The linearization performance of the traditional wideband digital predistortion (DPD) is seriously limited by the analog-to-digital converter (ADC) sampling rate in the feedback path. In this paper, a random demodulation based reduced sampling rate (RDRS) method is proposed to model and linearize the broadband power amplifiers (PAs). Compared with the existing spectrum extrapolation method, the proposed method can further reduce the sampling rate, but still maintain good performances. The simulation results demonstrate that the proposed method is well suitable for the ultra wideband digital predistortion.

A random demodulation based reduced sampling rate method for wideband digital predistortion

The analog-to-information conversion may be unable to provide satisfactory results when signals of interest show better sparsity in the fractional Fourier domain. In this paper, we redesign the random demodulator to sample generalized bandlimited signals. The fractional Fourier transform and fractional shift-invariant operator are applied to establish a relationship between time-frequency representations and observation vector. The mixing vector constructed by random phase sequence has been shown satisfying the incoherent condition. The reconstruction of representation is based on the orthogonal matching pursuit. The performance of the proposed sampling method is verified by the probability of successful reconstruction and mean squared error.

Generalized Random Demodulator Associated with Fractional Fourier Transform

Compressive sensing (CS) shows that, when a signal is sparse or compressible with respect to some basis, a small number of compressive measurements of the original signal can be sufficient for exact (or approximate) recovery. Distributed CS (DCS) takes advantage of both intra- and intersignal correlation structures to reduce the number of measurements required for multisignal recovery. In most cases of audio signal processing, only mixtures of the original sources are available for observation under the DCS framework, without prior information on both the source signals and the mixing process. To recover the original sources, estimating the mixing process is a key step. The underlying method for mixing matrix estimation reconstructs the mixtures by a DCS approach first and then estimates the mixing matrix from the recovered mixtures. The reconstruction step takes considerable time and also introduces errors into the estimation step. The novelty of this paper lies in verifying the independence and non-Gaussian property for the compressive measurements of audio signals, based on which it proposes a novel method that estimates the mixing matrix directly from the compressive observations without reconstructing the mixtures. Numerical simulations show that the proposed method outperforms the underlying method with better estimation speed and accuracy in both noisy and noiseless cases.

Compressive Blind Mixing Matrix Estimation of Audio Signals

In many applications, performing spectrum sensing after sampling at the Nyquist rate is inefficient because the signals of interest contain only a small number of significant frequencies relative to the band limit, although the locations of the frequencies may not be known a priori. For this type of signal, we focus on improve Random Demodulation, a novel technology inspired by Compressed Sensing theory to achieve sub-Nyquist sampling and boosting the efficiency of spectrum sensing. So, Synchronized Random Demodulation technology is proposed and a corresponding system is designed in this paper. It is noticed that two key elements play crucial role in this sensing system: synchronization among signals and accurate system impulse response. Then, a specialized synchronization mechanism and a cross-correlation method for measuring SRD system impulse response is developed to improve the system performance. Experiments demonstrate that the sampling rates for successful spectrum sensing and signal recovery can be as low as 4% of Nyquist rate.

Boost the efficiency of spectrum sensing using synchronized random demodulation

A novel undersampling architecture is employed to linearize the wideband power amplifiers (PAs) in this paper. The approach samples only I component of the feedback signal with a low sampling rate by means of the random demodulation-based reduced sampling rate (RDRS) method. It has lower complexity compared with some existing digital predistortion (DPD) systems, since a lower speed ADC in the feedback path is required. The least-square algorithm is adopted to extract the DPD coefficients. Experimental results show that the architecture achieves nearly the same linearization performance as the conventional methods and it is highly suitable for the broadband DPD.

A Novel Undersampling Architecture for Wideband Digital Predistortion

In this paper, a working hardware design with automatic synchronization function based on a random demodulation architecture is presented. It is a groping try to use compressive sensing theory to accomplish sub-Nyquist sampling for analog signal that are sparse in the frequency domain. This paper gives a description to the whole hardware system and also explains the effect of signals' synchronization and how to utilize OMP algorithm in a given hardware system. The experiment results show successful reconstruction of double-tone signals at sampling rates down to 1/4 of the Nyquist-rate.

A random demodulation hardware system with automatic synchronization function

This paper proposed a novel recovery algorithm for sparse wideband signals. Conventional recovery methods are mainly required the frequency support as a prior, or are based on some compressed sensing (CS) recovery algorithms, such as simultaneous orthogonal matching pursuit (SOMP), which identifies a single spectrum band one time, and couldn't refine the selected spectrum band that is wrong. The proposed algorithm adopts a backtracking strategy. In this strategy, the proposed algorithm can maintain the correct frequency supports and refine the wrong ones during next iteration. The expectation is that the recursive refinements of the estimate of the frequency support set will lead a higher success rate than SOMP. Simulation results demonstrate the proposed algorithm outperforms SOMP.

A blind spectrum recovery algorithm for sparse wideband signals based on backtracking

Compressed Sensing (CS) shows that, when signal is sparse or compressible with respect to some basis, only a small number of compressive measurements of original signal can be sufficient for exact (or approximate) recovery. While in some cases, only the mixtures of original sources are available for observation without knowing the priori information of both the source signals and the mixing process. To recover the original sources, estimating the mixing process is a key step. In this paper, we estimate the mixing matrix in the compressive measurement domain based on gradient ascent. The innovation lies in that, we recover the mixing matrix directly from the observed compressive measurements of mixture signals, without recovering the mixture signals at first. Numerical experiments show that the mixing matrix can be well estimated via the proposed method.

Compressive blind mixing matrix recovery algorithm based on gradient ascent

In nowadays, fault diagnosis method for analog circuit based on support vector machines, has become a hot topic in research field of fault diagnosis. However, in practical application of this method, the imbalanced problem occurred in fault sample dataset has greatly influenced its effectiveness. To remedy this problem, this paper proposed an improved Support Vector Machines method based on biased empirical feature mapping. In the new method, biased discriminant analysis was applied in empirical feature space, to make all normal samples far away from center of fault samples, so that the overall fault diagnosis ability can be improved. Through theoretical analysis and empirical study on actual electronic circuit fault diagnosis problem, we show that our method augments the diagnosis accuracy rate effectively.

Gang Wang

Papers

A random demodulation hardware system with automatic synchronization function

A blind spectrum recovery algorithm for sparse wideband signals based on backtracking

Boost the efficiency of spectrum sensing using synchronized random demodulation

Compressive blind mixing matrix recovery algorithm based on gradient ascent

Electronic circuit fault diagnosis methods based on improved Support Vector Machines