Sharif Digital Repository

In this thesis, we study optimal scheduling in a multi-hop wirelessly-powered network. We propose an online control policy for energy transfer from the energy access point to the wireless nodes and for data transfer among the nodes. With our proposed control policy, all data queues of the nodes are stable, while the average energy consumption of the network is shown to be within a bounded gap of the minimum energy required for stabilizing the network. Our proposed policy is designed using a quadratic Lyapunov function to capture the limitations on the energy consumption of the nodes imposed by their bachery levels. We show that under the proposed control policy, the bachlog level in the data... 

Most of current the MAC protocols for Wireless Sensor Networks (WSNs) concern about energy efficiency rather than low latency. The main source of energy consumption in the MAC layer is idle listening, and most of these protocols use duty cycling for reducing energy consumption. But duty cycling method increases packet delivery latency. This problem is very critical in real time applications where packets should meet their deadline. There are some solutions for this problem. For example a node forwards data packets multi hop away instead of one hop in a single cycle. In this research, an efficient cross layer MAC protocol for real time applications is proposed. In this protocol multi hop... 

Intercultural Communicative Competence has been the focus of much academic inquiries in the field of second/foreign language learning in the last decades since learners’ ability to communicate authentically in the target language has been recognized among the scholars. This study aimed to address this issue in the Iranian academic environment where there was almost a gap in not touching the cultural issues of target language in a focused and organized manner in language classes at universities. For that purpose, an intercultural syllabus was designed for a 14-week course of English short story in the Minor Program at Sharif University of Technology. The syllabus was an integration of various... 

Complex networks have recently become a very noticeable subject because of their remarkable theory on real networks such as rumor spreading. This thesis contributes to a prevention strategy which uses a maximal clique algorithm to find overlapping communities and also uses inverse targeting immunization strategy to immune these communities. By the using maximal clique algorithm combined with inverse targeting strategy we have achieved that the TIK method (targeted immunization using k-clique percolation) obtains 10% of better immunization than HD (highest degree) method which is one of the best methods in targeted immunization strategies. The TIK also demonstrates that rumor spreading is... 

It is of fundamental importance to find algorithms obtaining optimal performance for learning of statistical models in distributed and communication limited systems. In this thesis, we aim at characterizing the best learning strategies over distributed datasets such that the communications between storing machines are minimized. We have addressed two problems in distributed setting: learning of Gaussian processes, and structure learning of Gaussian Graphical Models (GGM). The performance of the proposed methods are analyzed theoritically and verified experimentally. The experimental results show that with spending few bits the proposed distributed methods have close performance to the... 

Nowadays, the explosion of the volume data provides more accuracy in machine learning models. But, Working with a vast amount of data is not easy, especially in a situation that data are distributed over the systems. In such systems, designing distributed learning algorithms that in communication efficient setting demand reliable and more accurate results, are so important. We studied sparse structure learning of Gaussian graphical model in a situation that our data are distributed over the system and each machine has a dimension of data. Each local machine should send its data to a central machine and the central machine is responsible for learning the structure. For reliable learning under... 

Reinforcement learning is a learning method that uses reward and penalty feedbacks, having no information about the right action. In this method, agent gets the state of environment and selects an action among its permissible set of actions, regarding its policy and the given state. Environment, expresses an evaluation, in form of a reinforcement signal and a change in state, as a response for agent’s action. Afterward, the agent updates its policy considering received signal in order to maximize its long term reward. Reinforcement learning rapidly converges to the optimal solution, only if there are few states and actions, but there are lots of domains that consist of too many states and... 

Wireless Sensor Networks (WSNs) attracted lots of researchers in recent years, these sensors are small and have limited computational resources and are not expensive in comparision with conventional sensors. These sensors can sense, measure, gathering information from environment and send them to user, they are typically characterized by limited communication capabilities due to tight energy and bandwidth constraints. As a result, WSNs have inspired a resurgence in research on decentralized algorithms, structure of these networks are variable and dynamic, sensors may deleted from the network or added to the network, so using models which don’t require precies specifications of environment is... 

Distributed learning has been a popular area of machine learning for researchers according to having access to an unprecedented amount of data distributed over many clients in a network. Moreover, high dimensional learning, when the dimensions of data are high but the effective features are low, has great usage in learning especially in medical science. In this paper, we attempted to develop a method for learning high-dimensional sparse problems in a distributed manner, whit a star shape network. Our main focus in this research is on optimizing the communication flow in the network. The prominent idea of our proposed method is to extract the main feature in the first step, then continuing... 

Ising model is a Markov Random Field (MRF) with binary random variables which has a vast literature in both theoretical and practical sides. In this thesis, we investigate two important statistical problems on this model. Learning the structure of MRFs has a long history and had a significant progress in the recent years. The goal of this problem is to find the independence graph of MRF using the samples generated from it. Specifically, we focus on the structure learning of ising models. Important algorithms for finding the structures had been reviewed. Additionally, we introduced information-theoretical and computational limitations of this problem. The second problem is community detection... 

