A System-C based Micro architectural Exploration Framework for Latency, Power and Performance Trade-offs of On-Chip Interconnection

We describe a System-C based framework we are developing, to explore the impact of various architectural and microarchitectural level parameters of the on-chip interconnection network elements on its power and performance. The framework enables one to choose from a variety of architectural options like topology, routing policy, etc., as well as allows experimentation with various microarchitectural options for the individual links like length, wire width, pitch, pipelining, supply voltage and frequency. The framework also supports a flexible traffic generation and communication model. We provide preliminary results of using this framework to study the power, latency and throughput of a 4x4 multi-core processing array using mesh, torus and folded torus, for two different communication patterns of dense and sparse linear algebra. The traffic consists of both Request-Response messages (mimicing cache accesses)and One-Way messages. We find that the average latency can be reduced by increasing the pipeline depth, as it enables higher link frequencies. We also find that there exists an optimum degree of pipelining which minimizes energy-delay product.

Joint evaluation of multiple speech patterns for speech recognition and training

We are addressing the novel problem of jointly evaluating multiple speech patterns for automatic speech recognition and training. We propose solutions based on both the non-parametric dynamic time warping (DTW) algorithm, and the parametric hidden Markov model (HMM). We show that a hybrid approach is quite effective for the application of noisy speech recognition. We extend the concept to HMM training wherein some patterns may be noisy or distorted. Utilizing the concept of ``virtual pattern'' developed for joint evaluation, we propose selective iterative training of HMMs. Evaluating these algorithms for burst/transient noisy speech and isolated word recognition, significant improvement in recognition accuracy is obtained using the new algorithms over those which do not utilize the joint evaluation strategy.

Joint decoding of multiple speech patterns for robust speech recognition

We are addressing a new problem of improving automatic speech recognition performance, given multiple utterances of patterns from the same class. We have formulated the problem of jointly decoding K multiple patterns given a single Hidden Markov Model. It is shown that such a solution is possible by aligning the K patterns using the proposed Multi Pattern Dynamic Time Warping algorithm followed by the Constrained Multi Pattern Viterbi Algorithm The new formulation is tested in the context of speaker independent isolated word recognition for both clean and noisy patterns. When 10 percent of speech is affected by a burst noise at -5 dB Signal to Noise Ratio (local), it is shown that joint decoding using only two noisy patterns reduces the noisy speech recognition error rate to about 51 percent, when compared to the single pattern decoding using the Viterbi Algorithm. In contrast a simple maximization of individual pattern likelihoods, provides only about 7 percent reduction in error rate.

Comparative Study of Vocal Communication in Two Asian Leaf Monkeys, Presbytis johnii and Presbytis entellus

This paper presents comparative data on the vocal communication of two Asian leaf monkeys, the Nilgiri langur (Presbytis johnii) and South Indian common langur (Presbytis entellus), based on sound recordings and behavioural observations of free-ranging groups. Spectrographical analyses revealed a repertoire of 18 basic patterns for Nilgiri langurs, and 21 basic patterns for common langurs. The repertoires of the two langur species consist of both discretely structured vocal patterns, in which alterations of the physical parameters are restricted to intra-class variation, and those in which structural variations cause intergradation between different sections of the repertoire. Qualitative assessments of group scans indicate that in both species vocal behaviour is characterized by pronounced sex-differences in the use of the different elements of the vocal repertoire. Comparison of data available from different populations of P. entellus suggests population-specific modifications on both structural and behavioural levels. Moreover, characteristic elements of the vocal systems of the two Asian species demonstrate striking similarities to those described for the African black-and-white colobus.

Vocal communication of wild bonnet macaques (Macaca radiata)

Field observations and spectrographic analyses of sound recordings of South Indian bonnet macaques revealed a vocal repertoire of at least 25 basic patterns. The repertoire consists of well separated sound classes and acoustic categories connected by structural intergradation. Besides structural variations within and between different elements of the repertoire, the vocal system ofMacaca radiata is characterized by regular combinations of particular basic patterns. These combinations occurred not only between calls of similar structure and function but also between calls usually emitted in entirely different social contexts. According to the qualitative analysis, sex-specific asymmetries of the vocal behaviour were less pronounced than age-dependent characteristics. The comparison of clear call vocalizations ofMacaca radiata andM. fuscata revealed significant species-specific differences on the structural and the behavioural level. Evaluations of the structural features of alarm calls of various macaque species imply marked differences between members of thefascicularis group andsinica group on one hand and thesilenus group andarctoides

Multi-Pattern Viterbi Algorithm for joint decoding of multiple speech patterns

Joint decoding of multiple speech patterns so as to improve speech recognition performance is important, especially in the presence of noise. In this paper, we propose a Multi-Pattern Viterbi algorithm (MPVA) to jointly decode and recognize multiple speech patterns for automatic speech recognition (ASR). The MPVA is a generalization of the Viterbi Algorithm to jointly decode multiple patterns given a Hidden Markov Model (HMM). Unlike the previously proposed two stage Constrained Multi-Pattern Viterbi Algorithm (CMPVA),the MPVA is a single stage algorithm. MPVA has the advantage that it cart be extended to connected word recognition (CWR) and continuous speech recognition (CSR) problems. MPVA is shown to provide better speech recognition performance than the earlier techniques: using only two repetitions of noisy speech patterns (-5 dB SNR, 10% burst noise), the word error rate using MPVA decreased by 28.5%, when compared to using individual decoding. (C) 2010 Elsevier B.V. All rights reserved.

