Diana Cristina González

Analysis of the Multifractal Nature of Speech Signals

Frame duration is an essential parameter to ensure correct application of multifractal signal processing. This paper aims to identify the multifractal nature of speech signals through theoretical study and experimental verification. One important part of this pursuit is to select adequate ranges of frame duration that effectively display evidence of multifractal nature. An overview of multifractal theory is given, including definitions and methods for analyzing and estimating multifractal characteristics and behavior. Based on these methods, we evaluate the utterances from two different Portuguese speech databases by studying their singularity curves (τ(q) and f(α)).We conclude that the frame duration between 50 and 100 ms is more suitable and useful for multifractal speech signal processing in terms of speaker recognition performance [11].

Distributed Suboptimal Schemes for TAS/SC and TAS/LS in Fixed-Gain AF Relaying Systems

We design and analyze two distributed suboptimal schemes for transmit-antenna selection and link selection in a dual-hop, fixed-gain, amplify-and-forward relaying system, composed by one multi-antenna source, one single-antenna destination, and one single-antenna relay. The proposed schemes share the same antenna-selection policy, but differ from each other in the way the direct or relaying link is selected for communication. In a first scheme, the link is selected after transmission, at the destination; in a second scheme, it is selected before transmission, at the source. A great advantage of the proposed schemes over the optimal centralized solution is their low and constant delay/feedback overhead, regardless of the number of transmit antennas. In addition, the second scheme brings an improved spectral efficiency, once it saves one time slot when selecting the direct link. We derive analytical lower and upper bounds for the outage probability of the first scheme, in single-fold integral form, and exact closed-form expressions for the outage probability and mean spectral efficiency of the second scheme. We also perform an asymptotic analysis, showing that the first scheme achieves full diversity order, whereas the second scheme achieves full diversity order minus one, as a penalty for its improved spectral efficiency.

Distributed TAS/MRC and TAS/SC Schemes for Fixed-Gain AF Systems With Multiantenna Relay: Outage Performance

Although transmit-antenna selection (TAS) is an alternative low-cost strategy to capture the advantages of multi-antenna systems, its application may require a large amount of feedback transmissions. Owing to this fact and in order to reduce such requirement, herein we analyze the outage performance of two distributed TAS (DAS) schemes. One scheme employs maximal-ratio combining (DAS/MRC) at the destination, whereas the other employs selection combining (DAS/SC). We consider a dual-hop fixed-gain amplify-and-forward relaying network, equipped with multi-antenna source/relay nodes and a single-antenna destination. A lower bound expression is derived for the outage probability of each investigated scheme. Importantly, the derived bounds prove to be very tight approximations to the exact outage performance. Also, capitalizing on a strikingly interesting property of Stirling numbers of the second kind, we provide closed-form asymptotic expressions for the obtained bounds. Our results show that the diversity order of the proposed distributed schemes is identical to that of their optimal centralized counterparts, namely Nt + min(Nrr, Nrt), with Nt, Nrr, and Nrt denoting the number of transmit antennas at the source, the number of receive antennas at the relay, and the number of transmit antennas at the relay, respectively. In addition, as the relay approaches the destination, the outage performance of the proposed schemes approaches the optimal one.

Asymptotically Optimal Power Allocation for WSNs With Mutually Correlated Sensing Data

It is well known that distributed source coding can improve the reliability and energy efficiency of wireless sensor networks by exploiting the correlation among sensing data. In this letter, we strengthen the benefits of such an approach by proposing a simple, general, efficient power allocation scheme for an arbitrary amount of sensors. The proposed power allocation is asymptotically optimal at high signal-to-noise ratio.

Improving MFCC based ASI system performance using novel multifractal cascade features

In this work we use a set of multifractal features, namely Variable Variance Gaussian Parameter (WGP), extracted from a cascade model of speech signals to improve the performances of a traditional speaker recognition approach. We describe in detail the stochastic cascade model used to represents these WGP features as well as the proper feature extraction procedure. The evaluation of the discriminative capability of the WGP features is carried out in two steps. First we implement an automatic text-independent speaker identification system based only on the WGP features and Gaussian mixture model (GMM) classifiers. Then, we evaluate classification strategies that jointly use both the WGP and traditional mel-frequency cepstrum coefficients (MFCCs) features under two multimodal fusion schemes, namely score-level and feature-level fusion. Experimental tests reveal that the WGP features are discriminant and capable of improving the performance of MFCC based ASI systems.

Distributed transmit-antenna selection in variable-gain relaying systems

Recently, distributed transmit-antenna selection schemes have attracted great interest, since they capture the essential benefits of multi-antenna systems while reducing their cost, complexity, delay, and feedback overhead. In those distributed schemes, the antenna selection is based on local channel-state information, in contrast to their optimal centralized counterparts, which require knowing the channel state of all links. Herein, we design two such distributed schemes for a dual-hop variable-gain amplify-and-forward relaying system with one multi-antenna source, one single-antenna relay, and one single-antenna destination. The two schemes differ in the diversity method used at the destination, namely, selection combining or maximal-ratio combining, and in the selection rule accordingly. In addition to conceiving these new schemes, we analyze their outage performance. Since an exact analysis proves intractable, we tackle the outage probability in terms of lower-bound expressions and their asymptotes at high signal-to-noise ratio. Importantly, the derived bounds turn out to be almost indistinguishable from the true performance, assessed via simulation. Our results reveal that the proposed distributed schemes achieve the same diversity order of their optimal centralized counterparts and perform closely to these, specially when the relay is near the source or destination.

An Efficient Power Allocation Scheme for Multirelay Systems With Lossy Intra-Links

The so-called chief executive officer problem suggests that the source message can be recovered at the destination by merging a set of corrupted replicas forwarded by multiple relays, as long as these replicas are sufficiently correlated with the original message. In this paper, we build on Slepian-Wolf’s correlated source coding theorem to design a simple, yet efficient power allocation scheme for a multirelay system, in which the direct link is unavailable to convey information. In such a system, the replicas forwarded by the relays are allowed to contain intra-link errors due to previous unreliable hops, and the destination is supposed to retrieve the source message by jointly decoding all received replicas. Importantly, the proposed power allocation is asymptotically optimal at high signal-to-noise ratio.

VVGP features for speaker verification using i-vector framework

Spectral parameters alone, especially the Mel-Frequency Cepstral Coefficients (MFCCs) and perceptual linear prediction (PLP) coefficients, have shown good performance in speaker recognition. However, the cepstrum carries only linear information. In this paper, we study the performance of the Variable Variance Gaussian Parameter (VVGP) in a state-of-the-art i-vector speaker verification system. Experimental results on the Ynoguti 2 database indicate that VVGP features is complementary to MFCCs and can improve recognition accuracy.