Master of Science / Department of Computer Science / Cornelia Caragea / Doina Caragea / Social media has changed the way people communicate information. It has been noted that social media platforms like Twitter are increasingly being used by people and authorities in the wake of natural disasters. The year 2017 was a historic year for the USA in terms of natural calamities and associated costs. According to NOAA (National Oceanic and Atmospheric Administration), during 2017, USA experienced 16 separate billion-dollar disaster events, including three tropical cyclones, eight severe storms, two inland floods, a crop freeze, drought, and wild re. During natural disasters, due to the collapse of infrastructure and telecommunication, often it is hard to reach out to people in need or to determine what areas are affected. In such situations, Twitter can be a lifesaving tool for local government and search and rescue agencies. Using Twitter streaming API service, disaster-related tweets can be collected and analyzed in real-time. Although tweets received from Twitter can be sparse, noisy and ambiguous, some may contain useful information with respect to situational awareness. For example, some tweets express emotions, such as grief, anguish, or call for help, other tweets provide information specific to a region, place or person, while others simply help spread information from news or environmental agencies. To extract information useful for disaster response teams from tweets, disaster tweets need to be cleaned and classified into various categories. Topic modeling can help identify topics from the collection of such disaster tweets. Subsequently, a topic (or a set of topics) will be associated with a tweet. Thus, in this report, we will use Latent Dirichlet Allocation (LDA) to accomplish topic modeling for disaster tweets dataset.

A multi-agent system consists of a number of agents, which may include software agents, robots, or even humans, in some application environment. Multi-robot systems are increasingly being employed to complete jobs and missions in various fields including search and rescue, space and underwater exploration, support in healthcare facilities, surveillance and target tracking, product manufacturing, pick-up and delivery, and logistics. Multi-agent task allocation is a complex problem compounded by various constraints such as deadlines, agent capabilities, and communication delays. In high-stake real-time environments, such as rescue missions, it is difficult to predict in advance what the requirements of the mission will be, what resources will be available, and how to optimally employ such resources. Yet, a fast response and speedy execution are critical to the outcome. This thesis proposes distributed optimisation techniques to tackle the following questions: how to maximise the number of assigned tasks in time restricted environments with limited resources; how to reach consensus on an execution plan across many agents, within a reasonable time-frame; and how to maintain robustness and optimality when factors change, e.g. the number of agents changes. Three novel approaches are proposed to address each of these questions. A novel algorithm is proposed to reassign tasks and free resources that allow the completion of more tasks. The introduction of a rank-based system for conflict resolution is shown to reduce the time for the agents to reach consensus while maintaining equal number of allocations. Finally, this thesis proposes an adaptive data-driven algorithm to learn optimal strategies from experience in different scenarios, and to enable individual agents to adapt their strategy during execution. A simulated rescue scenario is used to demonstrate the performance of the proposed methods compared with existing baseline methods.

She, Yingni. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2009. / Includes bibliographical references (leaves 74-78). / Abstracts in English and Chinese. / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Scope --- p.2 / Chapter 1.2 --- Motivation --- p.5 / Chapter 1.3 --- Structure of the Thesis --- p.6 / Chapter 2 --- Literature Review --- p.7 / Chapter 2.1 --- Intelligent Agents and Multiagent Systems --- p.8 / Chapter 2.1.1 --- Intelligent Agents --- p.8 / Chapter 2.1.2 --- Multiagent Systems --- p.10 / Chapter 2.2 --- Minority Game --- p.13 / Chapter 2.2.1 --- Minority Game --- p.13 / Chapter 2.2.2 --- Characteristics of Minority Game --- p.14 / Chapter 2.2.3 --- Strategies for Agents in Minority Game --- p.18 / Chapter 2.3 --- Resource Allocation --- p.22 / Chapter 2.3.1 --- Strategies for Agents in Multiagent Resource Allocation --- p.23 / Chapter 3 --- Individual Agent´ةs Wealth in Minority Game --- p.26 / Chapter 3.1 --- The Model --- p.26 / Chapter 3.2 --- Motivation --- p.27 / Chapter 3.3 --- Inefficiency Information --- p.28 / Chapter 3.4 --- An Intelligent Strategy --- p.31 / Chapter 3.5 --- Experiment Analysis --- p.32 / Chapter 3.6 --- Discussions and Analysis --- p.35 / Chapter 3.6.1 --- Equivalence to the Experience method --- p.36 / Chapter 3.6.2 --- Impact of M' and S' --- p.38 / Chapter 3.6.3 --- Impact of M and S --- p.41 / Chapter 3.6.4 --- Impact of Larger Number of Privileged Agents --- p.48 / Chapter 3.6.5 --- Comparisons with Related Work --- p.48 / Chapter 4 --- An Adaptive Strategy for Resource Allocation --- p.53 / Chapter 4.1 --- Problem Specification --- p.53 / Chapter 4.2 --- An Adaptive Strategy --- p.55 / Chapter 4.3 --- Remarks of the Adaptive Strategy --- p.57 / Chapter 4.4 --- Experiment Analysis --- p.58 / Chapter 4.4.1 --- Simulations --- p.58 / Chapter 4.4.2 --- Comparisons with Related Work --- p.62 / Chapter 5 --- Conclusions and Future Work --- p.69 / Chapter 5.1 --- Conclusions --- p.69 / Chapter 5.2 --- Future Work --- p.71 / A List of Publications --- p.73 / Bibliography --- p.74

