVD for each test angle in a scanning range. Recently, we proposed using the Frobenius norm instead of the SVD to reduce the computational effort of the SEAD method. However, the resulting DOA estimation performance remained inferior to that using the SVD. In order to improve de robustness of the Frobenius Spectrum, this paper analyzes the components of the Frobenius norm aiming to realize advantages and disadvantages of using it. Then, a comparison of the resulting method to other SEAD based methods and MODEX has shown that Frobenius Spectrum is more sensitive for close angles, however, it has shown to be more robust for distant angles.

A New Approach for the Differential Spectrum Using the Frobenius Norm

The SEAD method relies on the difference between the two largest singular values of an augmented spatial covariance matrix in order to generate a Differential Spectrum that provides accurate DOA estimation even for low values of SNR. However the SEAD method is highly dependent on the SVD, such that it has to be performed for each test angle on a sweeping range. We have found that the induced matrix 2-norm by vector, i.e. The largest singular value, corresponds to the dominant contributor to the Differential Spectrum. On the other hand, as the Frobenius norm requires far less computations than the SVD, in this paper we analyze the use of the Frobenius norm to yield a spectrum that allows estimating the DOA angles. The major contribution of this proposition is the ease of performing the calculation of the Frobenius norm as opposed to performing multiple SVD.

A New Approach for Brain Source Position Estimation Based on the Eigenvalues of the EEG Sensors Spatial Covariance Matrix

Direction of Arrival (DOA) estimation methods, like MUSIC, can be applied to EEG signals for brain source localization. However, they show a severe degradation at small signal-to-noise ratios on the EEG sensors and for large amounts of brain sources. Inspired on the SEAD method, this article introduces a new method that analyses the eigenvalues of a modified spatial covariance matrix of the EEG signals to produce a two-dimensional spectrum whose peaks more robustly estimate the source positions on a horizontal section of the brain. The key approach is to select the eigenvalues that are less affected by the noise and use them to produce the spectrum. To assess the accuracy and robustness of the proposed method, we compared its root-mean-square-error performance at different noise conditions to those of MUSIC and NSF. The proposed method showed the lowest estimation errors for different amounts of brain sources and grid densities.

Comparative Study between the Discrete-Frequency Kalman Filtering and the Discrete-Time Kalman Filtering with Application in Noise Reduction in Speech Signals

This article aims to carry out a comparative study between discrete-time and discrete-frequency Kalman filters. In order to assess the performance of both methods for speech reconstruction, we measured the output segmental signal-to-noise ratio and the Itakura-Saito distance provided by each algorithm over 25 different voice signals. The results show that although the two algorithms performed very similarly regarding noise reduction, the discrete-time Kalman filter produced smaller spectral distortion on the estimated signals when compared with the discrete-frequency Kalman filter.

Adaptive Beamforming for Moving Targets Using Genetic Algorithms

Context: This works investigates the use of Genetic Algorithm (GA) for beamforming on a Code Division Multiple Access (CDMA) environment under different Signal-to-Noise Ratios (SNR), assuming a reference signal is known. Method: The GA is a method inspired in evolutionary principles to optimize an objective function by choosing the best candidates of a population. The population is randomly generated to ensure high diversity and get a global optimization. On the other hand, the Least Means squares (LMS) algorithm is an adaptive algorithm with guaranteed convergence as long as a reference signal is known. Results: The GA converged faster than the LMS in all tested scenarios. Besides, GA achieved best results in pointing the beam for uncorrelated static sources. Additionally, proper tuning of GA parameters allowed fast convergence and improved tracking of moving targets. Conclusions: The simulation results confirm that the GA is able to obtain a convergent and accurate tool for beamforming and tracking of moving targets, given a reference signal. Hence, GA turns to be promising in replacing LMS on Smart Antenna Systems for increasing channel capacity.

Spatial Filtering Based on Differential Spectrum for Improving ML DOA Estimation Performance

Recently, maximum spatial eigenfiltering allowed significant improvements on the estimation performance of maximum likelihood direction-of-arrival (DOA) estimators for closely spaced sources. However, that eigenfilter may greatly attenuate widely spaced sources as SNR decreases, leading to severe performance degradation. Since the differential spectrum shows prominent spectral peaks around the true DOA angles even at very low SNR, we originally propose using it to derive two eigenvalue-based finite-impulse response spatial filters to overcome that problem. The first one employs the frequency sampling approach, whereas the second one includes moving average modeling, to fit the frequency response to the differential spectrum. Both the amount of spectral samples and the filter orders were carefully chosen to control aliasing in time domain and reduce the overall computational effort. The simulation results showed that the proposed filters preserved signal passbands of DOA sources, even where the maximum spatial eigenfilter failed. Compared with that filter, our propositions significantly reduced the threshold SNR for widely spaced sources while performing very similarly for closely spaced sources, at the expense of an small increase in runtime. To the best of our knowledge, the proposed filters are the first to allow spatial filtering in maximum likelihood DOA estimation independently of source separation.