Distributed Computing and Internet Technology

In recent years, several works have demonstrated how the study of biology can benefit from an algorithmic perspective. Since biological systems are often distributed in nature, this approach may be particularly useful in the context of distributed computing. As the study of algorithms is traditionally motivated by an engineering and technological point of view, the adaptation of ideas from theoretical distributed computing to biological systems is highly non-trivial and requires a delicate and careful treatment. In this review, we discuss some of the recent research within this framework and suggest several challenging future directions.

The recent wave of sensor-rich, Internet-enabled, smart mobile devices such as the Apple iPhone has opened the door for a novel paradigm for monitoring the urban landscape known as participatory sensing. Using this paradigm, ordinary citizens can collect multi-modal data streams from the surrounding environment using their mobile devices and share the same using existing communication infrastructure (e.g., 3G service or WiFi access points). The data contributed from multiple participants can be combined to build a spatiotemporal view of the phenomenon of interest and also to extract important community statistics. Given the ubiquity of mobile phones and the high density of people in metropolitan areas, participatory sensing can achieve an unprecedented level of coverage in both space and time for observing events of interest in urban spaces. Several exciting participatory sensing applications have emerged in recent years. For example, GPS traces uploaded by drivers and passengers can be used to generate realtime traffic statistics. Similarly, street-level audio samples collected by pedestrians can be aggregated to create a citywide noise map. In this talk, we will provide a comprehensive overview of this new and exciting paradigm and outline the major research challenges.

Energy consumption is a critical aspect of mobile applications. Excessive drain of battery is one of the key complaints of handheld device users and a success bottleneck for many mobile applications. This paper looks at energy-efficiency from the application development point of view and reviews some answers to questions like what are the key principles in energy consumption in mobile applications and what kind of software solutions can be used to save energy? These issues are discussed in particular in the context of 3G cellular communication. Applications of the ideas are illustrated in energy efficient mobile peer-to-peer systems; BitTorrent and Kademlia DHT.

Given the vast volume of data that needs to be stored reliably, many data-centers and large-scale file systems have started using erasure codes to achieve reliable storage while keeping the storage overhead low. This has invigorated the research on erasure codes tailor made to achieve different desirable storage system properties such as efficient redundancy replenishment mechanisms, resilience against data corruption, degraded reads, to name a few prominent ones. A problem that has mainly been overlooked until recently is that of how the storage system can be efficiently populated with erasure coded data to start with. In this paper, we will look at two distinct but related scenarios: (i)

migration to archival

- leveraging on existing replicated data to create an erasure encoded archive, and (ii)

data insertion

- new data being inserted in the system directly in erasure coded format. We will elaborate on coding techniques to achieve better throughput for data insertion and migration, and in doing so, explore the connection of these techniques with recently proposed locally repairable codes such as self-repairing codes.

The design and functional complexity of medical software has increased during the past 50 years, evolving from the use of a metronome circuit for the initial cardiac pacemaker to functions that include electrocardiogram (EKG) analysis, laser surgery, and networked systems for monitoring patients across various healthcare environments. Software has become ubiquitous in healthcare applications, as is evident from its prevalent use for controlling medical devices, maintaining electronic patient health data, and enabling healthcare information technology (HIT) systems. As the software functionality becomes more intricate, concerns arise regarding efficacy, safety and reliability. It thus becomes imperative to adopt an approach or methodology based on best engineering practices to ensure that the possibility of any defect or malfunction in these devices is minimized.

Given a set

${\cal V}$

of

n

sensor nodes distributed on a 2-dimensional plane and a source node

$s \in{\cal V}$

, the

interference problem

deals with assigning transmission range to each

$v \in{\cal V}$

such that the members in

${\cal V}$

maintain connectivity predicate

${\cal P}$

, and the maximum/total interference of the network is minimum. We propose algorithm for both

minimizing maximum interference

and

minimizing total interference

of the networks. For minimizing maximum interference we present optimum solution with running time

$O(({\cal P}_n + n^2) \log n)$

for connectivity predicate

${\cal P}$

like strong connectivity, broadcast (

s

is the source),

k

-edge(vertex) connectivity, spanner, where

$O({\cal P}_n)$

is the time complexity for checking the connectivity predicate

${\cal P}$

. The running time of the previous best known solution was

$O({\cal P}_n \times n^2)$

[3].

For minimizing total interference we propose optimum algorithm for the connectivity predicate broadcast. The running time of the propose algorithm is

O

(

n

). For the same problem, the previous best known result was 2(1 + ln (

n

 − 1))-factor approximation algorithm [3]. We also propose two heuristics for minimizing total interference in the case of strongly connected predicate and compare our results with the best result available in the literature. Experimental results demonstrate that our heuristic outperforms existing results.

The Integration of MANETs and infrastructure networks such as Internet, extends network coverage and increases the application domain of MANET. Several strategies exist for integration of two networks. Most of them presume that a non-adversarial environment prevails in the network. However such an ideal scenario is not always guaranteed. Nodes many behave maliciously due to scarcity of resources, congestion, or malicious intentions. In this paper a trust based, load aware, and secure gateway discovery for IIM is proposed. It employs the concept of mutual trust and authentication among nodes to prevent malicious activities. However a notable exception is that a node may experience packet-drop due to congestion on route or overflow in the interface queue of an intermediate node; this is not a malicious behaviour. In order to avoid such false malicious behaviours, we employ an effective and adaptive load balancing scheme to avoid congested routes. Thus it would be a novel strategy that ensures the routing in the Integrated Internet and MANET to be trustworthy, secure, efficient, and robust.

MapReduce has emerged as an important programming model with clusters having tens of thousands of nodes. Hadoop, an open source implementation of MapReduce may contain various nodes which are heterogeneous in their computing capacity for various reasons. It is important for the data placement algorithms to partition the input and intermediate data based on the computing capacities of the nodes in the cluster. We propose several enhancements to data placing algorithms in Hadoop such that the load is distributed across the nodes evenly. In this work, we propose two techniques to measure the computing capacities of the nodes. Secondly, we propose improvements to the input data distribution algorithm based on the map and reduce function complexities and the measured heterogeneity of nodes. Finally, we evaluate the improvement of the MapReduce performance.

This paper presents a system to recommend a learning path to an e-learner of a virtual university according to the assessment of linear combination of learner specific parameters and system specific parameters. An online virtual university offers various courses. But learners of this university often face problems due to several constraints of the course. Online recommendation of learning path is an important research issue for virtual learning systems because no fixed learning path will be appropriate for all learners. Generally, inappropriate courseware leads a learner to cognitive overload or disorientation during learning processes, thus it results in reducing learning performance. The developed system implements a simple approach to recommend a learning path to guide a learner from any point of the course. Experimental result also supports the system by manifesting desired output.

Given a set of sensors in a plane or in higher dimension, the strong minimum energy topology problem is to assign transmission range to each of the sensor nodes, so as to minimize the total power consumption. Here the constraint is that the network must be strongly connected. This problem is known to be NP-hard. As this problem has lot of practical application, several approximation algorithms and heuristics has been proposed. There exist a MST based 2-approximation algorithm for this problem having running time complexity of

O

(

n

2

log

n

). In this paper we propose a simple parallel version of the 2-approximation algorithm. We prove that this parallel algorithm is a NC-algorithm and is also cost optimal for a dense graph. We prove that the algorithm has a time complexity

O

(log

n

) and work complexity

O

(

n

2

) when the number of processor used is

O

(

n

2

).

Node localization in Wireless Sensor Networks has been a crucial problem in its domain. It refers to finding the absolute co-ordinates of the nodes in a network using some available information. In this paper we present a novel approach to solve the localization problem using controlled initial placement of nodes followed by conflict minimization. We use the actual location of certain nodes in the deployment field (anchor nodes), along with the neighborhood information of all the nodes to calculate the unknown locations. First we select a subset of nodes with healthy no. of anchors in neighborhood. These are then initialized at the mean locations of their neighboring anchors. Finally we apply conflict minimization (guided by a cost function) on these initial estimates to get their most probable actual locations. The process is executed iteratively, with placed nodes being considered as anchor nodes. The proposed scheme is tested using simulation on a sensor network of 400 nodes showing that it gives accurate and consistent location estimates of the nodes, and mitigates some of the basic shortcomings of previous approaches. It has a no. of advantages among which scalability, high convergence probability towards solution and less time complexity are most notable.

In this paper, we extend recent works on concrete and abstract semantics of structured query languages by considering recursive queries too. We show that combining abstraction of data and widening operators that guarantee the convergence of the computation may be useful not only for static analysis purposes, but also as a sound and effective tool for query language transformations.

We study the problem of using monadic second order (MSO) logic to reason about certain distributed infinite state systems that communicate by exchanging messages. The success of MONA [6] and similar tools shows that a nonelementary worst-case complexity of MSO logic is not relevant in the practice of model checking. Indeed, MSO logic enjoys a satisfactory expressive power for many applications since it subsumes many temporal and program logics. Towards obtaining decidable MSO theory, we consider systems that are not explicitly product-based but, still have an underlying component-based structure.

With the fastest growth of World Wide Web it is quite difficult to track and understand users’ need for the owners of a website. Hence, an intelligent analyzer is proposed to find out the browsing patterns of a user. Moreover the pattern, which is revealed from this deluge of web access logs must be interesting, useful, and understandable. In this paper, a two phases learning algorithm with a modified kernel for radial basis function neural networks is proposed to classify the web pages on time of access and region of access. In phase one a meta-heuristic approach known as differential evolution is used to reveal the parameters of the modified kernel. The second phase focus on optimization of weights for learning the networks. The simulation result shows that the proposed learning mechanism is evidently producing better classification accuracy vis-à-vis radial basis function neural networks.

This paper proposes a distributed algorithm for circle formation by a system of mobile robots. Each robot observes the positions of other robots and moves to a new position. Eventually they form a circle. The robots do not store past actions. They are anonymous and cannot be distinguished by their appearance and do not have a common coordinate system (origin and axis) and chirality (common handedness). Most of the earlier works assume the robots to be dimensionless (points). In this paper a robot is represented as a unit disc (

fat robot

). The robots are assumed to be transparent in order to achieve full visibility. However, a robot is considered as a physical obstacle for another robot. The robots execute the algorithm asynchronously.

This paper presents a finite-buffer single server queue where packets arrive according to a batch Markovian arrival process (BMAP). Partial batch acceptance strategy has been analyzed in which the incoming packets of a batch are allowed to enter the buffer as long as there is space. When the buffer is full, remaining packets of a batch are discarded. The server serves till system is emptied and after that it takes a maximum

H

number of vacations until it either finds at least one packet in the queue or the server has exhaustively taken all the vacations. We obtain some important performance measures such as probability of blocking for the first-, an arbitrary- and the last-packet of a batch, mean queue lengths and mean waiting time of packet. The burstiness of the correlated traffic influences the probability of packet loss which makes buffer management to satisfy the Quality of Service (QoS) requirements.

Tuple Space based Mobile Middleware (TSMM), with tuple space as its coordination medium, provides multiple decoupled behaviors for coordinating interactions between different agents of supported applications. However, maintaining consistency in TSMM is a challenging problem, considering its underlying infrastructure with unpredictable host mobility, sporadic network dynamics, and unreliability in communication links. Existing TSMM maintains consistency by coupling interacting agents, which in turn reduces decoupling abilities of TSMM, thereby restricting development of robust and flexible applications. This paper addresses consistency problems while decoupling agent interactions in TSMM, which renders complete decoupling of interactions. It proposes mechanisms to resolve consistency problems in a fully-decoupled TSMM. Both OUT-consistency and IN-consistency problems are handled in proposed mechanisms. This paper also suggests an approach for formalizing proposed consistency mechanisms in TSMM in order to appropriately analyze reliability and robustness of TSMM as coordination platform for mobile applications. Formalization is carried out using Mobile UNITY.