Comparative Analyses of Age- and Sex-Specific Patterns of Vocal Behaviour in Four Species of Old World Monkeys

Sound recordings and behavioural data were collected from four primate species of two genera (Macaca, Presbytis). Comparative analyses of structural and behavioural aspects of vocal communication revealed a high degree of intrageneric similarity but striking intergeneric differences. In the two macaque species (Macaca silenus, Macaca radiata), males and females shared the major part of the repertoire. In contrast, in the two langurs (Presbytis johnii, Presbytis entellus), many calls were exclusive to adult males. Striking differences between both species groups occurred with respect to age-specific patterns of vocal behaviour. The diversity of vocal behaviour was assessed from the number of different calls used and the proportion of each call in relation to total vocal output for a given age/sex class. In Macaca, diversity decreases with the age of the vocalizer, whereas in Presbytis the age of the vocalizer and the diversity of vocal behaviour are positively correlated. A comparison of the data of the two genera does not suggest any causal relationship between group composition (e.g. multi-male vs. one-male group) and communication system. Within each genus, interspecific differences in vocal behaviour can be explained by differences in social behaviour (e.g. group cohesion, intergroup relation, mating behaviour) and functional disparities. Possible factors responsible for the pronounced intergeneric differences in vocal behaviour between Macaca and Presbytis are discussed.

Use of running fractal dimension for the analysis of changing patterns in electroencephalograms

Running fractal dimensions were measured on four channels of an electroencephalogram (EEG) recorded from a normal volunteer. The changes in the background activity due to eye closure were clearly differentiated by the fractal method. The compressed spectral array (CSA) and the running fractal dimensions of the EEG showed corresponding changes with respect to change in the background activity. The fractal method was also successful in detecting low amplitude spikes and the changes in the patterns in the EEG. The effects of different window lengths and shifts on the running fractal dimension have also been studied. The utility of fractal method for EEG data compression is highlighted.

Wavelet transform analysis of truncated fringe patterns in 3-D surface profilometry

Wavelet transform analysis of projected fringe pattern for phase recovery in 3-D shape measurement of objects is investigated. The present communication specifically outlines and evaluates the errors that creep in to the reconstructed profiles when fringe images do not satisfy periodicity. Three specific cases that give raise to non-periodicity of fringe image are simulated and leakage effects caused by each one of them are analyzed with continuous complex Morlet wavelet transform. Same images are analyzed with FFT method to make a comparison of the reconstructed profiles with both methods. Simulation results revealed a significant advantage of wavelet transform profilometry (WTP), that the distortions that arise due to leakage are confined to the locations of discontinuity and do not spread out over the entire projection as in the case of Fourier transform profilometry (FTP).

Monitoring urbanization and its implications in a mega city from space: Spatiotemporal patterns and its indicators

Rapid and invasive urbanization has been associated with depletion of natural resources (vegetation and water resources), which in turn deteriorates the landscape structure and conditions in the local environment. Rapid increase in population due to the migration from rural areas is one of the critical issues of the urban growth. Urbanisation in India is drastically changing the land cover and often resulting in the sprawl. The sprawl regions often lack basic amenities such as treated water supply, sanitation, etc. This necessitates regular monitoring and understanding of the rate of urban development in order to ensure the sustenance of natural resources. Urban sprawl is the extent of urbanization which leads to the development of urban forms with the destruction of ecology and natural landforms. The rate of change of land use and extent of urban sprawl can be efficiently visualized and modelled with the help of geo-informatics. The knowledge of urban area, especially the growth magnitude, shape geometry, and spatial pattern is essential to understand the growth and characteristics of urbanization process. Urban pattern, shape and growth can be quantified using spatial metrics. This communication quantifies the urbanisation and associated growth pattern in Delhi. Spatial data of four decades were analysed to understand land over and land use dynamics. Further the region was divided into 4 zones and into circles of 1 km incrementing radius to understand and quantify the local spatial changes. Results of the landscape metrics indicate that the urban center was highly aggregated and the outskirts and the buffer regions were in the verge of aggregating urban patches. Shannon's Entropy index clearly depicted the outgrowth of sprawl areas in different zones of Delhi. (C) 2014 Elsevier Ltd. All rights reserved.

A new load distribution strategy for linear network with communication delays

In this paper, we propose a new load distribution strategy called `send-and-receive' for scheduling divisible loads, in a linear network of processors with communication delay. This strategy is designed to optimally utilize the network resources and thereby minimizes the processing time of entire processing load. A closed-form expression for optimal size of load fractions and processing time are derived when the processing load originates at processor located in boundary and interior of the network. A condition on processor and link speed is also derived to ensure that the processors are continuously engaged in load distributions. This paper also presents a parallel implementation of `digital watermarking problem' on a personal computer-based Pentium Linear Network (PLN) topology. Experiments are carried out to study the performance of the proposed strategy and results are compared with other strategies found in literature.

Analysis and design of beam-shaped rectangular polyrod antennas

Theoretical and experimental investigations on the near field and radiation characteristics show a fairly good agreement which justifies the TE(11)(x) mode of excitation. Eight polyrod antennas of different configurations were built and tested as functions of taper angles, straight and curved axial lengths, and frequency of excitation. It is found that the radiation patterns. cross-polarization level, beamwidth and gain could be controlled not only by the axial length and taper angles but also by shaping the axis of the polyrods in order to realize an optimum design

An Efficient Load Distribution Strategy for a Distributed Linear Network of Processors with Communication Delays

In this paper, we present an improved load distribution strategy, for arbitrarily divisible processing loads, to minimize the processing time in a distributed linear network of communicating processors by an efficient utilization of their front-ends. Closed-form solutions are derived, with the processing load originating at the boundary and at the interior of the network, under some important conditions on the arrangement of processors and links in the network. Asymptotic analysis is carried out to explore the ultimate performance limits of such networks. Two important theorems are stated regarding the optimal load sequence and the optimal load origination point. Comparative study of this new strategy with an earlier strategy is also presented.