Master of Science, The Ohio State University, 2022, Industrial and Systems Engineering

Decades of cognitive systems engineering research has revealed that implementing human-machine teams into complex environments can consequently result in challenges that negatively impact human-machine teams. Such challenges and conflicts amongst team members can readily be observed in human-machine teams where agents are assigned heterogeneous tasks because the agents' individual goals may have a tendency to conflict and compete with one another in their shared environment. This conflict may also be magnified if the agents of our heterogeneously tasked human-machine team do not share a common goal and are not equipped with the resources to manage their differences. In our study, we set out to determine how the performance of our heterogeneously tasked agents in our simulated human-machine team was impacted in our full-motion video and intelligence analysis. By using joint-performance activity graphs, various statistical analyses, constant comparative analysis, and human-machine teaming heuristic analysis, we were able to determine that the performance of our human-machine team was not significantly different from the performance of our participants who worked alone. This led us to the conclusion that the machine agent insufficiently aided their human agent's decision making during the full motion video analysis and the design of the machine failed to adhere to known Human-Machine Teaming heuristics. Lastly, this holistic analysis revealed that the machine agent acted as if it did not have any knowledge of the ultimate goal of their human agent, and due to its limited capabilities, the machine was unable to contribute information in relation to the overarching goal. Even though the architecture of the human-machine team in this study failed to adhere to various human-machine teaming heuristics, failing to adhere to and implement the team so that both the agents' individual tasks meaningfully contributed the shared goal was determined to be the most criti (open full item for complete abstract)

Committee: Michael Rayo (Advisor); Samantha Krening (Committee Member); Michael Rayo (Committee Member) Subjects: Design; Engineering; Industrial Engineering; Systems Science

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

This thesis describes the concept of sensor networks which has been made viable by the convergence of MEMS system technology and efficient routing protocols. Sensor nodes possess finite, non-renewable energy that they expend in sensing a multitude of modalities including temperature, moisture, pressure, light and infrared radiation. A radio-interconnected collection of such sensors forms a sensor network and the information collected from the network is transmitted for analysis at a distant location termed as the sink. The main purpose of a sensor network is to gather information about the various parameters of the area in which it is deployed and to transmit this information to the sink for appropriate utilization. A wireless sensor node is capable of only a limited amount of communication and processing. Therefore, unlike traditional networks, where the objective is to maximize channel throughput, the chief consideration in a sensor network is to extend the system lifetime as well as system robustness. Wireless ad hoc and sensor networks are comprised of energy–constrained nodes. This limitation has led to the dire need for energy-aware protocols to produce an efficient network. Heterogeneity is introduced in a wireless sensor network by having a large number of low power sensor nodes and a small number of more powerful nodes to serve as cluster heads. We propose a self-tuning scheme that improves the lifetime of a heterogeneous wireless sensor network by appropriately scheduling the transmission rate of individual sensor nodes in the network. We consider a distance based sleep scheduling problem for equal energy consumption rates in low power sensor nodes and evaluate the optimal settings required in a heterogeneous sensor network. We evaluate the efficiency of our proposed algorithm based on an analytical model and perform simulations to verify the adequacy of our scheme in terms of important network parameters and compare with existing heterogeneous sensor netw (open full item for complete abstract)

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

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

The problem of real-time scheduling has been very well researched for the single processor case and the identical multiprocessor case. There are a number of efficient algorithms and schedulability tests that enable the system designer to model his/her application as a set of real-time tasks. However, it may very well be the case that the multiprocessor system consists of processors with different speeds and configurations. This heterogeneity makes the scheduling problem much more complex. Real-time scheduling on heterogeneous multiprocessors has not received much attention in the scheduling literature. We propose several algorithms for scheduling a set of precedence related tasks on heterogeneous multiprocessors. In the first section, we propose a reliability driven algorithm for scheduling periodic tasks on heterogeneous systems. We relax the assumption that all the processors in the system are equally reliable. We introduce a new metric called cost of dependability. The periodic tasks are assigned to processors in such a way that the reliability of the system is maximized. We observe that our scheme generate schedules that are consistently more reliable than the schedules generated by algorithms that do not take the reliability of the processors into account. In the second section, we propose a duplication based algorithm for scheduling a set of soft deadline tasks on a system of heterogeneous multiprocessors. We observe that communicating tasks assigned to different processors have to incur a delay by using the inter-processor communication channel. By duplicating any of these two tasks on either assigned processor, we can cut down on that delay. Hence tasks can start (and hence finish) earlier. This leads to a larger number of tasks meeting their deadlines and hence an increase in the guarantee ratio of the real-time application. The next section extends the concept of task duplication to a set of hard deadline tasks. The deadlines are hard in that even a single (open full item for complete abstract)

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

Doctor of Philosophy, The Ohio State University, 2011, Electrical and Computer Engineering

