During the past decade, the Internet has witnessed a dramatic increase in the popularity of Peer-to-Peer (P2P) applications. This has caused a significant growth in the volume of P2P traffic. This trend has been particularly alarming for the Internet Service Providers (ISPs) that need to cope with the associated cost but have limited control in routing or managing P2P traffic. To alleviate this problem, researchers have proposed mechanisms to reduce the volume of external P2P traffic for individual ISPs. However, prior studies have not examined the global effect of P2P applications on the entire network, namely the traffic that a P2P application imposes on individual underlying Autonomous Systems (ASs). Such a global view is particularly important because of the large number of geographically scattered peers in P2P applications. This dissertation examines the global effect of P2P applications on the underlying AS-level Internet. Toward this end, first we leverage a large number of complete overlay snapshots from a large-scale P2P application, namely Gnutella, to characterize the connectivity and evolution of its overlay structure. We also conduct a case study on the performance of BitTorrent and its correlation with peer- and group-level properties. Second, we present and evaluate Respondent-driven sampling as a promising technique to collect unbiased samples for characterizing peer properties in large-scale P2P overlays without requiring the overlay's complete snapshot. Third, we propose a new technique leveraging the geographical location of peers in an AS to determine its geographical footprint and identify the cities where its Points-of-Presence (PoPs) are likely to be located. Fourth, we present a new methodology to characterize the effect of a given P2P overlay on the underlying ASs. Our approach relies on the large scale simulation of BGP routing over the AS-level snapshots of the Internet to identify the imposed load on each transit AS. Using our methodology, we characterize the impact of Gnutella overlay on the AS-level underlay over a 4-year period. Our investigation provides valuable insights on the global impact of large scale P2P overlay on individual ASs. This dissertation includes my previously published and co-authored material.

With growing number of devices connected to the network, there is a greater need for use of Peer-to-Peer (P2P) networks and distributed P2P protocols.Devices participating in the P2P network do not usually need to use any central server that links up connections. It has many advantages but it needs to use so-called overlay network that consists of protocols used for traffic routing and decision making. Protocols used in today’s P2P networks are mostly not considerate of particular participating nodes and all the nodes in the network are usually equal. This can have negative impacts on network performance. In order to avoid or reduce some unwanted negative impacts, it would be advantageous if the overlay network could route traffic and make decisions according to context information like battery levels or connection speeds. This work reviews a few popular P2P overlay networks and based on that it introduces an improvement of one of them – Chord. The structure of the improved version of the Chord protocol called Context-Aware Chord is described. Then results of the evaluation are presented. With a use of the improved protocol, nodes can participate longer in the network and throughput of lookup messages is improved.

Serverless Edge Computing is an extension of the serverless computing paradigm that enables the deployment and execution of modular software functions on resource-constrained edge devices. However, it poses several challenges due to the edge network's dynamic nature and serverless applications' latency constraints. In this work, we introduce EdgeFn, a lightweight distributed data store for the serverless edge computing system. While serverless comput- ing platforms simplify the development and automated management of software functions, running serverless applications reliably on resource-constrained edge devices poses multiple challenges. These challenges include a lack of flexibility, minimum control over management policies, high data shipping, and cold start latencies. EdgeFn addresses these challenges by providing distributed data storage for serverless applications and allows users to define custom policies that affect the life cycle of serverless functions and their objects. First, we study the challenges of existing serverless systems to adapt to the edge environment. Sec- ond, we propose a distributed data store on top of a Distributed Hash Table (DHT) based Peer-to-Peer (P2P) Overlay, which achieves data locality by co-locating the function and its data. Third, we implement programmable callbacks for storage operations which users can leverage to define custom policies for their applications. We also define some use cases that can be built using the callbacks. Finally, we evaluate EdgeFn scalability and performance using industry-generated trace workload and real-world edge applications. / Master of Science / Serverless Edge Computing is an extension of the serverless computing paradigm that enables the deployment and execution of modular software functions on resource-constrained edge devices. However, it poses several challenges due to the edge network's dynamic nature and serverless applications' latency constraints. In this work, we introduce EdgeFn, a lightweight distributed data store for the serverless edge computing system. While serverless comput- ing platforms simplify the development and automated management of software functions, running serverless applications reliably on resource-constrained edge devices poses multiple challenges. These challenges include a lack of flexibility, minimum control over management policies, high data shipping, and cold start latencies. EdgeFn addresses these challenges by providing distributed data storage for serverless applications and allows users to define custom policies that affect the life cycle of serverless functions and their objects. First, we study the challenges of existing serverless systems to adapt to the edge environment. Sec- ond, we propose a distributed data store on top of a Distributed Hash Table (DHT) based Peer-to-Peer (P2P) Overlay, which achieves data locality by co-locating the function and its data. Third, we implement programmable callbacks for storage operations which users can leverage to define custom policies for their applications. We also define some use cases that can be built using the callbacks. Finally, we evaluate EdgeFn scalability and performance using industry-generated trace workload and real-world edge applications.