Li, Kin Fai. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2008. / Includes bibliographical references (leaves 95-99). / Abstracts in English and Chinese. / Abstract --- p.i / Acknowledgement --- p.iv / Chapter 1 --- Introduction / Motivation --- p.1 / Chapter 2 --- Literature Review --- p.7 / Chapter 2.1 --- Introduction --- p.7 / Chapter 2.2 --- Dynamic Spectrum Access Networks --- p.8 / Chapter 2.3 --- Reliability --- p.10 / Chapter 2.3.1 --- Reliability in Wireless Networks --- p.10 / Chapter 2.3.2 --- Reliability in Wireline Networks --- p.11 / Chapter 2.4 --- Capacity Planning in Wireless Mesh Networks --- p.14 / Chapter 2.4.1 --- Interference Model --- p.14 / Chapter 2.4.2 --- Link Capacity Constraint --- p.15 / Chapter 2.4.3 --- Feasible Path --- p.16 / Chapter 2.4.4 --- Optimal Capacity Allocation in DSA Net- works and Wireless Mesh Networks --- p.16 / Chapter 2.5 --- Chapter Summary --- p.18 / Chapter 3 --- Lifetime Aware Routing without Backup --- p.19 / Chapter 3.1 --- Introduction --- p.19 / Chapter 3.2 --- System Model --- p.20 / Chapter 3.3 --- Lifetime Distribution of a Path without Backup Protection --- p.22 / Chapter 3.3.1 --- Exact Lifetime Distribution --- p.23 / Chapter 3.3.2 --- The Chain Approximation --- p.24 / Chapter 3.4 --- Route Selection without Backup Protection --- p.26 / Chapter 3.4.1 --- NP-Hardness of Finding Maximum Lifetime Path --- p.26 / Chapter 3.4.2 --- The Minimum Weight Algorithm --- p.28 / Chapter 3.4.3 --- Greedy Algorithm --- p.28 / Chapter 3.4.4 --- GACA - The Greedy Algorithm using the Chain Approximation --- p.32 / Chapter 3.5 --- Simulation Results --- p.33 / Chapter 3.5.1 --- Tightness of the Chain Approximation Bound for Vulnerable Area --- p.33 / Chapter 3.5.2 --- Comparison between Greedy and GACA using Guaranteed Lifetime --- p.36 / Chapter 3.5.3 --- Factors impacting the performance of GACA --- p.37 / Chapter 3.6 --- Chapter Summary --- p.43 / Chapter 4 --- Prolonging Path Lifetime with Backup Channel --- p.44 / Chapter 4.1 --- Introduction --- p.44 / Chapter 4.2 --- Non-Shared Backup Protection --- p.45 / Chapter 4.2.1 --- Lifetime of a Path with Non-Shared Backup --- p.45 / Chapter 4.2.2 --- Route Selection for paths with Non-Shared Backup --- p.46 / Chapter 4.3 --- Shared Backup Protection --- p.47 / Chapter 4.3.1 --- Sharing of Backup Capacity --- p.48 / Chapter 4.3.2 --- Lifetime of a Path with Shared Backup --- p.48 / Chapter 4.3.3 --- Route Selection for paths with Shared Backup --- p.50 / Chapter 4.4 --- Simulation Results --- p.50 / Chapter 4.4.1 --- Tightness of Failure Probability Upper Bound for Backup Protection --- p.51 / Chapter 4.4.2 --- Comparison between the Shared Backup and Non Shared Backup schemes --- p.53 / Chapter 4.5 --- Chapter Summary --- p.54 / Chapter 5 --- Finding Capacity-Feasible Routes --- p.55 / Chapter 5.1 --- Introduction --- p.55 / Chapter 5.2 --- Constructing an Edge graph --- p.56 / Chapter 5.3 --- Checking Capacity Feasibility under each Protec- tion Scheme --- p.58 / Chapter 5.3.1 --- No Backup Protection --- p.59 / Chapter 5.3.2 --- Non-Shared Backup Protection --- p.59 / Chapter 5.3.3 --- Shared Backup Protection --- p.60 / Chapter 5.4 --- Chapter Summary --- p.62 / Chapter 6 --- Performance Evaluations and Adaptive Protec- tion --- p.63 / Chapter 6.1 --- Introduction --- p.63 / Chapter 6.2 --- Tradeoffs between Route Selection Algorithms --- p.64 / Chapter 6.3 --- Adaptive Protection --- p.66 / Chapter 6.3.1 --- Route Selection for Adaptive Protection --- p.67 / Chapter 6.3.2 --- Finding a Capacity-Feasible Path for Adaptive Protection --- p.68 / Chapter 6.4 --- Comparison between No Protection and Adaptive Protection --- p.69 / Chapter 6.5 --- Chapter Summary --- p.71 / Chapter 7 --- Restoration Capacity Planning and Channel Assignment --- p.72 / Chapter 7.1 --- Introduction --- p.72 / Chapter 7.2 --- System Model --- p.74 / Chapter 7.2.1 --- Channel Assignment Model --- p.74 / Chapter 7.2.2 --- Presence of Primary Users --- p.75 / Chapter 7.2.3 --- Link Flow Rates --- p.76 / Chapter 7.2.4 --- Problem Formulation --- p.77 / Chapter 7.3 --- Simulation Results --- p.79 / Chapter 7.3.1 --- "Comparison between ""Shared Backup"" and “No Restore Plan"" using Guarantee Percentage and Reduced Capacity" --- p.80 / Chapter 7.3.2 --- Comparison using Traffic Demand Scaling Factor g and Guarantee Fraction p --- p.81 / Chapter 7.3.3 --- Comparison between Optimal Channel Assignment and Random Channel Assignment --- p.84 / Chapter 7.4 --- Chapter Summary --- p.86 / Chapter 8 --- Conclusion and Future Works --- p.87 / Chapter A --- Proof of Theorem (3.1) in Chapter3 --- p.90 / Chapter B --- Proof of Theorem (4.1) in Chapter4 --- p.92 / Bibliography --- p.95

