Contemporary Computing

This paper presents a dynamic model of the blended biogeography based optimization (BBO) for land cover feature extraction. In the blended BBO, the habitats represent the candidate problem solutions and the species migration represents the sharing of features (SIVs) between candidate solutions according to the fitness of the habitats which is called their HSI value [9]. However, it is assumed that these SIVs i.e. the number of solution features, remain constant for every habitat [10] and the HSI for each habitat depends only on the immigration and the emigration rates of species [9]. This paper extends the blended BBO by considering the fact that the no. of SIVs or the decision variables may not remain constant for all candidate solutions (habitats) that are part of the Universal habitat. Since the characteristics of each habitat vary greatly hence, comparing all the habitats using the same set of SIVs may be misleading and also may not lead to an optimum solution. Hence, in our dynamic model, we consider the fact that HSI of a solution is affected by factors other than migration of SIVs i.e. solution features, also. These other factors can be modeled as several definitions of HSI of a habitat, each definition based on a different set of SIVs which simulates the effect of these other factors. We demonstrate the performance of the proposed model by running it on the real world problem of land cover feature extraction in a multi-spectral satellite image.

Maximum weight clique problem is an NP-Hard problem which seeks the fully connected subgraph of maximum weight in a given vertex weighted graph G. In this paper, we have used a recently proposed metaheuristic technique called Grenade Explosion Method (GEM) to solve the maximum weight clique problem. GEM was originally designed for continuous optimization problems. We have suitably modified the GEM so that it can be applied to a discrete optimization problem. To our knowledge this is the first approach which uses GEM for the discrete optimization. Computational results on the benchmark instances show the effectiveness of our proposed GEM approach.

The Minimum Capacitated Dominating Set (CAPMDS) problem is the problem of finding a dominating set of minimum cardinality with the additional constraint that the nodes dominated do not exceed the capacity of the dominating node. The capacity can be uniform across all nodes or variable. Being a generalization of the Minimum Dominating Set problem, this problem also

$\cal NP$

-hard. In this paper, we present a heuristic and a couple of its variants for solving the CAPMDS problem. Our heuristics work for both uniform and variable capacity graphs. We show that the heuristic proposed is better than the variants, in general. However, for Unit Disk Graphs with high degree of connectivity and uniform capacity, one of the variants performs better. For general graphs, the proposed heuristic is far superior to the variants.

Subgraph isomorphism problem is an NP-hard problem and the available algorithms are of exponential time complexity. Hence these are not efficient for real world applications. A number of heuristic methods are proposed in the literature in this field. Ullmann[6] proposed a solution for subgraph isomorphism problem in 1976, which is being referred till today. Ullmann’s algorithm is refined to get better algorithms in current literature. Cordella et al.[7] proposed an algorithm VF2, that improves Ullmann’s refinement. In this project, we propose a heuristic to be applied to Ullmann’s algorithm in order to reduce the search space. We show that the proposed heuristic performs better than both Ullmann’s and VF2 algorithm. The testing is done using a graph generation software[12]. Further the heuristic algorithm is tested on the benchmark data set [4]. Both the experiments show that our proposed heuristics perform better for all type of graphs given in the benchmark data set.

This paper presents optimal solution of the unit maintenance scheduling problem in which the cost reduction is as important as reliability. The objective function of the algorithms used to address this problem, considers the effect of economy as well as reliability. Various constraints such as spinning reserve, duration of maintenance crew are being taken into account while dealing with such type of problems. In our work we apply the Teaching learning based optimization algorithm on a power system with six generating units. Numerical results reveal that the proposed algorithm can find better and faster solutions when compared to other heuristic or deterministic methods.

successful indexing/categorization of images greatly enhance the performance of content based retrieval systems by filtering out irrelevant classes. This rather difficult problem has not been adequately addressed in current image database systems. In this paper we have introduced a novel feature for classification of image data by taking the one dimensional representation of it (time series) as our input data. Here we have chosen local shape feature instead of global shape feature for the said purpose which enhances its consistency in case of distorted and mutilated shapes.

The POWER (PrObabilistic Weighted Ensemble of Roles) model[1] is a Bayesian mixture model where a single data point is generated by multiple components with varying degrees of influence. This is unlike standard mixture models where each data point is generated by a single mixture component. The POWER model allows for capturing various hidden and complexly overlapping components and roles of entities that contribute to generating the data points. However, the POWER model suffers from a very slow learning time. The highest complexity time of the parameters’ learning steps is

$O\left(n \cdot k \cdot d^{2}\right)$

, with

n

being the number of data points,

k

the number of components of the model and

d

the number of data attributes.

