K. V. S. V. N. Raju

Understanding User Behavior using Web Usage Mining

usage mining is about analyzing the user interactions with a web server based on resources like web logs, click streams and database transactions. It helps in discovering the browsing patterns of the user and in relating the pages visited by him. This knowledge can be of help in making business decisions, refining the web site design to derive personalized pages. Web usage mining uses the data mining and the incremental mining techniques to extract the user usage patterns and study the visiting characteristics of the user. Normally, the source for all this data is the web server log, which changes dynamically. Conventional data mining techniques were proposed to be inefficient, as they need to be re-executed every time the log changes which in turn requires multiple database scans. This paper discusses an algorithm to suit the dynamically changing log scenario, based on association rule mining with incremental technique. The algorithm proved to be more efficient as there are redundant no of scans of database; it avoids the generation of candidate keys and allows interactive mining with different supports. The experimental results proving this claim are given in this paper.

Formal Verification of IEEE802.16m PKMv3 Protocol Using CasperFDR

IEEE 802.16m is the standard representing the security architecture for multi hop relay of broadband wireless access. The security sublayer is provided within IEEE 802.16m MAC layer for privacy and access control, which includes the Privacy and Key Management (PKM) protocol. This paper models the PKMv3 key agreement protocol using CasperFDR and analyzes the output. A few Attacks are found in this version. The specifications through which these attacks are found are presented.

Secure Group Key Distribution Using Hybrid Cryptosystem

In group-oriented applications like conferencing, chat groups and interactive gaming myriad messages are sent from one or more sources to multiple users. Multicasting is the optimum technique for such group oriented applications with effective network resource utilization. But maintaining security is a critical issue in group oriented protocols with frequent membership changes. Confidentiality can be achieved through changing the key material, known as rekeying every time a new member joins the group or existing member leaves from the group. Many techniques have been proposed earlier for this purpose. In centralized approach, a single key server is responsible to generate and distribute keys. In decentralized approach, a hierarchy of key managers distributes the keys. In distributed key-agreement protocol, the group members collectively generate and distribute a group key. This paper uses combination of both de-centralized and key agreement approaches to prevent a single point of failures and to improve the reliability as well as the performance of the system. This paper proposes new a technique(SGKP-1), using hybrid key trees, has certain advantages like secure channel establishment for the distribution of the key material, reducing the storage requirements and burden at each member, minimization of time requirement to become a new member of a group. The computational complexity further reduced using both the combination of public and private key crypto systems.

An Effective Calculation of Reputation in P2P Networks

With the advent of sophisticated networking technologies and the related applications, more and more computers are getting hooked to the Internet. This is mainly for utilizing several services ranging from information sharing to electronic transactions. P2P networks which allow decentralized systems, have posed problems related to trust when transactions have to be carried out. Current literature proposes several solutions for trust management and reputation computation. The solutions base their assessment of reputations on the number of successful transactions or on the similarity of the feedbacks. There are some concerns in the feedback ratings if we are not considering the issues like number of transactions, frequency of transactions with the same peer and different peers, age of transaction, how frequently a given peer attends a common vendor, and the number of common vendors between the pairs. This paper puts forward a reputation computation system addressing these concerns. It implicitly allows detection of malicious peers. It also incorporates a corrective mechanism, if the feedbacks are from more number of malicious peers. The implementations and the results that support our claims are also presented.

Integration Testing of Multiple Embedded Processing Components

Integration testing of Complex Embedded systems (CES) and associated interconnection network has not been discussed much in the literature. This paper focuses on integration testing among Embedded Processing Components (EPCs) that are (loosely coupled) interconnected via I/O ports. This paper models EPC as a deterministic FSM and describes its fault model that captures errors in the interface between the Hardware and software. The EPC integration with rest of ES (or another EPC) can be viewed as a system under test (SUT) composed of two parts: EPC1 that requires integration testing with the other EPC2. This paper models EPC integration testing as a general fault localization problem [15] between Communicating FSM. An example shows that the integration testing of two subsystems is a subset of general fault diagnostic problem and the faulty machine can be identified during integration.

Trust Based Participant Driven Privacy Control in Participatory Sensing

Widespread use of sensors and multisensory personal devices generate a lot of personal information. Sharing this information with others could help in various ways. However, this information may be misused when shared with all. Sharing of information between trusted parties overcomes this problem. This paper describes a model to share information based on interactions and opinions to build trust among peers. It also considers institutional and other controls, which influence the behaviour of the peers. The trust and control build confidence. The computed confidence bespeaks whether to reveal information or not thereby increasing trusted cooperation among peers.

Credibility Based Reputation Calculation in P2P Networks

Trust management and reputation computation is a major issue in P2P networks. This paper presents a method for reputation calculation based on credibilities of peers. The credibility of a peer is found based on feedback similarity, vendor similarity, interactions similarity and transaction age. The reputation computation system also allows a corrective mechanism, if the feedbacks are from more number of malicious peers. Results that support our claims are presented.

SenSim: sentence similarity based on the concept of relevance

Sentence similarity measure is an important concern for researchers for text retrieval in areas such as text mining, web information retrieval, decision making and question matching. Existing methods for computing sentence similarity have been adopted and derived from word similarity measure and concentrated on multidimensional space. In this proposed work, words from the sentences are considered and their respective taxonomies are built using Word Net. The evolved taxonomies are merged to develop Hierarchical Ontologies. A comparison is done using an empirical formula (SenSim) on the Hierarchical Ontology developed from the two sentences and found that our proposed method gives fairly good sentence similarity measure.

Optimal exposure sets for high dynamic range scenes

The dynamic range of many natural scenes is far greater than the dynamic range of the imaging devices. These scenes present a challenge to the consumer digital cameras. The well-known technique to capture the full dynamic range of the scene is by fusing multiple images of the same scene. Usually people combine three or five different exposures to get the full dynamic range of the scene. Some cameras like Pentax K-7, always combines fixed exposures together to produce the output result. However, this should be adaptive to the scene characteristics. We propose an optimal solution for dynamically selecting the exposure sets.

Fair and secure m-cash

The paper proposes an efficient M-commerce digital cash framework in which customers can get digital cash from any bank and vendors can encash digital cash at any bank. Thus, the system provides flexible and anonymous digital cash. The proposed scheme has the following features: (i) Realizes practically feasible anonymous digital cash by using group signatures; (ii) Achieves fairness in goods exchanged between customers and vendors by using trusted, tamper-resistant hardware; (iii) Installs e-wallet as a secure Java smart card applet making it practical for ease of implementation and security; (iv) Does not require an on-line trusted third party (TTP); (v) Allows dispute resolution when necessary. The paper describes the protocol and the underlying framework, and analyzes protocol in terms of security and feasibility.