PHD, Kent State University, 2020, College of Arts and Sciences / Department of Computer Science

Traditionally, the multi-party telepresence system is supported by one or more servers called Multipoint Control Unit(MCU). These servers are expensive, involve the third party in the system, and also bottleneck for large scale implementation. So, this dissertation presents protocols for autonomous Peer-to-Peer(P2P) implementation of Crowd-scale Telepresence System. The protocols use multiple features from widely adopted P2P network, Gnutella. The proposed protocols and strategies are designed based on the Principle of Distributed Computing (PDC) and the Principle of Priority-based Resource Allocation(PPRA). These principles are considered to address three of the four identified challenges of CMTS implementation, (1) Computational Challenge, (2) Temporal Challenge, and (3) Overcrowding Challenge. The fourth one is the visual challenge which is left for future work. The PDC is used to address the first two challenges by distributing of MCU's workloads among participating peers. The MCU consists of a Multipoint Controller(MC) and one or more Multipoint Processors(MP). For distributed MCU, the optimal placement of MC and MPs in the P2P overlay network is necessary, which is time-consuming because of exponential search space. So, a phase-based design approach is considered. For optimal placement of MC, three incremental protocols, such as GAncestor, ZePoP, and ZePoP-ε are presented. Then, multiple methods are discussed to place the MPs around the optimal MC. For supporting the desired frame rate, two versions of progressive timer management schemes are used at MPs. The protocol ZePoP-ε is designed based on PPRA that emphasis to properly utilize the limited resources of the P2P network. Thus, PPRA is used to address the overcrowding challenge as well as the temporal challenge. It is used to design a profit-based stream collection mechanism of ZePoP-ε for maximizing a Dynamic Role and Demand based Index (DRDI) in bounded waiting time. The proposed protocols and methods co (open full item for complete abstract)

Committee: Javed I Khan (Advisor); Cheng Chang Chang Lu (Committee Member); Gokarna P Sharma (Committee Member); Murali Shanker (Committee Member); Jun Li (Other) Subjects: Computer Science

PhD, University of Cincinnati, 2020, Engineering and Applied Science: Computer Science and Engineering

In recent days, revolution of the internet and the telecommunication with new smart devices have made big data booming fame. Tons of data are stored for each one of us in different servers like Facebook, Tweeter, and other apps. However, if collecting data looks as an easy task, then classifying and processing big data is a difficult challenge that ought to be addressed among many issues. Taxonomy is a mechanism of data labels that can perform economic and effective analysis of associated values. In term of processing and classifying big data, most of the large companies are busy trying to find out the best and easy way to do that. While they succeeded in some respects they are still struggling on other sides. Moreover, some of those technology companies like Microsoft produce Azure cloud service that can help in term of big data (storage, process, and develop). Amazon also have their own too. In this work we used Azure cloud. This work is to focus on this important field where big data is used for health care in order to diagnose diseases before they occur. One of the major academic work on health data (Breast cancer) could be divided into two big objectives like statistic work and analytical work, with an emphasis on predictive analytics as it is emerging as a transformative tool that can enable more proactive and preventative treatment option. Moreover, in this work, we create a new platform that works perfectly with big data to create our model and then the sub-aim is to compare our results with other researchers. Later on, in this research we focus on pandemic of COVID-19 wand we used a dataset from internet to read it and analysis to understand and predict whether the patient going to release, isolate or decease. The second aim of this research is to focus on the economy datasets. Loans have been continuously growing up to trillions of dollars in the US, which makes it difficult for the lenders to study every case individually. Borrowers tend to acquir (open full item for complete abstract)

Committee: Dharma Agrawal D.Sc. (Committee Chair); Rui Dai Ph.D. (Committee Member); Chia Han Ph.D. (Committee Member); Wen-Ben Jone Ph.D. (Committee Member); Haider K. Raad Ph.D. (Committee Member) Subjects: Computer Science

PhD, University of Cincinnati, 2010, Engineering and Applied Science: Computer Science and Engineering