In this paper we have presented an algorithm to construct a rooted tree with base station as root connecting all the sensor nodes. The tree is constructed with the aim of maximizing the network life-time. It is assumed that all the nodes have same initial energy, but they can adjust their transmission range and thus the amount of energy needed for transmission may vary. The cost of a node is the amount of energy spent by it in each data gathering round. While determining the lifetime of a node, the energy lost in the process of constructing the tree is also considered. Thus the lifetime of a node is its residual energy (initial energy minus energy spent in exchanging messages for construction) divided by its cost. The lifetime of the network is the minimum of the lifetimes of all the nodes in the sensor network. It is also assumed the sensed data is aggregated so that nodes send a fixed sized message to its parent in each data gathering round. The algorithm works in two phases: In the first phase, an initial tree is constructed where the path from a sensor node to the base station consists of least possible number of hops. In the second phase (called fine-tuning) nodes may change their parents if that lead to a reduction in maximum cost of the three nodes involved (the node, its present parent, its future parent). The algorithms for both the phases (initial tree construction and fine-tuning) are distributed where nodes take decision Bbased on the status of its neighbors. Experimental results show that fine-tuning leads to considerable improvement in life-time of the network. The lifetime computed is significantly higher than those obtained by other well-known algorithms for data gathering.

Many organizations today are turning towards the contemporary concept of Cloud Computing, which entails the transfer of their applications (processing both sensitive and non-sensitive data) into online repositories or clouds. Encryption schemes based on homomorphic functions can provide the technology required to protect the confidentiality of this data. Their limitation, however, lies in their ability to process numeric values only. This paper, therefore, is focused on proposing a new encryption scheme relying on character string splitting and ciphering with additional components on the user and cloud side to support querying and modifying character string functions. Ultimately, what eventuates is an environment where a cloud is unaware of the contents of any input character strings, and of the output data during processing.

The Software Transactional Memory (STM) is a promising alternative to lock based concurrency control. Three well studied progress conditions for implementing STM are, wait-freedom, lock-freedom and obstruction-freedom. The wait-freedom is the strongest progress property. It rules out the occurrence of deadlock and starvation but impose too much implementation overhead. The obstruction freedom property is weaker than wait-freedom and lock-freedom. Obstruction Free Transactional Memory (OFTM) is simpler to implement, rules out the occurrence of deadlock and has faster performance in absence of contention. A transaction

T

k

that opens an object, say X, for updating may be in the active state even after completion of update operation on X. Hence, object X cannot be accessed by any other transaction as the current transaction

T

k

is still active. At this point, if another transaction

T

m

wants to acquire object X at this point (for read/write), either

T

k

needs to be aborted or

T

m

must wait till

T

k

finishes. Both of these approaches are detrimental for the performance of the system. Besides, OFTM does not allow

T

m

to wait for the completion of

T

k

. In this paper, a new OFTM implementation methodology has been proposed to allow the second transaction

T

m

to proceed immediately without affecting the execution of the first transaction

T

k

. The proposed approach yields higher throughput as compared to existing OFTM approaches that calls for aborting transactions.

Routing with energy consideration has paid enormous attention in the field of Wireless Sensor Networks (WSNs). In Some WSNs, some high energy sensors called relay nodes are responsible to route the data towards a base station. Reducing energy consumption of these relay nodes allow us to prolong the lifetime and coverage of the WSN. In this paper, we present a Genetic algorithm based routing scheme called GAR (Genetic Algorithm-based Routing) that considers the energy consumption issues by minimizing the total distance travelled by the data in every round. Our GA based approach can quickly compute a new routing schedule based on the current network state. The scheme uses the advantage of computational efficiency of GA to quickly find out a solution to the problem. The experimental results demonstrate that the proposed algorithm is better than the existing techniques in terms of network life time, energy consumption and the total distance covered in each round.

Utilization of MPUs in a computing cluster node for multimedia stream processing is considered. Non-linear increase of processor utilization is described and a related class of algorithms for multimedia real-time processing tasks is defined. For such conditions, experiments measuring the processor utilization and output data loss were proposed and their results presented. A new formula for prediction of utilization was proposed and its verification for a representative set of tasks was performed.

Grid computing has made it possible for users to create complex applications by combining globally distributed data and analysis components and deploy them on geographically distributed resources for execution. Introduction of ad-hoc Virtual Organizations based on on-demand service provisioning further enhances this architectural concept. Job-based paradigms or reliance on relatively static UDDI lead to a failure in offering the complete dynamism of a heterogeneous distributed Grid. A possible alternative is the use of dynamic peer-to-peer (P2P) techniques within a Web Service based Grid to introduce the ability of the network to adapt to resource volatility already established in P2P-based content-delivery models. In this paper, we present the architecture of a demand-driven Web Service deployment framework that allows sharing of data and computing capacity using P2P technology as its backbone. We focus on various issues such as resource availability, scalability and abstraction. Demand-driven resource allocation is based on request parameters and availability of the resources to create the basis for a fully dynamic virtual market place of computational resources.

People in a society in some context lend an ear to their respective world members and so form a network of listening that is passively embedded in the society. This paper defines a model of listening that explains spread of a message creating listen pathways to reach people in the network. A metric listen weight, measures the message relevance perceived by a listener. And accumulation of this weight to a message while traversing a network projects the emergent relevance and intensity of the message spread. We propose a method to engineer listening process to make a society inclusive of listening.

Social Systems and Computational Social Science - the terms often used interchangeably for the same concept speaks on emergence of a new area of research for both social and computing scientists with a hope of better understanding of social phenomena and using the same for delivering services to society. Further, the area assumes larger research interest for exploring behavior of netizens. This paper makes an attempt to follow the contours of research in this area with a proposal on designing of social systems.

In absence of line-of-sight (LOS) signal, localization is a real challenge since multi-path effect takes place due to reflections and/or scattering of signals. In this paper we have assumed that there are few anchor nodes present in the network which broadcast ultrasonic and electromagnetic signals with their known positions. Based on receiving those ultrasonic signals and using received signal strength of the electromagnetic signals, a localization technique has been proposed for finding position of other sensor nodes. On the assumption that a signal may be reflected at most twice before reaching a node, our technique provides on an average 67% reduction of the area of the zone in which the sensor node will be bound to reside.

P2P networking has emerged as a successful Internet based computing paradigm, which can provide an inexpensive platform for distributed computing. The success of P2P network is directly affected by the selfish behaviours of the peers of the P2P network. To solve this issue, we propose a modified grade based incentive mechanism to reward the cooperating peers and to encourage the free rider to become a good contributor based on the peer contribution to network function. Further if the free rider continues to act against the principles of P2P network, it will lose its credit points and eventually it will be eliminated from the P2P network. It also addresses the issue of starvation and introduces the mechanism of trading of incentives to avoid starvation. We design a simulation to verify this approach and the results shows the improvement of fairness in resource sharing and robustness in P2P network.

Social networking has evolved as a basic amenity in today’s interconnected world. Users of social media tools do not always keep up with privacy policies and its adverse effects. It is very common that even experienced users are often caught unaware of actions that happen behind the interface screens of their interconnection devices. Many-a-time mobile application developers take advantage of such user complacency and leak location information from the device (and hence information about the user) to other applications. Though there has been considerable alerts raised on the issue of location information leakage, there are situations wherein applications sneak through these ‘walls’ and connect with devices / applications to extract / query desired information from the firmware. In this work, we provide a in-depth review of literature on this emerging area of social interest, and propose a four-layer context-based authentication framework (4-CBAF) to address location privacy concerns. The 4-CBAF framework provides a facility for users to share pertinent information only if the user specifically authorizes such information sharing. The 4-CBAF is intelligent enough to reduce the number of human interventions that a user should attend to. The effectiveness of the proposed 4-CBAF is also demonstrated for check-in application for Facebook using smart devices.