In this paper, we propose an approximation to the POWER model weight parameter learning algorithm that reduces the computational time significantly. The overall complexity of the new algorithm is

$O\left(n \cdot k \cdot d \cdot p\right)$

in practice (where

p

is an approximation parameter much smaller than

d

). This allows the model learning time to be linear in the number of attributes of the data and provides a significant speedup over the original algorithm.

We demonstrate the accuracy of our approximation technique using synthetic and real datasets. We also provide experimental results of the approximate POWER model on a dataset of NIPS papers and a dataset of wireless web access patterns and show that the model learnt are similar. An implementation of the approximate POWER model on the dataset of NIPS papers is about 27 times faster than the original version.

Grid environment is a type of parallel and distributed system that provides resources for computing and storage. Grid Scheduling is the technique of mapping jobs to resources such that resources are utilized efficiently to meet user demands. But minimization of time experienced by user for task completion is a thought provoking issue today for optimal scheduling in the grid network. This paper presents a dynamic optimization algorithm for minimizing the average turnaround time resulting in fast completion of jobs submitted to the grid. The algorithm considers arrival time of job and the waiting time encountered at resource while selecting the target resource for execution.

Most of the networks that we see in our day to day lives today are types of dynamic networks i.e. networks which keep on changing in very short period of time. One example of such a network is the network formed by Bluetooth connectivity. Due to limited range of wireless trans-receivers, a mobile node can communicate with the mobile nodes that lie within its range. Hence, in order to communicate with nodes in far-off regions, forwarding of data have to be done. Our methodology is based on this concept. Our proposition will first apply an algorithm to cluster the existing nodes in two types of clusters called INTER and INTRA clusters and then perform routing with the help of routing algorithm as proposed by us. The information regarding change in topology is maintained in the form of lists [data structure] which helps in routing the data effectively.

The quality of text-to-speech (TTS) synthesis systems can be improved by controlling the intensities of speech segments in addition to durations and intonation. This paper proposes linguistic and production constraints for modeling the intensity patterns of sequence of syllables. Linguistic constraints are represented by positional, contextual and phonological features, and production constraints are represented by articulatory features associated to syllables. In this work, feedforward neural network (FFNN) is proposed to model the intensities of syllables. The proposed FFNN model is evaluated by means of objective measures such as average prediction error (

μ

), standard deviation (

σ

), correlation coefficient (

γ

X

,

Y

) and the percentage of syllables predicted within different deviations. The prediction performance of the proposed model is compared with other statistical models such as Linear Regression (LR) and Classification and Regression Tree (CART) models. The models are also evaluated by means of subjective listening tests on the synthesized speech generated by incorporating the predicted syllable intensities in Bengali TTS system. From the evaluation studies, it is observed that prediction accuracy is better for FFNN models, compared to other models.

In this paper, an approach is proposed to accurately predict the locations of phrase breaks in a sentence for a Bengali text-to-speech (TTS) synthesis system. Determining the positions of phrase breaks is one of the most important tasks for generating natural and intelligible speech. In order to approximate the break locations, a feed-forward neural network (FFNN) based approach is proposed in the current study. For acquiring prosodic phrase break knowledge, morphological information along with widely-used positional and structural features are analyzed. The importance of all the features is demonstrated using a model-dependent feature selection approach. Finally the phrase break predicting model is implemented with the selected optimal set of features and incorporated inside a Bengali TTS system built using Festival framework [1]. The proposed FFNN model is developed using the optimally selected morphological, positional and structural features. The performance of the proposed FFNN model is compared with widely used Classification and Regression Tree (CART) model for prediction of breaks and no-breaks. The FFNN model is evaluated objectively on the basis of precision, recall and a harmonized measure - F score. The significance of the phrase break module is further analyzed by conducting subjective listening tests.

The present paper studies the propagation of torsional surface wave in a homogeneous isotropic substratum lying over a viscoelastic half space under the influence of rigid boundary. Dispersion relation has been obtained analytically in a closed form. The effect of internal friction, rigidity, wave number and time period on the phase velocity has been studied numerically. Dispersion equation thus obtained match perfectly with the classical dispersion equation of Love wave when derived as a particular case.

In this work, we propose a SaaS model that provides service to ordinary investors, unfamiliar with finance models, to evaluate the price of an option that is currently being traded before taking a decision to enter into a contract. In this model, investors may approach a financial Cloud Service Provider (CSP) to compute the option price with time and/or accuracy constraints. The option pricing algorithms are not only computationally intensive but also communication intensive. Therefore, one of the key components of the methodology presented in this paper is the topology-aware communication between tasks and scheduling of tasks in virtual machines with the goal of reducing the latency of communication between tasks. We perform various experiments to evaluate how our model can map the tasks efficiently to reduce communication latency, hide network latency ensuring that all virtual machines are busy increasing response time of users.

Creation of metamorphic relations, and building up a dynamic system to carry out automated metamorphic testing has been aimed. Modeling notation has been adopted to design and extract metamorphic properties of the system, which have been tested during run time. Real time data for system execution has been used to generate follow up test data. Thus, parallel execution and testing system properties helps to achieve reduction in testing cost and effort. The complete methodology and its effectiveness for testing systems for which no reliable test oracle exists, has been demonstrated with the help of suitable case study, and the results have been verified.

Analyst reports are confirming that adoption of SOA is growing; the actual goal of SOA is to help align IT capabilities with business goals. In today’s competitive scenario where business demand changes very frequently, the expectation from technology is raised to level where we are expecting the business processes are developed in such a manner that they can adapt the frequent changes without affecting the overall organization business architecture. Thus the need to assume business processes as a smart services that can be loosely coupled. Thus need of service oriented architecture arises. This paper is a review of articles and research work that have undergone in the past 1 decade (i.e. from 2001 - 2011). The source of data is from most prestigious journal and website covering area of SOA, In this paper we have identified the factors that are relevant to SOA implementation and up to how much extents each factor is crucial to SOA implementation is also identified.

Much research has been reported in the field of gender classification through face recognition techniques making use of variation in skin textures and distances between the nose, eyes and mouth and so forth. However, to the best of our knowledge, no work has yet been done on recognition of males and females on the basis of lip colour. In this paper, we have proposed a two-stage methodology on the above subject. In the first stage, termed as the Lip Contour Extraction Stage, the colour contrast between the skin and the lip enables the lip contour extraction. The mask so obtained is made noiseless thus perfecting the contour. The next stage, termed as the Gender Classification stage, the extracted contour is classified for genders on the basis of pronounced variation in the hue and saturation values of the lip colour. These variations, though not discernible through the naked eyes, are analysed using training machines like SVM and Neural Network. The promising experimental results showing an overall accuracy of 85 % while an accuracy of 90% for Asian and 80% for White, for gender classification, can pave way to the future of lip recognition in Biometric systems.

Predicting strength of a relationship (also known as Tie Strength Problem) has been a trivial research area amongst sociologists for decades. However, considering the recent trends in internet behavior of people along with the development of so called

social web

, makes it popular amongst web scientists to work on this as a potential research topic with new perspectives. Real life is a complex social dynamic system comprising individuals starting of either as

strong acquaintances

or

weak acquaintances

and move towards

strong

or

weak ties

with passage of time. In this paper we validate the existence of varying degree of relationship individuals have on Facebook using unsupervised machine learning techniques like divisive hierarchical clustering and statistical techniques like SSE ; analyzing strength of the boundaries that distinguish them. We have realized this on a feature rich dataset of more than 100 nodes collected during 10th of July, 2011 to the 9th of September 2011 using a Facebook application. We provide descriptive error analysis interviews focussing on the clustered structure, obtaining it with an accuracy of 90%. The paper concludes by illustrating how modeling tie strength can improve social media design elements, including privacy controls, message routing and information prioritization in databases. Potential usage of this work can be in making complex recommender systems, lead generation marketing and in organizational or telecom network.

Focused crawling is considered to be an important strategy to reduce search space and give more relevant links to a user, based on search queries. Existing web crawlers work only on the basis of full string matching of query keywords with words present in various tags or fields in the web pages. But a particular keyword can have different meanings in different contexts depending on its usage as verb, noun etc. For example

‘fly’ refers to an insect if used as a noun and refers to an act of moving in the air if used as a verb.

Most of the existing search engines work on semantic context, based on string matching of keywords but not based on contextual senses of keywords. Further, general crawling strategy of various crawlers is forward oriented, giving less consideration to the backward links of the page. There is a strong need to work on a crawling strategy that overcomes these gaps. In this paper a mechanism that evaluates the web document on the basis of contextual senses (verb, noun etc.) of the keywords contained in the downloaded page is being proposed. Moreover back-link to a web page has also been analyzed with reference to a specific page providing links related to the page. Consequently, more number of relevant links related to one topic is displayed to the user.