In this dissertation, we present a novel approach to the problem of distributed (peer-to-peer) backup. Our approach requires that data not be transferred more than two-hops from it source and that each peer store exactly the same amount of data as it distributes to be backed up. These two requirements address two import features of any distributed backup solution - trust and fairness. In a social network, the hop distance requirement means that in the worst case, a peer's data is backed up in the local storage of a friend of a friend (FoaF). Our assumption is that this offers a higher degree of trust than simply choosing a random peer. We achieve fairness through the requirement that peers store exactly the same amount of data that they distribute for backup. To facilitate this requirement, our approach uses symmetric exchanges of data. This not only supports fairness, but also enhances trust by introducing a vested interest between peers to preserve the data that they are storing. We call our approach the fair two-hop exchange scheme, or FTHES. We show that existing f-factor theory and algorithms can be used to compute an FTHES. Then we introduce and prove a fundamental existence theorem which states that an FTHES always exists under two fairly weak conditions. This theorem leads to a linear time sequential algorithm and an efficient distributed algorithm. We also prove a theorem stating that at most 2n-3 exchanges are needed to backup all of the data in our scheme and later conjecture that this may actually have a lower bound of n. Finally, we present an application of the FTHES in a content management system.

Committee: Kenneth Berman PhD (Committee Chair); Fred Annexstein PhD (Committee Member); Jerome Paul PhD (Committee Member); Yiming Hu PhD (Committee Member); Dan Ralescu PhD (Committee Member) Subjects: Computer Science

MS, University of Cincinnati, 2006, Engineering : Computer Science

This thesis addresses the problem of neighbor selection and congestion in a peer-to-peer network constructed with a mixture of public and guarded nodes. Current overlay construction typically relies on a nearest neighbor algorithm and two-way communication between peers. In many environments two-way communication is not a possibility. Most neighbor selection algorithms are not able to construct an overlay under such conditions since not all peers are able to send and receive messages. We propose a neighbor selection algorithm that incorporates a node's lifetime and number of resources to overcome these obstacles. We evaluate the performance of this neighbor selection algorithm by measuring the congestion in the network. In actual environments where firewalls exist our neighbor selection algorithm has been shown to reduce the total congestion by 2.5% compared to a nearest neighbor algorithm.

Committee: Dr. Fred Annexstein (Advisor) Subjects: Computer Science

MS, University of Cincinnati, 2003, Engineering : Computer Science

This thesis introduces the concept of service oriented architecture in P2P domain and implement a Java framework for building a service-sharing P2P network. In our network peers will share services, which are defined as fine grained, modular and self contained applications. This is analogous to the way peers share files in P2P networks like Napster. Using our framework, services are capable of communicating and collaborating with other services and form bigger and wholesome applications from smaller modules. Some examples of services are natural language translation, currency conversion, and graphic rendering of scenes. Peers acting as service providers publish their services to a central register and are responsible for classifying themselves into groups and provide a description of what they do. Peers acting as clients search for available services on the register and are given the remote references of the possible matches with whom a direct contact could be established. Programming of sharable services in P2P domain requires special considerations. The task of providing a service must be keptsimple and yet the service model should be flexible to allow for different kinds of services. Using a Java RMI based class library, we attempt to provide the necessary infrastructure and some useful features to facilitate programming of sharable services. These include asynchronous remote method invocation, remote object creation, a volunteering system to avoid freeloading of resources, wrapper for interfacing with programs in legacy code and encrypted method invocations. This service sharing framework can easily be adapted for building a parallel meta-computer using available peers. This is illustrated by implementing a parallel graphic rendering application, in which the nature of overhead involved per peer was shown to be effectively constant.