Crowdsourcing is a distributed problem-solving paradigm. Service oriented crowdsourcing paradigm involves both consumers and service providers. A consumer requests for a service (task); a provider provides that service (does that task); and the providers are paid by consumers for the service as per their satisfaction. The challenge is to select a service provider from a list of providers which can provide maximum satisfaction to the consumer for that service. This work outlines an architectural model using SLURM tool for efficient management of crowd. At the center of this work, we proposed a novel idea of adaptive task scheduling which is based on the customer satisfaction feedbacks. Our approach improves efficiency, and decreases the cost of service to consumers. Experimental results demonstrate the viability of our approach.

Advanced active-safety-critical automotive applications require close coordination of different activities including sensing, processing and actuations, typically performed by different Electronic Control Units (ECU) connected over a communication network. It is imperative that ECUs executing these tasks have a common reference of time. In a distributed system, a periodic resynchronization is a common approach to ensure this. In this paper, we have proposed a new protocol for clock synchronization keeping resource-constraints automotive systems. Our specific contributions include: (a) as compared to standard treatment of drift as a linear function of time, we use a realistic non-linear model of drift, and (b) in order to minimize communication overhead, we propose an algorithm to anticipate the time at which a specific ECU would go out of sync and participate only such identified ECUs for resynchronization instead of all ECUs, as traditionally done. Analytical results show that the proposed protocol incurs minimal communication as well computational load on the ECUs.

Energy-aware network management is extremely important in wireless sensor networks as sensors in the network are powered by battery and it may not be possible to recharge the batteries of sensors after they are deployed. Topology control problem deals with the transmission power assignments to the nodes of a wireless sensor network so that the induced graph topology satisfies some specified properties. Given a set of sensors in the plane, the

Strong Minimum Energy Topology (SMET)

problem is to assign transmit power to each sensor such that the sum total of powers assigned to all the sensors is minimized subject to the constraint that the induced topology containing only bidirectional links is strongly connected. This will allow the sensors to communicate with each other, while conserving battery power as much as possible leading to increased network life time. So this problem is significant in both theory and application. However, the SMET problem is known to be NP-hard. Several heuristics have been proposed for SMET problem by various researchers. In this paper we propose a local search based heuristic for the SMET problem. We prove that our local search based heuristic is a 2-approximation algorithm. We compare our algorithm with several heuristic algorithms available in the literature. Simulation result shows that the local search based heuristic performs better than the existing heuristics.

Scheduling is an emergent area in Grid Environment. It is essential to utilize the processors efficiently and minimize the schedule length. In Grid Environment, tasks are dependent on each other. We use Directed Acyclic Graph (DAG) to solve task scheduling problems. In this paper, we have proposed a new scheduling algorithm called M-Level Sufferage-based Scheduling Algorithm (MSSA) for minimizing the schedule length. It has two-phase process: m-level and sufferage value. M-level is used to calculate the earliest time. Sufferage is used to assign priority and select an optimal machine. MSSA always gives optimal or sub-optimal solution. Our result shows better results than other scheduling algorithms such as MET, MCT, Min-Min and Max-Min with respect to scheduling length and resource utilization.

Preserving Privacy is crucial in distributed environments wherein data mining becomes a collaborative task among participants. Solutions proposed on the lines of cryptography involve use of classical cryptographic constructs in data mining algorithms. Applicability of solutions proposed depends on the adversary model in which it is able to preserve privacy. Existing cryptography based solutions for privacy preserving clustering aim to achieve privacy in presence of semi honest adversary model. For the practical applicability of the solutions in real world settings, support of malicious adversary model is desirable. As per our literature survey, the existing research lacks any fool proof solution for privacy preserving distributed clustering in malicious adversary model. In this paper, we propose privacy preserving distributed K-Means clustering of horizontally partitioned data that supports privacy in malicious adversarial model. The basic construct involves use of secret sharing mechanism clubbed with code based zero knowledge identification scheme. We use secret sharing for privately sharing the information and code based identification scheme to add support against malicious adversaries.

Inspired by the cell motion expressed by endocytosis and exocytosis, we propose a class of membrane systems which uses elementary membrane division and mobility of membranes. We show that this class of mobile membranes using only elementary division and mobility can provide a semi-uniform polynomial solutions for the 4

QBF

problem, ascending to the fourth level in the polynomial hierarchy.

Program Slicing is a popular technique that assists in various software maintenance activities like debugging, program comprehension and regression testing. It is a decomposition technique used for the extraction of program statements affecting the values computed at some point of interest. We propose a technique for computing slices of XML documents. Given a valid XML document, we produce a new XML document (a slice) containing the relevant information in the original XML document according to some criterion. We output a new DTD such that the computed slice is valid with respect to this DTD. Our technique first slices the associated DTD and the DTD slice is used as a slicing criterion in order to produce the associated XML slice.

Web Services are proving to be a convenient way to integrate distributed software applications. As service-oriented architecture is getting popular, vast numbers of web services have been developed all over the world. But it is a challenging task to find the relevant or similar web services using web services registry such as UDDI. Current UDDI search uses keywords from web service and company information in its registry to retrieve web services. This information cannot fully capture user’s needs and may miss out on potential matches. Underlying functionality and semantics of web services need to be considered. In this study, we explore the resemblance among web services using WSDL document features such as WSDL Content and Web Services name. We compute the similarity of web services and use this data to generate clusters using K-means clustering algorithm. This approach has really yielded good results and can be efficiently used by any web service search engine to retrieve similar or related web services.

In the modern epoch, one of the most imperative issues is the nuisance of day-to-day survival of the physically disabled people. Recent development in science and technology has provided a helping hand towards those physically challenged people in the form of different hearing enhancement tools for deaf people, vision enhancement technology for blind people and different audio-vision combinational devices for deaf-blind people. But in true sense, assistive technologies, that too within budget for the lonely deaf-blind people has not been sufficient at all for many years. In our paper, we have tried to introduce SPARSHA which is a low cost refreshable Braille device for deaf-blind and blind people to communicate with other deaf-blind people, blind people and with the nondisabled people. SPARSHA is an electronic device which is connected with a computer and acquires the signal corresponding to alphabet, digit or special symbols and displays the corresponding Braille to represent those alphabet, digit or special symbols. There are six pin to represent the character equivalent to the Braille; similarly we have used six pins to represent SPARSHA. These pins are movable and they can be individually controlled .They can go downward or can go upward. From computer an equivalent signal representing alphabet, digit etc. is sent to the device and the corresponding character is displayed in Braille. Therefore, SPARSHA is a very cost effective and portable Braille display which may provide an affordable way to the blind or deaf-blind people who are facing trouble in communication with the other disabled or non disabled persons in their daily life.

Nowadays information published in the internet has become a common knack for all. As a result volume of information has become huge. To handle that huge volume information, Web researchers are introduced various types of search engines. Efficiently Web-page crawling and resource repository building mechanisms are an important part of a search engine. Currently, Web researchers are already introduced various types of Web search crawler mechanism for the various search engines. In this paper, we have introduced a new design and development mechanism of domain-specific Web search crawler, which uses multilevel domain classifiers and crawls multiple domain related Web-pages, uses parallel crawling, etc. Two domain classifiers used to identify domain-specific Web-pages. These two domain classifiers are used one after the other, i.e., two levels. That’s why we are calling this Web search crawler is a multilevel domain-specific Web search crawler.

Protecting individual sensitive specific information has become an area of concern over the past one decade. Several techniques like

k-

anonymity and

l

-diversity employing generalization/suppression based on concept hierarchies (CHTS) were proposed in literature. The anonymization effectiveness depends on the CHT chosen from the various CHTS possible for a given attribute. This paper proposes a model for constructing dynamic CHT for numerical attributes which can be: 1) generated on the fly for both generalization/suppression; 2) dynamically adjusted based on a given

k

. The anonymized data using our method yielded 12% better utility when compared to existing methods. The results obtained after experimentation support our claims and are discussed in the paper.

In the article we evaluate different text representation methods used for a task of Wikipedia articles categorization. We present the Matrix’u application used for creating computational datasets of Wikipedia articles. The representations have been evaluated with SVM classifiers used for reconstruction human made categories.

Supporting transactions within file systems entails very different issues than those in Databases, wherein the size of writes per transaction are smaller. Traditional file systems use a scheme similar to database management systems for supporting transactions resulting in suboptimal performance. Ext[6] based file systems either involve duplication of blocks, resulting in a reduced write throughput or provide only metadata consistency. The performance provided by a Log-structured file system on traditional hard disk drives is poor due to non-sequential reads that require movement of the read head.

This work presents an implementation of transaction support for log-structured file systems on flash drives. The technique makes use of the copy-on-write capabilities of the hitherto existing log- structured file systems. The major improvement is in the reduction in the overall write-backs to the disk. We provide protocols for recovery from transaction aborts and file system crash. The transaction support and recovery has been implemented in a flash file system[1].

Object oriented databases (OODBMS) are widely used for applications which require support of complex relationships on data. Transactions access the database for data or schema simultaneously. The runtime transactions read/modify the data. The design time transactions read/modify the schema. In a continuously evolving domain, more number of design time transactions is executed along with runtime transactions. Parallel execution of runtime and design time transactions affects the consistency of the database. Hence a concurrency control technique is needed to preserve the database consistency. Several concurrency control techniques using multi-granular lock models have been proposed in the literature. Multi-granular lock model provides better concurrency and is simple to implement. But high concurrency is provided to dynamic databases at the cost of heavy maintenance overhead. In this paper, a concurrency control scheme is proposed using access control lists to provide better concurrency, without the maintenance overhead. The performance of proposed scheme is compared with existing work. It is found that the proposed scheme gives better response time than the existing work and also eliminates the overhead of maintenance of access vectors.

Recent advances in Internet technologies have populated web with large number of services. For effective retrieval of services we have proposed SMap [10] specification that enables service provider to describe a service concealing its business logic. The details of a service includes its constituents, promotion and associated conditions. Here we propose a scheme to store SMaps in a relational database. For service retrieval a query processing system is also proposed. Retrieved services are ranked based on a score computed from structural characteristics of services.

This work demonstrates use of Materialized Views to enhance query performance for partitioned RDF data. Given a query, our system determines which views or combinations thereof can be used to answer it. Break- even analysis for the proposed system has been done based on view materialization and refreshment costs. The system performance was evaluated for 7 query types, 3 having Sub-Obj joins. It shows that our approach reduces query response time by an average of 26% for all query types w.r.t response time using just vertical partitioning. Specifically, for queries with Sub-Obj joins, the average reduction is by 37%. On scaling data up 8 times, the reduction changed from 37% to 79% for queries with Sub-Obj joins, and from 26% to 51% on an average for all query types. With the proposed technique, Semantic Web Applications shall be more interactive since queries having Sub-Obj. joins are expected for them.

The search results offered currently by majority of search portals are horizontal by nature. This denotes that these search engines intend to index as much web pages as possible and present search results based on these web pages. These results often offer generalized results. Focused Crawlers were built to download web pages relevant only to a pre-specified topic. Searching on these kinds of pages is called as Vertical Search, as it attempts to drill down on a single topic, rather than exploring a plethora of other pages on web which are related to search query in one way or another. In this paper, we propose an algorithm which helps a focused crawler decide whether a web page should be downloaded on not. The selection algorithm proposed in this paper makes use of semantic properties of the content to arrive at a decision.