Doctor of Philosophy (PhD), Ohio University, 1996, Electrical Engineering & Computer Science (Engineering and Technology)

The stringent timing requirements imposed by distributed multimedia applications have raised questions about the adequacy of continuous media support in the current commercial operating systems. The main objective of this research is to study the requirements, also known as Quality of Service (QOS), of multimedia applications and develop a QOS management scheme to support an efficient multimedia networking environment. An integrated QOS management architecture is proposed to maintain synchronization among different continuous media objects. The primary goal of this research is to present a set of key application QOS parameters and map these requirements through all the layers of our proposed integrated QOS management framework. Emphasis is placed on four performance criteria for continuous media communication: Throughput, transmission delay, delay variations, and error rates. End-to-end QOS guarantees are ensured by dynamic QOS control that is orchestrated by a protocol entity called the QOS negotiation agent. The QOS negotiation agent expands on the supplier-consumer paradigm, where the consumer requests a service (product) from the supplier and the supplier delivers the service if the consumer agrees on the cost. The QOS negotiation agent is an end-point resource manager that orchestrates the required resources for tasks performed in the application and transport subsystems. Using different negotiation protocols, the consumer side QOS agent negotiates QOS requirements with the network resource management and the supplier side QOS agent. The QOS negotiation agent was programmed as a Windows NT background process handling all the negotiation threads by monitoring the operating system and network resources while the multimedia application was running in the foreground. To test the effectiveness of the QOS negotiation agent, we developed an AVI video player which could be executed with the QOS negotiation capabilities in either enabled or disabled mode. The experiment (open full item for complete abstract)

Committee: Mehmet Celenk (Advisor) Subjects:

Master of Science (MS), Ohio University, 2005, Computer Science (Engineering)

The existing Quality-of–Service Resource Allocation Model (QRAM) is a generic method of allocating resources to applications while maximizing the utility of the system; it has been extended to the network domain to allocate applications with bandwidth on multiple links. When the utility of an application depends on response time, the QRAM network model is limited since it does not consider the impact of link load on response time. It further assumes that the network propagation delay is negligible, which also impacts application response. The thesis proposes an extension to the QRAM network model, (EQRAM), which addresses both these limitations. We also show that the response time achieved by assigning resources to an application using EQRAM is better than the response time achieved when using QRAM. This thesis also proposes a novel means of obtaining the resource-utility table for the applications quickly using OPNET.

Committee: Carl Bruggeman (Advisor) Subjects: Computer Science

Master of Science in Electrical Engineering, Cleveland State University, 2021, Washkewicz College of Engineering

In this thesis, we draw attention to the problem of cross-service attacks, that is, attacks that exploit information collected about users from one service to launch an attack on the same users on another service. With the increased deployment and use of what fundamentally are integrated-services networks, such as 4G/LTE networks and now 5G, we expect that cross-service attacks will become easier to stage and therefore more prevalent. As running example to illustrate the effectiveness and the potential impact of cross-service attacks we will use the problem of account association in 4G/LTE networks. Account association attacks aim at determining whether a target mobile phone number is associated with a particular online account. In the case of 4G/LTE, the adversary launches the account association attacks by sending SMS messages to the target phone number and analyzing patterns in traffic related to the online account. We evaluate the proposed attacks in both a local 4G/LTE testbed and a major commercial 4G/LTE network. Our extensive experiments show that the proposed attacks can successfully identify account association with close-to-zero false negative and false positive rates. Our experiments also illustrate that the proposed attacks can be launched in a way that the victim receives no indication of being under attack.

Committee: Ye Zhu (Committee Chair); Yongjain Fu (Committee Member); Sui-Tung Yau (Committee Member) Subjects: Computer Engineering; Computer Science; Electrical Engineering; Information Technology; Technology

Master of Science in Engineering, University of Akron, 2020, Electrical Engineering

This thesis proposes a reinforcement learning (RL) driven software-defined networking (SDN) routing scheme for the situation-awareness and intelligent networking management. Firstly, the existing SDN network monitoring technique is applied to track the quality of service (QoS) parameters (link delay and packet loss). Afterward, the QoS data are fed to the RL algorithm in order to achieve situation awareness in SDN routing. The performance of the proposed RL-enabled routing scheme is evaluated in the simulation section by considering various network scenarios, including network congestion. Finally, the end-to-end delay, the episode reward, and the probability of path selection are recorded for each case. According to the outcomes, the proposed scheme intelligently select the efficient data path according to the current state of the network. Moreover, the end-to-end delay is compared with the Dijkstra algorithm, demonstrating the superiority of RL-enabled dynamic routing strategy over static. Additionally, the scalability of the algorithm is tested with multiple controller SDN.

Committee: Jin Wei-Kocsis (Advisor); Kye-Shin Lee (Committee Member); Hamid Bahrami (Committee Member) Subjects: Computer Engineering; Computer Science; Electrical Engineering

MS, University of Cincinnati, 2015, Engineering and Applied Science: Computer Science

Continued evolution of Wireless Body Area Networks (WBANs) has made effective monitoring of vital parameters of a person much faster and efficient, thereby providing better personal healthcare. Sensor nodes of a WBAN acquire critical physiological parameters like heartbeat, neural activity, limb motion, muscle movement and fatigue, temperature, etc. that are monitored by a physician. Important factors in the acceptance of WBAN performance are energy efficiency and the Quality of Service (QoS) supported for such critical data that impact human lives. The sensor nodes of a WBAN are highly constrained in terms of their battery life. Most of the work till date on WBANs uses a star topology which employs single hop communication. This work discusses various factors that affect energy efficiency in a WBAN and establishes the need for a multi-hop tree based topology. It also studies the need for QoS in WBANs and existing support provided by the current Body Area Sensor Network (BASN) Standard. This thesis tackles the all important challenge of providing QoS in autonomous and neighbor-aware multi-hop WBANs in significant detail spread across multiple chapters. In case of independent, autonomous multi-hop WBANs, the aforementioned issue is resolved by implementing a two layer priority-mapping scheme over a reactive Media Access Control (MAC) layer designed to alter durations of the access phases involved as per QoS requirement. In the case of neighbor-aware WBANs, a framework is defined under which a cooperative inter-WBAN routing scheme is implemented through power based weight assignment and fault detection is carried out by employing Kosaraju's two-pass algorithm that discovers the strongly connected components in the network deployment graph.

Committee: Dharma Agrawal D.Sc. (Committee Chair); Raj Bhatnagar Ph.D. (Committee Member); Prabir Bhattacharya Ph.D. (Committee Member) Subjects: Computer Science

PhD, University of Cincinnati, 2015, Engineering and Applied Science: Computer Science and Engineering

Wireless Body Area Network (WBAN) is a low-power Personal Area Network involving sensor nodes (SNs) that sense and relay physiological data to a central station. WBANs are new and still evolving. We try to address three open research areas involving WBANs. The limited energy budget in WBANs necessitates energy conservation to prolong the network lifetime. The first challenge we try to address is related to improvement of the lifetime of a WBAN, given the small sizes of body sensor nodes (SNs) and the limited battery power that they run on. We proposed a dual-prediction framework for improvement of network lifetime. The framework allows for minimizing data transmission involving four important body parameters by reconstructing their information by time series prediction at reception. A sample elimination algorithm further optimizes the framework performance. We enhanced the framework by reducing the sampling frequency and implementing the algorithm on top, increasing the network lifetime further. The missing samples were reconstructed by interpolation at the receiver. We probed the effects of adaptive sampling and evaluated the increase in battery lifetime in WBANs. We then tried to test the behavior of a WBAN in the presence of other WBANs around it and check the issues faced by WBANs. Wireless systems can face severe interference problems if they use the same communication channels at a time. There are issues related to data routing because the critical nature of WBAN data requires assured communication of body data. For optimum network utilization, efficient scheduling of transmissions in multiple co-existing WBANs is important in order to avoid intra and inter-WBAN interference and for a graceful coexistence. We propose that inter-WBAN interference can be avoided by a QoS based MAC scheduling approach and that intra-WBANs interference can be circumvented by fuzzy scheduling of intra-WBAN transmissions. We also propose to use interference to the benefit of WBA (open full item for complete abstract)