This dissertation presents research into the development of high performance dataflow middleware and applications on heterogeneous, distributed-memory supercomputers. We present coarse-grained state-of-the-art ad-hoc techniques for optimizing the performance of real-world, data-intensive applications in biomedical image analysis and radar signal analysis on clusters of computational nodes equipped with multi-core microprocessors and accelerator processors, such as the Cell Broadband Engine and graphics processing units. Studying the performance of these applications gives valuable insights into the relevant parameters to tune for achieving efficiency, because being large-scale, data-intensive scientific applications, they are representative of what researchers in these fields will need to conduct innovative science. Our approaches shows that multi-core processors and accelerators can be used cooperatively to achieve application performance which may be many orders of magnitude above naive reference implementations. Additionally, a fine-grained programming framework and runtime system for the development of dataflow applications for accelerator processors such as the Cell is presented, along with an experimental study showing our framework leverages all of the peak performance associated with such architectures, at a fraction of the cognitive cost to developers. Then, we present an adaptive technique for automating the coarse-grained ad-hoc optimizations we developed for tuning the decomposition of application data and tasks for parallel execution on distributed, heterogeneous processors. We show that our technique is able to achieve high performance, while significantly reducing the burden placed on the developer to manually tune the relevant parameters of distributed dataflow applications. We evaluate the performance of our technique on three real-world applications, and show that it performs favorably compared to three state-of-the-art distributed programmi (open full item for complete abstract)

Committee: Umit Catalyurek PhD (Advisor); Fusun Ozguner PhD (Committee Member); Charles Klein PhD (Committee Member) Subjects: Computer Engineering; Computer Science

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

This dissertation presents several methods to more efficiently use the computational resources availableon a Heterogeneous Chip Multiprocessor (H-CMP). Using task scheduling techniques, three challenges to the effective usage of H-CMPs are addressed: the emergence of reconfigurable hardware in general purpose computing, utilization of the network on a chip (NoC), and fault tolerance. To utilize reconfigurable hardware, we introduce the Mutually Exclusive Processor Groups reconfiguration model, and an accompanying task scheduler, the Heterogeneous Earliest Finish Time with Mutually Exclusive Processor Groups (HEFT-MEG) scheduling heuristic. HEFT-MEG schedules reconfigurations using a novel back-tracking algorithm to evaluate how different reconfiguration decisions affect previously scheduled tasks. In both simulation and real execution, HEFT-MEG successfully schedules reconfiguration allowing the architecture to adapt to changing application requirements. After an analysis of IBM's Cell Processor NoC and generation of a simple stochastic model, we propose a hybrid task scheduling system using a Compile- and Run-time Scheduler (CtS and RtS) that work in concert. The CtS, Contention Aware HEFT (CA-HEFT), updates task start and finish times when scheduling to account for network contention. The RtS, the Contention Aware Dynamic Scheduler (CADS), adjusts the schedule generated by CA-HEFT to account for variation in the communication pattern and actual task finish times, using a novel dynamic block algorithm. We find that using a CtS and RtS in concert improves the performance of several application types in real execution on the Cell processor. To enhance fault tolerance, we modify the previously proposed hybrid scheduling system to accommodate variability in the processor availability. The RtS is divided into two portions, the Fault Tolerant Re-Mapper (FTRM) and the Reconfiguration and Recovery Scheduler (RRS). FTRM examines the current processor availability and remap (open full item for complete abstract)

Committee: Fusun Ozguner (Advisor); Umit Catalyurek (Committee Member); Eylem Ekici (Committee Member) Subjects: Computer Science; Electrical Engineering

Doctor of Philosophy, Case Western Reserve University, 1993, Computer Engineering

It is increasingly the case that organizations rely upon diverse computer systems to perform a variety of knowledge-based tasks. This not only presents technical issues of interoperability and integration, but philosophical issues of how cooperation and interaction between computational entities is to be realized. Cooperating systems are systems that work together toward a common end. To develop cooperating heterogeneous systems the technical and philosophical must find common ground. The concepts of cooperation must find realization in technically sound system architectures. The heterogeneity and complexity of knowledge sources make it difficult to design and use an integrated system. This is particularly true of systems that are constructed of pre-existing components, each of which has its own task, design, and structure. We advocate adding a uniform meta-layer between knowledge sources and the rest of the system. The layer consists of a family of interpreters, one for each knowledge source, and meta-knowledge. A system architecture to integrate and control diverse knowledge sources is presented. The architecture is based on the meta-level properties of the logic programming language Prolog. We describe an implementation of the architecture, a Framework for Logic Programming Systems with D istributed Execution (FLiPSiDE). Based on the vanilla blackboard interpreter, we present interpreter enhancements that model the cooperative behavior of knowledge sources. Examples of interpreters to handle shared semantics and multiple interaction behaviors are shown. We introduce a dedicated trigger handler knowledge source to the design of control blackboard architectures. The advantages of such a knowledge source, including dynamic modification of triggers, are described. The support of human users is presented as an example of integrating heterogeneous users with the computational knowledge sources. The FLiPSiDE architecture is generalized to create a distributed Prolog env (open full item for complete abstract)

Committee: Leon Sterling (Advisor) Subjects: