M. Tahernezhadi

Frequency-domain periodic active noise control and equalization

This paper analyzes a frequency-domain periodic active noise control (ANC) system. The adaptive filter employs the frequency domain complex least mean square (LMS) algorithm driven by a unit value at each frequency bin. The synchronously sampled frequency-domain adaptive structure acts as a comb filter to cancel narrowband noises with harmonically related frequencies. A frequency-domain periodic active noise equalization (ANE) system, which reshapes the residual noise by controlling the output of the adaptive comb filter at each frequency bin, is also developed and analyzed in this paper. Computer simulations and real-time experiments conducted attest to the practical usefulness of the proposed ANC and ANE systems.

Development of a new OFDM transceiver without guard interval

This paper develops a new orthogonal frequency division multiplexing (OFDM) receiver model that does not use the guard interval. This proposed technique (termed as preprocessing technique) is further extended to the IEEE 802.11a standard for wireless local area networks (WLANs) in this paper. The paper also addresses the other major OFDM receiver design issues, i.e., symbol detection, carrier frequency offset estimation, and channel estimation. The preamble approved by the IEEE 802.11a Standard is used to accomplish synchronization and estimation at the receiver. The paper compares the performance of the conventional OFDM system, which uses the guard interval with that of the newly proposed OFDM system that does not use the guard interval.

Acoustic echo cancellation using subband technique for teleconferencing applications

In acoustic echo cancellation, the conventional adaptive FIR filter can have many taps, resulting in a large computational complexity and slow convergence of the algorithm. Multi-rate techniques can be used to overcome these disadvantages by adapting several shorter filters in subbands at a reduced rate. However, the down sampling process introduces aliased version of the original signals, especially when signals are critically down sampled. In this paper, the nonoverlapping filter banks with decimated auxiliary subbands are employed to alleviate distortion due to aliasing and spectral gaps, while a low computational load is maintained.

An improved subband acoustic echo canceller for teleconferencing applications

In this paper a novel subband acoustic echo canceller (AEC) is proposed. The proposed scheme consists in 8 non-overlapping main subbands and 8 auxiliary subbands. The auxiliary subbands are introduced in order to provide frequency coverage in the non-overlapping region between the neighboring main subbands. Central to computational efficiency of the proposed scheme is the decimation of the auxiliary bands. Simulation results provided show that the proposed subband AEC has faster convergence speed for tracking acoustic echo path as well as better dB reduction in the echo level in comparison to overlapping subband schemes.

Real-time implementation of an IIR acoustic echo canceller on ADSP21020

A practical IIR (pole-zero) lattice adaptive acoustic echo canceller (AEC) for teleconferencing application is proposed. The proposed algorithm consists in two parts: a forward lattice and an inverse lattice. Collectively, they are referred to as the LATIN (lattice and inverse lattice) configuration. While the forward lattice is responsible for acoustic echo cancellation, the inverse lattice is employed in the double-talk (DT) mode only as to undo the distortion of the near-end speech brought about by forward lattice when suppressing the acoustic echo. Assuming M poles and M zeros for the proposed AEC, the complexity of the proposed algorithm is approximately twice the complexity of an M-tap FIR gradient lattice algorithm. Real-time experimentation conducted on an ADSP21020 floating-point DSP chip attests to the stability and fast convergence of proposed IIR algorithm.

A transform domain approach to enhancement of evoked potential EEG signals

In this paper, a transform domain approach is employed for enhancement of a periodic signal in broadband noise. In particular, results are presented in application of evoked potential EEG signals for tracking certain harmonics of interest. The proposed structure provides superior performance compared to the conventionally used full-band algorithm known as the Fourier linear combiner (FLC).

Latency change estimation for evoked potentials via a new generalized cross correlation method

Variations in Evoked Potential (EP) can be used to detect changes in the neurological system. Experiments have shown a good correlation between latency variations and injury-related changes in the neurological system. Reliable detection and accurate estimation of latency changes in EP therefore become very important in monitoring the status of the neurological system. In this paper, the Generalized Cross Correlation (GCC) method is used to estimate latency changes in EP. Experimental results show that the performance of the GCC method can be improve significantly by using the Linear Prediction Coefficients (LPC) to estimate the power spectrum instead of the traditionally used Fast Fourier Transform (FFT).

Subband acoustic echo cancellation using a thin lattice structure

In this paper a novel subband acoustic echo canceller (AEC) is proposed. The proposed scheme consists in K non-overlapping main subbands and K auxiliary subbands implemented using DFT filter bank approach. The auxiliary subbands are introduced in order to provide frequency coverage in the non-overlapping region between the neighboring main subbands. Central to computational efficiency of the proposed scheme is the decimation of the auxiliary bands by a factor of 2 relative to the main bands. A gradient lattice structure, known as thin lattice with predictor part of considerably smaller size than its ladder (filtering) part, is utilized to adapt the taps of subband adaptive filter in the main and auxiliary bands. Simulation results for K=9 number of bands attest to superior tracking and echo attenuation of the proposed structure.

Performance evaluation of convolutional encoded partial differential space time OFDM using modified local splines

Convolutional encoding is an effective technique to increase the resistance of the system towards frequency-selective channels, especially fading. In this paper, we investigate a technique in which convolutional encoding and interleaving are combined with the partial differential space time orthogonal frequency division multiplexing (OFDM). The estimation of channel parameters is done using modified local splines method. Space diversity is applied to OFDM for overall improvement of the system performance. The signals transmitted through the time-varying channel experiences multipath fading and results in burst error condition. This burst error degrades the performance of the system and should be overcome by appending a convolutional encoder and an interleaver at the transmitter end and similarly using a deinterleaver and a Viterbi decoder at the receiver end. To combat intercarrier interference (ICI), channel estimation should be done. Another method which does not require channel estimation to overcome ICI is called differential detection. This method is only valid for low Doppler. For high Doppler conditions, differential detection is combined with channel estimation for space time OFDM. This paper compares the performance evaluation of convolutional encoded partial differential ST-OFDM with three other different schemes namely; partial differential ST-OFDM, convolutional encoded coherent ST-OFDM and coherent ST-OFDM. The simulation results show that convolutional encoded partial differential ST-OFDM performs better than other mentioned schemes. Performances have been evaluated in terms of bit error rate (BER), residual ICI power and signal power to ICI power ratio.

A DSP-based lattice-pole-zero acoustic ECHO canceller

In this paper, a practical pole-zero lattice adaptive acoustic echo canceller (AEC) for hands-free telephone is proposed. The proposed algorithm consists in two parts: forward lattice and inverse lattice. Collectively, they are referred to as LATIN (Lattice and Inverse Lattice) configuration. While the forward lattice is responsible for acoustic echo cancellation, the inverse lattice is employed in the double-talk (DT) mode only as to undo the distortion of the near-end speech brought about by the forward lattice filter when suppressing the acoustic echo. Assuming M poles and M zeros for the proposed AEC, the complexity of the proposed algorithm is approximately twice the complexity of an M-tap FIR gradient lattice algorithm. Real-time experimentation conducted on a floating-point DSP chip and simulation results attest to the stability, fast convergence, and high acoustic echo suppression of the proposed pole-zero algorithm.