Clustering has been used extensively for exploratory data analysis. GK clustering algorithm can provide a data partition that is more meaningful than the standard fuzzy c-means and its variants. In this paper we propose a novel approach towards fuzzy clustering which reduces the processing time significantly while keeping the results highly accurate. It is a matrix based approach using the concept of equivalent samples and the weighting samples. Equivalence is measured in terms of proximity of the samples and then weighted samples are used as an input to the modified GK clustering algorithm. Objective function and validation index estimates are used to assess the goodness of partition. Experimental results are shown to emphasize the benefits of the proposed technique in domains like Telecom where we have massive data sets to be processed for real time clustering and recommendation engines.

A data warehouse stores historical information, integrated from several large heterogeneous data sources spread across the globe, for the purpose of supporting decision making. The queries for decision making are usually analytical and complex in nature and their response time is high when processed against a large data warehouse. This query response time can be reduced by materializing views over a data warehouse. Since all views cannot be materialized, due to space constraints, and optimal selection of subsets of views is an NP-complete problem, there is a need for selecting appropriate subsets of views for materialization. An approach for selecting such subsets of views using Genetic Algorithm is proposed in this paper. This approach computes the top-T views from a multidimensional lattice by exploring and exploiting the search space containing all possible views. Further, this approach, in comparison to the greedy algorithm, is able to comparatively lower the total cost of evaluating all the views.

Tremendous growth in traffic is witnessed over the Internet where backbone links of several gigabits per second are commonly deployed. In order to handle these gigabit-per-second traffic rates, backbone routers must forward millions of packets per second on each of their ports. Routing tables of the core routers consists of IP addresses of the order of 200,000-500,000 and changes dynamically. A major challenge is to determine the next-hop address with as low as possible number of accesses of the routing table. IP address lookup in the routers uses the packets destination address to determine the next hop for each packet and is therefore crucial to achieve the required packet forwarding rates. IP address lookup is difficult because it requires a longest common prefix (LCP) match search. In the last couple of years, various algorithms for high-performance IP address lookup have been proposed. The objective of this paper is to use a specific data structure and develop the lookup algorithm that is required to meet the demands like fast lookup, memory efficiency and fast incremental updates. We have used a novel data structure y-fast trie for the routing table in this work. We adapted the algorithm for predecessor/successor search in x-fast trie via dynamic perfect hashing technique to find the longest common prefix between the incoming packets destination address and the next-hop address. By looking at this longest common prefix, we identify the next-hop address. As an improvement over this method, we also have used indirection using balanced BSTs (y-fast trie). On average the routing table creation takes 51703

μsec

for 100000 IP addresses in the method using indirection. Average lookup time using dynamic perfect table takes 0.83

μsec

.

In this work, spectral features are extracted from speech to perform emotion classification. Linear prediction cepstral coefficients, Mel frequency cepstral coefficients and their derivatives (velocity and acceleration coefficients) are explored as features. Gaussian mixture models are proposed as classifiers. The emotions considered in this study are anger, fear, happiness, neutral, sadness and surprise. The emotional speech database used in this work is both simulated and semi-natural in nature. The semi-natural database has been collected from the dialogues of actors/actresses in popular Hindi movies. Average emotion recognition performance, in the case of male and female speaker is observed to be around 65.3% and 72% respectively. Recognition performance for semi-natural and simulated databases has been compared.

Storing and querying are two important issues that need to be addressed while designing an information retrieval system for a large and high-dimensional data set. In this work, we discuss about tackling such data, specifically about the nearest neighbour search and the efficient storage layout to store such data. The data set used in the current work has been taken from an online source called ZINC, a repository for drug like chemical structures. Processing a high dimensional data is a tough task hence dimensionality reduction should be employed. Here for dimensionality reduction is achieved through a filter-based feature selection method, based on correlation fractal dimension (CFD) discrimination measure, is used. The number of dimensions using the correlation fractal dimension are reduced from 58 to 7. To identify the nearest neighbours for a given chemical structure Tanimoto similarity coefficient is used with these reduced set of features. The nearest neighbours identified using the Tanimoto measure are stored in a storage layout known as modified inverted file. Nearest neighbours for a query can be retrieved back from the storage layout, with just one read operation from the data file thereby reducing the time for retrieval.

This paper proposes Linear Prediction (LP) residual of speech signal for characterizing the basic emotions. LP residual is extracted from speech signal by LP analysis, by inverse filtering of the speech signal. LP residual basically contains higher order relations among the samples. Instant of glottal closure in a speech signal is known as an epoch. The significant excitation of vocal tract usually takes place at the instant of glottal closure. For analysing speech emotions, the LP residual samples chosen around glottal closure instants are used. A semi-natural database GEU-SNESC (Graphic Era University Semi Natural Emotion Speech Corpus) is used for modeling the emotions. This database is collected by recording dialogs of film actors from Hindi movies. In the study four emotions namely anger, happy, neutral and sadness are used. Auto-associative neural network models are used for characterizing the basic emotions present in the speech. Average emotion recognition of 66% and 59% is observed respectively for the epoch based and entire LP residual samples.

Currently unified modelling language (UML) is widely used for the specification and modeling of software. Model driven approach uses unified modeling language as platform independent model and converts it into platform specific model by adopting different strategies in the form of stereotype and metadata. However non-uniformity in strategy makes UML based model driven architecture (MDA) a challanging job. Also very less number of platform specific code is generated when UML platform independent model (PIM) is converted to platform specific model (PSM). A tool is proposed for design and implementation which is using eXtensible markup language (XML) as PIM. XML provides uniformity in description of different components. It also provides interoperability which is otherwise not achieved. Code density and code gain significantly increase when XML PIM is converted to PSM. A case study demonstrates the applicability of this tool.

LTE is a standard for wireless communication of high speed data rates, higher system throughput and lower latency for delay critical services. Basically belongs to 3GPP family and is based upon GSM and UMTS network technologies. Proper Handoff (HO) algorithm can make the system increased capacity, better coverage requirements. A new Handoff (HO) algorithm in LTE networks based on MCHO is being proposed in this paper for the improvement of performance in a fading environment based on cost-231 Walfisch Ikegami Model.

Tree topology based sensor node deployment in a region is a common approach. The network has a root called sink node and leaves known as end-devices. The end-devices sense the environmental phenomenon and forward it to the sink by single-hopping or multi-hopping. For it, device can either follow a fixed parent-child path depicted by Tree Routing protocols, or can utilize neighbor table to identify shortest path to the destination. The Enhanced Tree Routing (ETR) protocol is such a protocol that uses a structured node address assignment scheme. It uses neighbor table to find alternative one-hop neighbors link with minimum computation, other than parent-child links, for packet forwarding. The protocol is well suited for small and static tree topology and performs well. However, it lacks in focusing some issues like, how data is forwarded to sink i.e. raw-data converge cast or aggregated-data converge cast at each node, how to resolve multiple shortest path problem if network density increases and how to deal with changeable network topology. We, in this paper thus resolve some of the issues related to ETR protocol by proposing some new ideas and improvements.

With exponential growth in the number of computer applications and the size of networks, the potential damage that can be caused by attacks launched over the internet keeps increasing dramatically. A number of network intrusion detection methods have been developed with their respective strengths and weaknesses. The majority of research in the area of network intrusion detection is still based on the simulated datasets because of non-availability of real datasets. A simulated dataset cannot represent the real network intrusion scenario. It is important to generate real and timely datasets to ensure accurate and consistent evaluation of methods. We propose a new

real dataset

to ameliorate this crucial shortcoming. We have set up a testbed to launch network traffic of both attack as well as normal nature using attack tools. We capture the network traffic in packet and flow format. The captured traffic is filtered and preprocessed to generate a featured dataset. The dataset is made available for research purpose.

With rapid growth of Ad Hoc Networks consisting of low power computing devices, security will be an important factor for their full implementation. Because of scarcity of resources in terms of computing capability and energy efficiency, designing of computationally efficient group key management protocols with dynamic topology is a major concern. Teo and Tan [11] proposed an energy-efficient generalized circular hierarchical group model, but this approach suffers from: (i) exponential increase of key messages due to dynamic topology and (ii) energy loss because the vicinity of nodes in a subgroup is high. This work is an extension of Teo & Tan’s circular hierarchical model for fixed number of group members. The proposed modification overcomes these two weaknesses of Teo & Tan’s protocol. The proposed modifications make this protocol secure against replay, masquerading, spoofing, chosen ciphertext and impersonation attacks because of proper authentication and digital signatures. The comparative numerical and simulation analysis of proposed approach has been made with Teo & Tan, Wen-Lin-Hwang’s (WLH) and along with Tseng’s group key agreement approach. The analysis shows that proposed approach is well suited for low computational mobile devices with minimum delay. Through WLH protocol shows maximum throughput and minimum delay however it lacks in terms of security aspects.