Thesis (M. P. A.)--University of Hong Kong, 2007. / Title proper from title frame. Also available in printed format.

<p>In this thesis we investigate different algorithms for dynamic resource allocation in wireless networks. We introduce a general framework for modeling systems whichis applicable to many scenarios. We also analyze a specific scenario with adaptivebeamforming and show how it fits into the proposed framework. We then studytwo different resource allocation problems: Quality-of-Service (QoS) constraineduser scheduling and sum-rate maximization. For user scheduling, we select some“good” set of users that is allowed to use a specific resource. We investigatedifferent algorithms with varying complexities. For the sum-rate maximizationwe find the global optimum through an algorithm that takes advantage of thestructure of the problem by reformulating it as a D.C. program, i.e., a minimizationover a difference of convex functions. We validate this approach by showing that itis more efficient than an exhaustive search at exploring the space of solutions. Thealgorithm provides a good benchmark for more suboptimal algorithms to comparewith. The framework in which we construct the algorithm, apart from being verygeneral, is also very flexible and can be used to implement other low complexitybut suboptimal algorithms.</p>

This thesis consists of two studies on joint bandwidth and power allocation strategy for wireless communication networks. In the first study, joint bandwidth and power allocation strategy is proposed for wireless multi-user networks without relaying and with decode-and-forward relaying. It is shown that the formulated resource allocation problems are convex and, thus, can be solved efficiently. Admission control problem based on the joint bandwidth and power allocation strategy is further considered, and a greedy search algorithm is developed for solving it efficiently. In the second study, joint bandwidth and power allocation strategy is presented for maximizing the sum ergodic capacity of secondary users under fading channels in cognitive radio networks. Optimal bandwidth allocation is derived in closed-form for any given power allocation. Then the structures of optimal power allocations are derived. Using these structures, efficient algorithms are developed for finding the optimal power allocations. / Communications

Multichannel marketing is the practice of simultaneously offering information, goods, services, and support to customers through two or more synchronized channels. In this dissertation, I develop an integrated framework of multichannel marketing and develop models to assist managers in their marketing resource allocation decisions. In the first essay of the dissertation, I investigate the factors that drive customers multichannel shopping behavior and identify its consequences for retailers. In the second essay, I build on this work and develop a model that enables firms to optimize their allocation of marketing resources across different customer-channel segments. In the first essay, I develop a framework comprising the factors that drive consumers’ channel choice, the consequences of channel choice, and their implications for managing channel equity. The results show that customer-channel choice is driven in a nonlinear fashion by a customer demographic variable such as age and is also influenced by consumer shopping traits such as number of categories bought and the duration of relationship with a retailer. I show that by controlling for the moderating effects of channel-category associations, the influence of customers’ demographics and shopping traits on their channel choices can vary significantly across product categories. Importantly, the results show that multichannel shoppers buy more often, buy more items, and spend considerably more than single channel shoppers. The channel equity of multichannel customers is nearly twice that of the closest single channel customers (online or offline). In the second essay, I propose a model for optimal allocation of marketing efforts across multiple customer-channel segments. I first develop a set of models for consumer response to marketing efforts for each channel-customer segment. This set comprises four models, the first for purchase frequency, the second for purchase quantity, the third for product return behavior, and the fourth for contribution margin of purchase. The results show that customers’ responses to firm marketing efforts vary significantly across the customer-channel segments. They also suggest that marketing efforts influence purchase frequency, purchase quantity and monetary value in different ways. The resource allocation results show that profits can be substantially improved by reallocating marketing efforts across the different customer-channel segments.

Since semiconductor process is constantly being improved, the complexity of system-on-chip is rising daily and we can place more and more elements on the same chip area. The system designers have been searching new methodology that can handle the complex systems and the environment which can quickly simulate the system-on-chip. It is brought forward that is raising the level of abstraction, as the design methodology of Electronic-System-Level (ESL). But system designers still need to decide the system architecture (the bus and PE connection status), and judge if the system could meet the performance and cost constraints by simulation results. For the very complex system, system designers will cost more and more time owning to the growth of design space to get the best system architecture. In this thesis, we propose a synthesis method to support automatic ESL design and help system designers to decide system architecture from large design space in short time. The method uses fast estimation method to estimate bus topology and memory allocation that affect the processing-elements¡¦ communication. By this method, we can find better system architecture which meets all constraints with the same amount of processing-elements.

The cellular industry is undergoing a major paradigm shift from voice-centric, structured homogeneous networks to a more data-driven, distributed and heterogeneous architecture. One of the more promising trends emerging from this cellular revolution is femtocells. Femtocells are primarily viewed as a cost-effective way to improve both capacity and indoor coverage, and they enable offloading data-traffic from macrocell network. However, efficient interference management in co-channel deployment of femtocells remains a challenge. Decentralized strategies such as femtocell access control have been identified as an effective means to mitigate cross-tier interference in two-tier networks. Femtocells can be configured to be either open access or closed access. Prior work on access control schemes show that, in the absence of any coordination between the two tiers in terms of power control and user scheduling, closed access is the preferred approach at high user densities. Present methods suggest that in the case of orthogonal multiple access schemes like TDMA/OFDMA, femtocell access control should be adaptive according to the estimated cellular user density. The approach we follow, in this work, is to adopt an open access policy at the femtocell access points with a cap on the maximum number of users allowed on a femtocell. This ensures the femto owner retains a significant portion of the femtocell resources. We design an iterative algorithm for hybrid access control for femtocells that integrates the problems of uplink power control and base station assignment. This algorithm implicitly adapts the femtocell access method to the current user density. The distributed power control algorithm, which is based on Yates' work on standard interference functions, enables users to overcome the interference in the system and satisfy their minimum QoS requirements. The optimal allocation of femtocell resources is incorporated into the access control algorithm through a constrained sum-rate maximization to protect the femto owner from starvation at high user densities. The performance of a two-tier OFDMA femtocell network is then evaluated under the proposed access scheme from a home owner viewpoint, and network operator perspective. System-level simulations show that the proposed access control method can provide a rate gain of nearly 52% for cellular users, compared to closed access, at high user densities and under moderate-to-dense deployment of femtocells. At the same time, the femto owner is prevented from going into outage and only experiences a negligible rate loss. The results obtained establish the quantitative performance advantage of using hybrid access at femtocells with power control at high user densities. The convergence properties of the proposed iterative hybrid access control algorithm are also investigated by varying the user density and the mean number of femto access points in the network. It is shown that for a given system model, the algorithm converges quickly within thirty iterations, provided a feasible solution exists.