Committee: Dharma Agrawal D.Sc. (Committee Chair); Raj Bhatnagar Ph.D. (Committee Member); Prabir Bhattacharya Ph.D. (Committee Member); Chia Han Ph.D. (Committee Member); Marepalli Rao Ph.D. (Committee Member) Subjects: Computer Science

Doctor of Philosophy (PhD), Wright State University, 2009, Computer Science and Engineering PhD

The demand of network capacity has been increasing steadily with more users than ever connected to the Internet through broadband access and the popularity of video based applications, such as YouTube. Optical wavelength division multiplexing (WDM) networks are expected to form the next-generation backbone network and to fulfill the insatiable appetite for bandwidth.Wavelength routed WDM optical networks offer the granularity of switching at a fiber, waveband, or a wavelength level. The finest granularity offered is at a wavelength level by provisioning lightpaths for different clients/services. All-optical packet switching is still deemed technically infeasible and its competitiveness as a backbone technology is debatable. Optical burst switching (OBS) presents itself as a promising technology for bridging the gap between optical wavelength switching and optical packet switching. OBS operates at the sub-wavelength level and is designed to improve the bandwidth utilization of wavelengths by exploring statistical multiplexing to deal with bursty traffic, and is therefore more resource efficient than optical wavelength switching. In OBS networks, arriving data packets (e.g., IP packets) are assembled at the ingress OBS nodes to form a data burst. A burst control packet (CP) is sent on a control channel ahead of the data burst to reserve resources and configure the switches along the route traversed by the data burst. In this dissertation, we will explore several important and challenging issues in OBS networks in order to improve the utilization of network resource. To reduce the switching overhead, small bursts may be groomed to reduce resource waste and switching penalty. We have studied the per-hop burst grooming problem where bursts with the same next hop may be groomed together to minimize the number of formed larger bursts and strike a proper balance between burst grooming and grooming cost, assuming all the network nodes have the grooming capability. In order t (open full item for complete abstract)

Committee: Bin Wang PhD (Advisor); Krishnaprasad Thirunarayan PhD (Committee Member); Yong Pei PhD (Committee Member); Yuqing Chen PhD (Committee Member); Waleed Smari PhD (Committee Member) Subjects: Computer Science

MS, University of Cincinnati, 2001, Engineering : Computer Science

An ad hoc network is a collection of wireless mobile nodes dynamically forming a network without the use of any existing network infrastructure orcentralized administration. Dynamic routing protocols have been developed that allow such nodes to communicate using multi-hop routes, but existing protocols typically provide only best-effort service. In particular, they do not directly provide any way to control the resource usage in the network. This thesis focuses on the management of two such resources, viz., battery power and the bandwidth available to the nodes. Both are premium resources in a wireless network with mobile nodes. We develop our techniques as extensions of “on-demand” routing protocols because of their superior performance as reported in current literature. These are routing protocols that compute routes on an “as needed” basis. All protocols and techniques developed in this thesis are evaluated using extensive simulations using a multi-layer network simulator. To manage battery power, we develop a technique to make on-demand routing protocols energy-aware. The goal is to increase the operational lifetime of the network where mobile nodes are operating on battery power alone, and batteries cannot be recharged or replaced. The idea here is to balance the residual energy levels of the batteries in the mobile nodes by dynamically shifting routing loads from nodes with low battery levels to those with higher battery levels. The technique also switches off network interfaces strategically to conserve energy. Simulations show significant improvement of the energy budget of the network as a whole resulting in increased operational lifetime. The most improvement is noticed for high loads and modest mobilities. To understand the question of bandwidth usage better, we develop empirical techniques to determine network capacity and utilization when the network is running commonly used routing and medium access protocols. While the technique is general, we specifi (open full item for complete abstract)

Committee: Dr. Samir Das (Advisor) Subjects: Computer Science

MS, University of Cincinnati, 2001, Engineering : Computer Science and Engineering

The next generation of wireless mobile network has been designed to support acombination of both real-time service (such as voice, video transmission) and non real-time service (such as data transmission). To support such a wide range of traffic, Quality of Service (QoS) has to be taken into consideration while designing the infrastructure itself for the wireless mobile networks. One of the central issues characterizing the performance is how the handoff is handled. Two hybrid reservation handoff scheme, one is priority reservation and the other with preemptive procedure, for integrated real-time and non real-time service wireless mobile network have been proposed and analyzed in this dissertation. These handoff schemes are service dependent. The system is modeled by a multi-dimensional Markov chain and a numerical analysis is presented to estimate blocking probabilities of originating calls, forced termination probability of real-time service handoff requests calls, and average transmission delay of non real-time service calls. This scheme is also simulated using extensive runs and the results are observed to agree fairly closely. In the proposed scheme, forced termination probability of real-time service calls is seen to be significantly decreased. The probability of packet loss of non real-time transmission is made to be negligibly small, as a non real-time service handoff request in the queue can be transferred from the queue of current base station to another one.

Committee: Dr. Dharma Agrawal (Advisor) Subjects:

MS, University of Cincinnati, 2007, Engineering : Computer Science

The IEEE 802.11n standard provides high data transmission rates with advanced signal processing techniques. However, the higher layer protocols do not effec-tively leverage this increase in physical layer performance. There is often a lot of error control and other performance improvement mechanisms implemented at the MAC (Medium Access Control) and higher layers to account for wireless link transmission failures due to a noisy channel, or poor transmission quality. In this research work, the performance of higher layer schemes with respect to the new physical layer improvements of the IEEE 802.11n standard is evaluated. Also, two schemes, namely the Adaptive MAN (modulation, number of antennas, frame aggregation constant) and the Conditional FEC (forward error correction) scheme have been proposed and their performance has been evaluated through extensive simulations. The first scheme, namely the adaptive MAN scheme has been proposed, in which physical layer parameters such as the number of an-tennas are dynamically changed depending upon the channel conditions. This scheme shows reduced throughput variance, lesser retransmission requests and better channel utilization. The second scheme, called conditional FEC scheme has also been proposed to reduce the error control overhead under “good” chan-nel conditions. This scheme uses a combination of the adaptive physical layer parameters as well as the existing channel conditions to decide whether or not to use the FEC block of the physical layer. Simulation results show reduced latency and redundancy and thus improved overall system performance. Finally, the conclusions and future work is also presented.

Committee: Dr. Qing-An Zeng (Advisor) Subjects: Computer Science

MS, University of Cincinnati, 2007, Engineering : Computer Science

Handoff in the IEEE 802.11 wireless local area networks (WLANs) occurs whenever the mobile station (MS) changes its association from one access point (AP). Previous studies have shown that the handoff schemes employed in the IEEE 802.11 networks do not meet the strict delay constraints placed by many multimedia applications. The problem is further compounded by the fact that limited coverage areas of APs employed in existing IEEE 802.11 WLANs create frequent handoffs. In this research work, four novel handoff schemes namely, FHAP (Fast Handoff by Avoiding Probe wait), APFH (Adaptive Predictive Fast Handoff), DPH (Dynamic Predictive Handoff) scheme and AP-DPH (Adaptive Preemptive Dynamic Predictive Handoff) scheme are proposed for the IEEE 802.11 WLANs to reduce handoff latency and improve the QoS of multimedia applications. The performance of all the schemes are analyzed through extensive simulations.

Committee: Dr. Qing-An Zeng (Advisor) Subjects: Computer Science

PhD, University of Cincinnati, 2006, Engineering : Computer Science and Engineering

Most existing research in wireless sensor networks (WSNs) has been focused on proposing protocols to facilitate either energy effient communication or sensing operation in WSNs. We believe that there exist fundamental limits in the operations and the parameters of WSNs. Characterizing these fundamental limits, establishing the relationships between the parameters and the WSN performance, and identifying the tradeoffs among different performance measures can provide us insights in designing sensor networks. In this dissertation, the QoS properties of WSNs, including the coverage, reachability, and delay, are analyzed in an analytical framework. How the network parameters, such as the node density, sensing range, transmission range, impact the aforementioned QoS properties are investigated by abstracting the randomly deployed WSNs as random geometric graphs. Both homogenous and heterogenous sensor networks are addressed. Based on the analytical model, optimal sensor network parameters can be chosen to improve the network performance while achieving pre-specified QoS constraints. Besides this abstraction, practical issues affecting the QoS performance of WSNs are also examined by doing extensive simulations in this dissertation. Specifically, results are obtained for the QoS properties in the flooding of WSN when a cross-layer approach is adopted, and other practical issues such as channel fading and MAC layer contention are taken into account. The results show that exchanging more information across the layers can help improve effciency of the protocol. It is further shown that wireless fading helps to reduce the hop count, although it makes the transmission range less predictable and limits the improvement in the performance of such cross-layer design in WSNs.

Committee: Dr. Dharma Agrawal (Advisor) Subjects: Computer Science

MS, University of Cincinnati, 2005, Engineering : Electrical Engineering

The original wireless and mobile communication system used to carry voice calls only. In recent years, data service has been introduced into the system. The analysis in wireless communication has been performed extensively on real-time and non-real-time calls. Now traffic from Internet applications turns out to have more different types of quality of service (QoS) requirements. As we know, DiffServ (DS) provides an efficient way to work through different types of traffic in scalable wireline networks. To ensure end-to-end QoS support, we need to extend DiffServ from the wireline network into the wireless radio access network. The mobility natural of wireless communication system brings another level of complexity we need to take care of: the handoff event. Our proposed preemptive priority scheme deals with differenciated services in wireless and mobile communication system and gives higher priority to handoff calls than originating ones. Based on this system model, dynamic channel allocation scheme is presented. Giving the concept of logical channel or basic bandwidth unit (BBU), data service can benefit from multiple BBU assignment and reduce its service time. That in turn, allows the system to accept more user requests under certain traffic load. When traffic load is high, system can reclaim the additional channels assigned to these services to serve other incoming requests. Another topic of this thesis is about rate adaptation scheme. All previous approaches for resource management have a common assumption that the wireless channels are identical and static. Such homogeneous and time-invariant hypotheses reduce the complexity in theoretical analysis and simulation. However, a good radio frequence (RF) link allows higher rate of transmission with the help of enhanced coding and modulation methods. With rate adaptation scheme, we get better system utilization and can perform more specific preemption activities.

Committee: Qing-An Zeng (Advisor) Subjects:

PhD, University of Cincinnati, 2005, Engineering : Computer Engineering

One of the recent trends in the telecommunication industry is the integration of voice and data services in wireless and mobile networks. In the face of this more complex traffic mix, where each service may have different requirements, providing quality of service (QoS) guarantees poses a difficult challenge for network service providers. Call admission control (CAC) and channel allocation can help satisfy these guarantees. The objective of this research is to develop resource allocation schemes in integrated wireless and mobile networks. Our research work demonstrates that resource borrowing is an effective approach to achieve these targets. Several resource allocation schemes with channel borrowing have been proposed. The channels in each cell are divided into two parts and pre-allocated for real-time and non-real-time services. One type of service is allowed to borrow unused channels from the other under certain conditions provided that the QoS of the latter is not significantly affected. We take advantage of the delay insensitivity of non-real-time services and allow real-time service calls to preempt non-real-time service calls. We model the system using the Markov process to understand the effects of channel borrowing on QoS for different types of traffic. Widely used CDMA systems are mainly interference limited and the transmit power is an important resource. In such systems, resource borrowing occurs in terms of power borrowing. Based on the current measure of the received power, we develop an algorithm to estimate the updated total received power due to a new call. Using this algorithm, we propose a total received power based CAC for integrated CDMA systems. The IEEE 802.16 media access control (MAC) layer is designed to support desired level QoS for different services. We present and analyze an adaptive resources allocation algorithm based on QoS degradation. Unlike the preemptive channel borrowing, service degradation allows the low priority service to fr (open full item for complete abstract)

Committee: Dr. Dharma Agrawal (Advisor) Subjects:

MS, University of Cincinnati, 2005, Engineering : Computer Engineering

Next generation wireless and mobile cellular networks are expected to support multimedia applications. High user mobility and inherent instability of wireless channels make quality of service (QoS) provisioning a challenging issue in wireless and mobile networks. In this thesis, we propose a novel concept of time adaptability and introduce two adaptive time bandwidth allocation schemes, namely, adaptive time period (ATP) and flexible adaptive time period (FATP) for QoS provisioning in wireless and mobile cellular networks for variable bandwidth multimedia calls. With time adaptability, each call not only specifies the minimum and maximum bandwidth requirements in terms of basic bandwidth units (bbu), but also specifies a step size and an adaptive period. Step size is the bbu constituting each layer of adaptive multimedia. Adaptive period is the time period specified by a call over which the bandwidth change incurred by the call cannot exceed the step size for that call. This, accompanied by temporary bandwidth borrowing from the ongoing calls not only reduces forced termination probability of handoff calls and ensures that each call gets a fare share of the available system bandwidth but also helps to decrease large bandwidth fluctuations at the end mobile users running multimedia applications such as video conferencing or online gaming. Extensive simulation results show that our proposed schemes maintain a low forced termination probability for handoff calls and significantly reduce bandwidth variations.

Committee: Dr. Qing-An Zeng (Advisor) Subjects: Computer Science

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

Wireless sensor networks are being built to facilitate automated information gathering in military, industrial, environmental and surveillance applications. Many such applications of Sensor Networks require desired/pre-fixed QoS (Quality of Service: packetdata delivery within a defined deadline) guarantees and as well as high reliability. These applications demand high packet delivery ratio and are extremely delay sensitive and also demand reliability. However, certain factors limit the ability of the multihop sensor network to achieve the desired goals. These factors include the delay caused by network congestion, or hot regions in the network, limited energy of the sensor networknodes, packet loss due to the collisions and link failure andand mobilitylink failure. In this work, Our paper suggests an energy aware dual-path routing scheme for real timedeadline driven traffic is proposed, which balances node energy utilization, which is energy awareincreases the network lifetime, reducess the routing delay across the network by taking network congestion into account and increases the reliability of the packets reaching the destination by introducing minimal data redundancy. We als This paper also o introduces an adaptive prioritized Medium Access Layer (MAC) specifically to provide a differentiated service model for real time packets. Our intuitive claims areis well supported by simulationby simulation results.

Committee: Dr. Dharma P. Agrawal (Advisor) Subjects: Computer Science

MS, University of Cincinnati, 2003, Engineering : Electrical Engineering

Bluetooth works on a Master/Slave model supporting a 1Mbps nominal bandwidth. While this communication model is simple, it incurs longer delay due to sub-optimal packet forwarding and dual allocation of resources for slave-to-slave communication via the master. It also cannot serve demands exceeding this 1Mbps capacity. Group communication can only be supported by inefficient methods like multiple unicasts or a piconet-wide broadcast. We overcome these limitations by using Traffic Engineering concepts, providing QoS, admission control, maximal channel utilization and multi-device conversation with minimal interference. To overcome these problems we introduce four novel proactive techniques- Pseudo-Role-Switching(PRS), Dynamic-Slot-Allocation(DSA), Multi-Device Conversation(MDC) and Pseudo-PaRtitioning(PPR). We first achieve the maximum possible throughput using PRS and DSA and then dynamically partition the piconets using PPR. DSA also supports multicast-like communication. Through extensive simulations, we observe that by combining PRS or DSA with PPR, Bluetooth's throughput, delay and scalability is enhanced while interference and overhead is minimized.

Committee: Dr. Dharma Agrawal (Advisor) Subjects:

Master of Science, University of Toledo, 2010, Electrical Engineering

In this thesis the Quality of Service parameters for transmission of multimedia applications like voice, video and data are evaluated on ATM networks using various On-board switching algorithms and Queueing techniques. Due to the increasing interest in multimedia traffic it has become essential to consider satellite on-board switching capabilities for managing the multibeam input and output. Multimedia application has different requirements in terms of delay, jitter, and bandwidth and packet loss [1]. The network through which the data is transmitted and received has a large number of nodes, and the nodes have high variability and can cause blocking so blocking probability is another important parameter. Considering the expansion of internet there is demand for development of new protocols, architectures and network systems at the terrestrial level and also on the on-board processor keeping in mind the complexity level and cost factor. To improve these parameters different techniques like Differentiated services and On-board switching algorithms like Dijkstra's algorithm, Bellman-Ford algorithm and K-shortest path algorithm have been implemented on ATM networks like Banyan network, Benes network and Batcher-Banyan network using a single server M/G/1 queue. Simulation for MF-TDMA uplink and TDM downlink and evaluation of delay in algorithms, Queueing delay, Burstiness factor and Blocking probability has been performed. The results show that for small and medium size networks Bellman-Ford algorithm is effective in terms of delay whereas for large networks K-shortest path algorithm is most effective. To achieve low blocking probability and burstiness factor the Batcher-Banyan ATM network for single server M/G/1 queue is most efficient.

Committee: Jung Hwan Kim PhD (Committee Chair); Mohsin Jamali PhD (Committee Member); Ezzatollah Salari PhD (Committee Member) Subjects: Electrical Engineering

Master of Science, The Ohio State University, 2011, Computer Science and Engineering

Lately, there is a growing interest in the use of cloud computing for scientific applications, including scientific workflows. Key attractions of cloud include the pay-as-you-go model and elasticity. While the elasticity offered by the clouds can be beneficial for many applications and use-scenarios, it also imposes significant challenges in the development of applications or services. For example, no general framework exists that can enable a scientific workflow to execute in a dynamic fashion with QOS (Quality of Service) support, i.e. exploiting elasticity of clouds and automatically allocating and de-allocating resources to meet time and/or cost constraints while providing the desired quality of results the user needs. This thesis presents a case-study in creating a dynamic cloud workflow implementation with QOS of a scientific application. We work with MassMatrix, an application which searches proteins and peptides from tandem mass spectrometry data. In order to use cloud resources, we first parallelize the search method used in this algorithm. Next, we create a flexible workflow using the Pegasus Workflow Management System from ISI. We then add a new dynamic resource allocation module, which can use fewer or a larger number of resources based on a time constraint specified by the user. Finally we extend this to include the QOS support to provide the user with the desired quality of results. We use the desired quality metric to calculate the values of the application parameters. The desired quality metric refers to the parameters that are computed to maximize the user specified benefit function while meeting the time constraint. We evaluate our implementation using several different data-sets, and show that the application scales quite well. Our implementation effectively allocates resources adaptively and the parameter prediction scheme is successful in choosing parameters that help meet the time constraint.

Committee: Gagan Agrawal PhD (Advisor); Rajiv Ramnath PhD (Committee Member); Michael Freitas PhD (Committee Member) Subjects: Bioinformatics; Biomedical Engineering; Biomedical Research; Computer Engineering; Computer Science

Doctor of Philosophy, The Ohio State University, 2008, Computer Science and Engineering

Considerable research has focused on the problem of scheduling dynamically arriving independent parallel jobs on a given set of resources to improve the performance with respect to various system and user metrics. However, there has been little work on provision of Quality of Service (QoS) in space-shared parallel job scheduling, in the form of hard deadline guarantees and service differentiation. Both of these functionalities offered by system providers are very desirable to the user. On the other hand, revenue maximization along with the optimal management of resources is appealing to a service provider. This dissertation addresses these seemingly orthogonal aspects of parallel job scheduling in stages. At first a new scheme called QoPS is developed, to provide QoS in the form of response time guarantees. Essentially, QoPS implements an admission control mechanism for jobs, and provides deadline guarantees for all accepted jobs. Secondly, a pioneer model is proposed to enable proportional service differentiation (PSD) in job scheduling. A PSD framework would basically allow proportional allocation of resources across users based on relative priorities. In order to address the revenue issue, two different charging models are investigated, determined by the resource provider and user respectively. Since no QoS-enabled charging model is currently deployed at any supercomputer center, a new provider-determined charging model is proposed. In this context, the impact of user tolerance towards missed deadlines is studied, as well as various techniques to further improve the overall revenue. Alternatively, a user-centric and market-based revenue approach originally proposed for non-QoS scheduling is adapted for QoS-aware scheduling. Using this charging model, an extension to QoPS called DVQoPS is being developed, that considers the opportunity cost using a history-based predictive technique and thus maximizes the overall revenue while maintaining the deadline guarantees (open full item for complete abstract)

Committee: P. Sadayappan (Advisor); Dhabaleswar Panda (Committee Member); Rountev Atanas (Committee Member) Subjects: Computer Science