Recently Cloud robotics is a very vibrant research area due to its strategic application potentials. In this paper we have developed a basic cloud based architecture for knowing the localization information of a robot in a dynamic environment. Subsequently, this information could be useful to guide the robot in the desired path as trained by the central cloud. In this paper, Artificial Neural Network (ANN) is used for the training of locations with Radial Basis Function (RBF). The idea is to establish the communication between the cloud and robot over a large environment using the JAVA-RMI interface and identify the location from the images sent by the robot .This paper describes the Cloud As Software As a Service(SAAS).

Resource replication in distributed environment produces issues of secondary storage. Dereplication of resources is required when replication mechanism is hindered due to lack of secondary storage. This paper introduces dereplication approaches that depend upon last modification time, number of replica available and resource size. Comparative study shows that dereplication can be used to overcome the space overhead issue and reduces the dereplication time. Result shows that in case the space required is same but number of files to be dereplicated varies, dereplication time also varies depending on number of files to be dereplicated. Dereplication time will be more for case having large number of files. With the proposed approach, if file size increases by the multiple of 7, de-replication time will get increase just by the multiple of 1.5. This shows that dereplication time is decoupled from size of files that are dereplicated on the fly dynamically and does not increase proportionally with respect to file size.

Cloud computing is defined as delivering of computing resources as a service. Data security and access control are key components for any cloud service. The service level agreements are negotiated when service provider registers with an enterprise. This paper proposes an authentication and authorization interface to access a cloud service. Service selection is acquired via monitoring of security measures provided by a service provider through Security Service Level Agreements (Sec-SLAs). The enterprise and employee validation is performed through two level authentication mechanisms. Single sign on mechanisms for user and services makes the proposal more efficient. Features like denial of service, man in the middle attack and access control rights of employees are also handled. Security measures provided by service provider are handled by an enterprise, thereby, relieving the end user up to 20%-80% from the nitty-gritty of service providers in comparison to the approaches proposed in past, depending on application requirement.

The performance of any face recognition system degrades severely under varying illumination conditions. In this paper, a new approach of face recognition under varying lighting conditions has been proposed, in which a non-iterative neural network based classification has been used. Adaptive histogram equalization along with logarithm transform has been used to enhance to face image contrast. Further, down scaling of low-frequency discrete cosine transform coefficients (LFDCT) has been applied to suppress the effect of variable illuminations. These illumination normalized face images have been recognized using a neural network classifier whose network parameters have been calculated analytically. The performance of the developed approach has been evaluated on Yale and CMU PIE face databases, which reveals significant performance improvement by our approach.

Now a days the Internet has become common man’s communication channel and due to that ensuring security at all levels has become tedious.Denial of Service (DoS) attacks have grown to give rise to Distributed Denial of Service (DDoS) attacks. Due to the open access of Internet the software tools for generating bots are easily available. This has increased the span of DDoS. The traditional methods of DDoS detection fail to detect this emerging breed of attacks. In the recent past Shannon entropy analysis has been done for detection of intrusions in the computer network. Shannon entropy however has limitations in failing to detect attacks of very short duration. Generalized form of Non extensive Tsallis entropy has been tested to look into weaknesses of Shannon entropy. Secondly, there has been growth in the area of application of wavelets to signal processing. Because of their inherent nature wavelets beautifully capture the nature of traffic at multiple scales. We have tried to use Daubechies wavelets to measure Tsallis entropy with different moments and have detected the sudden changes induced in the traffic pattern because of DDoS attacks.

Eager replication is a technique of replication that ensures high consistency. Unfortunately, it degrades the system performance by increasing the response time and sacrificing availability. Lazy replication is a technique that provides high availability and ensures that database will eventually be in a consistent state. A formal rigorous reasoning is required to precisely understand the behavior of such techniques and to understand how they achieve the objectives. Event-B is a formal technique that is used for specifying and reasoning about complex systems. In this paper, we present a formal development of lazy replication system using Event-B. We outline our model for distributed environment where same database is replicated at all the sites. After updating the database locally within transactional framework messages are broadcast to other sites so that they can change their replicas.

IEEE 802.11e EDCA can be used to provide

Quality of Service

(QoS) in an ad hoc network at the

Medium Access Control

(MAC) layer. In this paper, we describe a model for the analysis of IEEE 802.11e EDCA for an ad hoc network. Our model is based on the differentiation among different classes of traffic based on the size of

Contention Window

(CW), and

Arbitration Inter-Frame Space

(AIFS). During our analysis, we focus on the following parameters: (i) average end-to-end delay, and (ii) throughput. To validate our analysis, we carried out simulations and found that the results obtained analytically are in accordance with those obtained through simulations.

Financial option pricing is a compute-intensive problem that requires real-time pricing for making decisions on investment portfolios or studying the risk value of a company’s assets. In this study, we report our experiences designing an algorithm for a complex option pricing problem on the Accelerated Processing Unit (APU), a state-of-the-art multi-core architecture. Using a naive algorithm, both the APU and GPU do not perform well as there is a non-optimal use of memory which limits our utilization of computational resources. To improve performance we examined two methods of optimization: (i) vectorization of the computational domain and (ii) loop unrolling of the computation. Through these two methods we achieve better performance and scalability with less powerful hardware than other GPU solutions currently available.

Maintenance of deployed software is an important phase of software lifecycle. Estimation of maintenance effort in object oriented software engineering is one of the major challenges. In object oriented software the maintenance effort is highly correlated with coupling among classes. It is widely accepted that there is strong relationship between high coupling and poor maintainability. The existing metrics sometimes do not depict the effect of the key factors which contribute significantly towards maintenance effort. Indirect coupling which manifests between two seemingly unrelated classes through hidden connections plays a major role in determining maintenance effort. This research proposes metrics which estimates the maintenance effort for software having indirect coupling between classes.

Adaptive Modular Active Leg (AMAL), a robotic Intelligent Prosthetic Limb has been developed at the Indian Institute of Information Technology Allahabad. The aim of the project was to provide the comfort of an intelligent prosthetic knee joint for differently abeled person with one leg amputated above the knee. AMAL provides him with the necessary shock absorption and a suitable bending of the knee joint oscillation. The bending and the shock absorption are provided by artificial muscles. In our case, it is the MR (Magneto Rheological) damper which controls the knee movement of an amputee. The feedback signal is provided by the heel’s strike sensor. AMAL has been kept simple with minimal feedback sensors and controls so that the product is economically viable for the patients. In this paper we describe the mechanical design, the electronic control with its successful testing on differently abeled persons.

We envision that in near future, Humanoid Robots will enter in the household. They have capability to change the quality of life and advance the human civilizations further. For this thing to happen, humanoid robot researchers need to solve many challenging problems. Towards this we have taken a significant step in developing indigenously a low cost research test bed where researchers can develop and test many technologies for Human Robot interactions. The basic issue which has been addressed is the development of a cost effective platform which will be stable and robust, having an open architecture so that various technologies like speech, vision, intelligence can be integrated with physical gestures. First we have described the developmental architecture of the SMART, and then the application of SMART as an intelligent robot which, for the time being, is capable of demonstrating our robotics and artificial intelligence laboratory to the visitors as a lab guide in an interactive manner.

In this paper we have discussed the method to detect skin identification. This method includes three main steps: First segmentation based on the pixel values, in which when the particular pixel values lies in the range it is skin area. Second is the post processing stage in which some area which is not the skin color but counted as a skin area is eliminated and some which is not counted as skin is added as possible. Third the segmented area is masked with original image (input image) and Finally Euclidean Distance is applied to find out Color difference between the segmented skin and mean of reference image based on threshold values classify the skin is infected or not.

Various methods and tools are used for genomic sequences annotation, each of which needs training data set and hence their accuracy is confined to specific type of organism. To surmount this problem, we proposed a hybrid method in which weighted annotated binary DNA sequences from different tools are convolved independently with multi scaled modified Gaussian function that generates set of multi scaled sequences for each tool. All the sequences of the same scale values from different tools are added based on each nucleotide position. Then this multi scaled sequences are normalized, scaled and combined together for each nucleotide position. By combining best predicted ranges among different predicted ranges from individual gene prediction tool, our proposed tool increases Exon level accuracy by 10 – 12 % whereas 2-4 % of missed and wrong exons can be identified in comparison to accuracy given by single gene predicting tool.

Compression is becoming very necessary in today’s time especially in the medical field. The reason for this is that the current analog film based medical images are very difficult to manage and can be easily damaged if exposed to sunlight. Digital images are more reliable and easier to manage but occupy a lot of computer space. Compressed medical images occupy less space and can be easily transmitted over the network in lesser amount of time. In this paper two algorithms have been discussed: The JPEG DCT image compression and the JPEG wavelet compression. The algorithms have been compared on the values obtained for the Mean Square Error and Peak Signal to Noise Ratio. The results show that wavelet compression gives better quality for the same compression ratio in comparison to DCT compression.

This paper presents a novel technique of using depth maps to recognize activities involving two persons. Very simple depth map based features are used to reduce processing power requirement. Kinect

®

SDK is used to identify the skeleton of up to two persons in the scene. Shape and temporal features are extracted from the region of interest. These features are subsequently passed to SVM Classifier for training and classification.

A unique approach for estimating the depth from a monocular moving camera has been synthesized. Good interest points are obtained using the Shi-Tomasi technique for every frame in real time. Lucas-Kanade method is applied on these interest points of two consecutive frames which computes feature of the frames and the maps the interest points. It utilizes mainly the Lucas approach for estimation of depth from continuous input of frames from the moving camera which helps in estimating relative depth amongst various objects in the scene. Two approaches have been designed to solve this problem.

Spoken Language Identification (SLI) is the process of identifying the language spoken by a speaker. Language identification has several applications in day-today life. It may be used in call centers (e.g., emergency and customer services), information directories (e.g., airport, hotel, and tourist attractions) dealing with speakers speaking different languages[1]. Humans perform language identification mainly based on the specific words(phonetic information) and pattern of pronunciation. Spectral features are known well to capture phonetic information from the speech utterances[2]. Therefore, in this work MFCC’s(Mel Frequency Cepstral Coefficients) are used. Language identification is mainly done using some pattern classifier namely GMM, SVM, ANN and HMM[3].

Cloud Computing is emerging out as the future of next generation architecture for information technology enterprises. But due to its popularity, it is vulnerable to various unwanted attacks. One of the solutions is intrusion detection system. The Existing architectures of IDS in cloud environment are deployed on the network periphery of each guest OS that offers high attack resistance at the cost of visibility. In this paper, we propose hybrid architecture for deployment of intrusion detection system which takes into account security at both the front end and the clusters. This Paper also includes a critical review of previously proposed architectures on deployment of Intrusion Detection Systems in Cloud Environment and a detailed description of the research Gaps identified. Our approach leverages VMware virtualization techniques using open nebula as a test bed for deploying our proposed system.

Vowel recognition is the focus of automatic speech recognition (ASR) and speaker identification systems because of spectrally well defined characteristics of vowels. The ability to recognize speech improves significantly by efficient vowel recognition, both by humans as well as by ASR systems [1]. Therefore vowel recognition has an important role in speech processing systems. Now-a-days telephonic vowel recognition has gain lot of research interest as telephones have been revolutionized the personal and professional communication. Therefore, in this work emphasis is given to vowel recognition from telephonic speech. Telephonic vowel recognition of English language is a challenging task as the same vowel has got different contextual pronounciation. Telephonic speech recognition is greatly benefitted by efficient vowel recognition. Recognition of vowels from telephonic speech has applications in speaker recognition for critical banking transactions, identification of criminals, access control systems and forensic applications. In the field like Biometrics, goal of vowel recognition is to verify an individual’s identity based on his or her voice. This is one of the trusted means of authentication as voice is one of the most natural forms of the communication.

Orthogonal frequency and code division multiplexing (OFCDM) technique has shown promising results in achieving high data rate while simultaneously combating multipath fading. The novelty of an OFCDM system is two-dimensional spreading and code multiplexing. These are achieved by using 1D-OVSF codes. Here, 2D-OVSF codes are applied in the OFCDM system to reduce multi code interference amongst code multiplexed channels.

The paper “Dynamic Optimization Algorithm for Fast Job Completion in Grid Scheduling Environment” authored by Monika Choudhary and Sateesh Kumar Peddoju, DOI 10.1007/978-3-642-32129-0_14, appearing on pages 86-94 of this publication has been retracted due to plagiarism. It is a duplicate of the paper titled “Turnaround time based job scheduling algorithm in dynamic grid computing environment” by the same authors, published in Proceedings of the CUBE International Information Technology Conference (CUBE 2012): DOI 10.1145/2381716.2381809.