Committee: Dr. John Franco (Advisor) Subjects: Computer Science

MS, University of Cincinnati, 2002, Engineering : Computer Science

File sharing has been the most popular service for which peer-to-peer (P2P) networks have been used in recent years and it is expected to remain so for a long time. A P2P file-sharing service makes each user's machine a peer in a network of peers and allows the users to share files. Users can issue queries to the network to find out the locations of the files of their interest. The average size of a P2P network is much larger than the average size of a client-server network. Each node in the network receives a lot of queries every second and so has to search through its file indices several times every second to obtain the results for the queries. The time taken by the nodes to respond to the queries can have a large impact on the overall performance of the network. Using proper indexing techniques can reduce the query response time significantly. This thesis work focuses on the study of advanced indexing techniques that can be used to index the filenames in P2P nodes that participate in a file-sharing service. As a case study, a special kind of P2P node, the supernode, was chosen. A supernode provides proxy and indexing services to nodes on slower network connections. Nodes connect to and disconnect from the supernode arbitrarily. So the problem of indexing filenames in the supernode becomes dynamic in nature. We consider two fundamentally different algorithmic models (the Merged Tree Model and the Vector Model) for dynamic indexing. The merged tree model is the model in which the indices obtained from the connecting nodes are combined with a single primary index maintained by the supernode. The vector model is the model in which the indices obtained from the connecting nodes are stored individually as a vector of indices. We provide a formal framework for analyzing the performance of the different models. Furthermore we use simulations to verify the formal framework and to determine precise constant factors. We conclude by demonstrating that a hybrid algorithm is (open full item for complete abstract)

Committee: Dr. Fred Annexstein (Advisor) Subjects: Computer Science

Bachelor of Science (BS), Ohio University, 2010, Computer Science

Massively multiplayer online games have been dramatically increasing inpopularity since their inception. Unfortunately, the commercial technologies to support these games have remained largely the same. Similarly, peer to peer technology has flourished dramatically. The marriage of massively multiplayer online worlds and peer to peer is the next logical step in the development of online worlds. Many such systems have been proposed but few take cheating into account making these systems infeasible for commercial deployment. New research in cheat detection and prevention attempts to mitigate some of these issues. These cheat prevention methodoligies have advantages and disadvantages in their approaches and implementations and are not entirely complete. This thesis evaluates the various cheating scenarios and prevention techniques, compares the most prominent of these systems, and briefly introduces a new Hybrid topology for a Pseudo-P2P system for future consideration and development.

Committee: Shawn Ostermann PhD (Advisor) Subjects: Computer Science

Master of Sciences, Case Western Reserve University, 2010, EECS - Computer and Information Sciences

Live streaming applications, especially those based on peer-to-peer networks, are becoming popular nowadays. It is widely known that there are still some performance challenges on transmission and scalability in peer-to-peer live streaming system. This thesis focuses on improving transmission efficiency in live media streaming and improving scalability in peer-to-peer live streaming systems. First, we improve transmission efficiency in live media streaming by studying chunk scheduling algorithms which include Greedy, Rarest First, Mixed, Random and our proposed Alternate algorithms, and delivery methods which include Push and Pull methods. Based on the evaluation of startup latency and streaming continuity for different chunk scheduling algorithms and delivery methods, we discuss how to make an optimal choice for better transmission efficiency. Second, we improve the scalability for peer-to-peer live streaming system by utilizing our incentive model, a bank incentive model, which can encourage peers to make more contribution in order to obtain extra benefits from their neighbors and the system. As well as applying encouragement to the peers, our incentive model can support multiple platforms and the extensibility of incentive strategies.

Committee: Shudong Jin (Committee Chair); Michael Rabinovich (Committee Member); Swarup Bhunia (Committee Member) Subjects: Computer Science

Master of Science, University of Akron, 2009, Computer Science

Reverse engineering malware has been an integral part of the world of security. At best it has been employed for signature logging malware until now. Since the evolution of new age technologies, this is now being researched as a robust methodology which can lead to more reactive and proactive solutions to the modern security threats that are growing stronger and more sophisticated. This research in its entirety has been an attempt to understand the in and outs of reverse engineering pertaining to malware analysis, with an eye to the future trends in security.Reverse engineering of malware was done with Nugache P2P malware as the target showing that signature based malware identification is ineffective. Developing a proactive approach to quickly identifying malware was the objective that guided this research work. Innovative malware analysis techniques with data mining and rough sets methodologies have been employed in this research work in the quest of a proactive and feasible security solution.

Committee: Kathy J. Liszka PhD (Advisor) Subjects: Computer Science; Engineering; Experiments; Systems Design