This thesis is concerned with an optimal trade-off between communication overhead and the number and size of neighborhoods in a distributed optimal control technique for large-scale systems that is based on the Jacobi iteration and two-layer architecture for communication. Although this method efficiently reduces computational overhead, continuous and proper data transfer between subsystems that are very often far from each other, is required to achieve an acceptable performance. However, limited transmission bandwidth and short communication range result in significant communication overhead. This causes significant time latency between measurement and applied calculated control actions,... 

The goal of this study is to investigate the effect of events such as earthquakes on businesses. Earthquakes can have a significant impact on the performance and profitability of businesses. Earthquakes can negatively affect the functionality of community infrastructure systems by damaging their components and this disruption in the functionality of the lifelines of the community can negatively affect the businesses that are dependent on their functionality. In addition, earthquakes will cause casualties and injuries, leading to a decrease in the workforce and the customers’ population. In such a situation, the activity of many businesses is disrupted and even stopped, which in turn has many... 

The purpose of this study is to present a suitable model for calculating the costs imposed on society due to deaths and injuries due to infrastructure failure due to earthquakes, which is used in the analysis of community resilience. Much research has been done to date to calculate the economic costs of infrastructure failure. However, no suitable analytical framework has been provided to quantify the costs of deaths and injuries to people due to earthquake failure. Previous research has provided methods such as the value of statistical life to quantify the social costs caused by accidents and environmental pollution. However, the literature review results in the field of seismic resilience... 

Nowadays Systems in different fields of research like computer science, biology, social networks, information networks, and economics are modeled as graphs. The graphs which model real world systems have very different topological characteristics than those of classic networks. One of the prominent characteristics of these networks, is that its not practical to describe a general model for their structure and behavior. As a consequence of this complexity in modeling and structure, these networks are called complex networks. One of the most important observations in complex networks is the presence of communities, it means that in such networks one can separate vertices in disjoint sets, such... 

Network methods have had profound influence in many domains and fields in the past decade. Community structure is a very important property of complex networks, but the precise definition of a community remains an unresolved problem and Community detection in complex networks has recently attracted wide attention in several research areas such as social networks, communication networks, and biological systems. Complex network studies grow by the scientific community, thanks to the increasing accessibility of real-world network data. The emphasis of use analysis on the network algorithms and community structure is one of the most important problems in the field of complex network. Many... 

In recent decades, a wide range of research has been done on the characteristics of networks in various domains. Community structure is one of the most important properties in networks. However, most community detection methods designed for unipartite networks. Bipartite networks are very important in many different fields because these networks explicitly shows conceptual connections between different types of entities, for example in bioinformatics science, business, social and economic systems. Recently, community detection in bipartite networks has risen attention by the technical community, due to the accessibility of huge bipartite network data from different domains and the spread of... 

Reliable and secure communication requires low error probability and low information leakage, respectively. But there are different metrics for error probability and information leakage. Two important reliability metrics are ϵ or zero probability of error. An ϵ-error criterion requires the (average or maximal) error probability to vanish as the blocklength increases, while a zero-error criterion, demands the error to be exactly zero for every given bloklength. Three important security metrics are weak, strong, or perfect secrecy. A weak notion of secrecy requires the percentage of the message that is leaked to vanish as the code blocklength increases, while a strong notion of secrecy... 

Nowadays sharing data between users and organizations can be difficult due to privacy and legal reasons. Therefore, real-world data is not fully exploited by machine learning methods. A novel model training method called Federated Learning was proposed to alleviate the aforementioned problem by enabling users to jointly train learning models without sharing data. Federated learning utilization faces many challenges among which high communication cost, statistical heterogeneity and system heterogeneity are most important. This research deals with these challenges by proposing two methods of Reduced Clustering Federated Learning (RCFL) and Weighted Federated Distillation (WFD). Moreover, we... 

Optical impairments, as a challenging obstacle in optical communication systems, have been under deep attention of optical engineers in recent years. These impairments can be divided into two categories of linear and nonlinear impairments. Dispersion is a significant linear impairment leads to broadening of the optical signal and interference between transmitted symbols during detection. Nonlinear impairments such as cross-phase modulation (XPM) and self-phase modulation (SPM) also distort the transmitted signal through the fiber, causing interference at the receiver and degrading the system's performance. An appropriate solution for compensating optical fiber impairments is to utilize... 

Today's interest growth to the underwater explorations and navigations demands to design appropriate underwater communication methods and systems is strongly felt. Although radio frequencies which are widely used in free space communications, they cannot be a decent candidate for underwater wireless communications, due to their sever attenuation in water environment. On the other hand, underwater acoustic communications has been widely studied and implemented in past decades, because of little attenuation in comparison with other means of communications in water environment. However, acoustic communications suffer from limited bandwidth, low propagation speed, and low security in compared...