This thesis presents an innovative strategy for distributed Graph Neural Network (GNN) training, leveraging a peer-to-peer network of heterogeneous edge devices interconnected through a Distributed Hash Table (DHT). As GNNs become increasingly vital in analyzing graph-structured data across various domains, they pose unique challenges in computational demands and privacy preservation, particularly when deployed for training on edge devices like smartphones. To address these challenges, our study introduces the Adaptive Load- Balanced Partitioning (ALBP) technique in the GraphDHT system. This approach optimizes the division of graph datasets among edge devices, tailoring partitions to the computational capabilities of each device. By doing so, ALBP ensures efficient resource utilization across the network, significantly improving upon traditional participant selection strategies that often overlook the potential of lower-performance devices. Our methodology's core is weighted graph partitioning and model aggregation in GNNs, based on partition ratios, improving training efficiency and resource use. ALBP promotes inclusive device participation in training, overcoming computational limits and privacy concerns in large-scale graph data processing. Utilizing a DHT-based system enhances privacy in the peer-to-peer setup. The GraphDHT system, tested across various datasets and GNN architectures, shows ALBP's effectiveness in distributed GNN training and its broad applicability in different domains and structures. This contributes to applied machine learning, especially in optimizing distributed learning on edge devices. / Master of Science / Graph Neural Networks (GNNs) are a type of machine learning model that focuses on analyzing data structured like a network, such as social media connections or biological systems. These models can help identify patterns and make predictions in various tasks, but training them on large-scale datasets can require significant computing power and careful handling of sensitive data. This research proposes a new method for training GNNs on small devices, like smartphones, by dividing the data into smaller pieces and using a peer-to-peer (p2p) network for communication between devices. This approach allows the devices to work together and learn from the data while keeping sensitive information private. The main contributions of this research are threefold: (1) examining existing ways to divide network data and how they can be used for training GNNs on small devices, (2) improving the training process by creating a localized, decentralized network of devices that can communicate and learn together, and (3) testing the method on different types of datasets and GNN models, showing that it works well across a variety of situations. To sum up, this research offers a novel way to train GNNs on small devices, allowing for more efficient learning and better protection of sensitive information.

In recent years, Internet-of-Things (IoT) devices generate a large amount of personal data. However, due to the privacy concern, collecting the private data in cloud centers for training Machine Learning (ML) models becomes unrealistic. To address this problem, Federated Learning (FL) is proposed. Yet, central bottleneck has become a severe concern since the central node in traditional FL is responsible for the communication and aggregation of mil- lions of edge devices. In this paper, we propose Scalable Federated Learning via Distributed Hash Table Based Overlays for network (Scaled) to conduct multiple concurrently running FL-based applications over edge networks. Specifically, Scaled adopts a fully decentral- ized multiple-master and multiple-slave architecture by exploiting Distributed Hash Table (DHT) based overlay networks. Moreover, Scaled improves the scalability and adaptability by involving all edge nodes in training, aggregating, and forwarding. Overall, we make the following contributions in the paper. First, we investigate the existing FL frameworks and discuss their drawbacks. Second, we improve the existing FL frameworks from centralized master-slave architecture by using DHT-based Peer-to-Peer (P2P) overlay networks. Third, we implement the subscription-based application-level hierarchical forest for FL training. Finally, we demonstrate Scaled's scalability and adaptability over large scale experiments. / Master of Science / In recent years, Internet-of-Things (IoT) devices generate a large amount of personal data. However, due to privacy concerns, collecting the private data in central servers for training Machine Learning (ML) models becomes unrealistic. To address this problem, Federated Learning (FL) is proposed. In traditional ML, data from edge devices (i.e. phones) should be collected to the central server to start model training. In FL, training results, instead of the data, are collected to perform training. The benefit of FL is that private data can never be leaked during the training. However, there is a major problem in traditional FL: a single point of failure. When power to a central server goes down or the central server is disconnected from the system, it will lose all the data. To address this problem, Scaled: Scalable Federated Learning via Distributed Hash Table Based Overlays is proposed. Instead of having one powerful main server, Scaled launches many different servers to distribute the workload. Moreover, since Scaled is able to build and manage multiple trees at the same time, it allows multi-model training.

The wide-spread coverage and ubiquitous presence of mobile networks has propelled the usage and adoption of mobile phones to an unprecedented level around the globe. The computing capabilities of these mobile phones have improved considerably, supporting a vast range of third party applications. Simultaneously, Peer-to-Peer (P2P) overlay networks have experienced a tremendous growth in terms of usage as well as popularity in recent years particularly in fixed wired networks. In particular, Distributed Hash Table (DHT) based Structured P2P overlay networks offer major advantages to users of mobile devices and networks such as scalable, fault tolerant and self-managing infrastructure which does not exhibit single points of failure. Integrating P2P overlays on the mobile network seems a logical progression; considering the popularities of both technologies. However, it imposes several challenges that need to be handled, such as the limited hardware capabilities of mobile phones and churn (i.e. the frequent join and leave of nodes within a network) intensive mobile networks offering limited yet expensive bandwidth availability. This thesis investigates the feasibility of extending P2P to mobile networks so that users can take advantage of both these technologies: P2P and mobile networks. This thesis utilises OverSim, a P2P simulator, to experiment with the performance of various P2P overlays, considering high churn and bandwidth consumption which are the two most crucial constraints of mobile networks. The experiment results show that Kademlia and EpiChord are the two most appropriate P2P overlays that could be implemented in mobile networks. Furthermore, Network Address Translation (NAT) is a major barrier to the adoption of P2P overlays in mobile networks. Integrating NAT traversal approaches with P2P overlays is a crucial step for P2P overlays to operate successfully on mobile networks. This thesis presents a general approach of NAT traversal for ring based overlays without the use of a single dedicated server which is then implemented in OverSim. Several experiments have been performed under NATs to determine the suitability of the chosen P2P overlays under NATed environments. The results show that the performance of these overlays is comparable in terms of successful lookups in both NATed and non-NATed environments; with Kademlia and EpiChord exhibiting the best performance. The presence of NATs and also the level of churn in a network influence the routing techniques used in P2P overlays. Recursive routing is more resilient to IP connectivity restrictions posed by NATs but not very robust in high churn environments, whereas iterative routing is more suitable to high churn networks, but difficult to use in NATed environments. Kademlia supports both these routing schemes whereas EpiChord only supports the iterating routing. This undermines the usefulness of EpiChord in NATed environments. In order to harness the advantages of both routing schemes, this thesis presents an adaptive routing scheme, called Churn Aware Routing Protocol (ChARP), combining recursive and iterative lookups where nodes can switch between recursive and iterative routing depending on their lifetimes. The proposed approach has been implemented in OverSim and several experiments have been carried out. The experiment results indicate an improved performance which in turn validates the applicability and suitability of ChARP in NATed environments.

Media streaming over the Internet is becoming increasingly popular. Currently, most media is delivered using global content-delivery networks, providing a scalable and robust client-server model. However, content delivery infrastructures are expensive. One approach to reduce the cost of media delivery is to use peer-to-peer (P2P) overlay networks, where nodes share responsibility for delivering the media to one another. The main challenges in P2P media streaming using overlay networks include: (i) nodes should receive the stream with respect to certain timing constraints, (ii) the overlay should adapt to the changes in the network, e.g., varying bandwidth capacity and join/failure of nodes, (iii) nodes should be intentivized to contribute and share their resources, and (iv) nodes should be able to establish connectivity to the other nodes behind NATs. In this work, we meet these requirements by presenting P2P solutions for live media streaming, as well as proposing a distributed NAT traversal solution. First of all, we introduce a distributed market model to construct an approximately minimal height multiple-tree streaming overlay for content delivery, in gradienTv. In this system, we assume all the nodes are cooperative and execute the protocol. However, in reality, there may exist some opportunistic nodes,  free-riders, that take advantage of the system, without contributing to content distribution. To overcome this problem, we extend our market model in Sepidar to be effective in deterring free-riders. However, gradienTv and Sepidar are tree-based solutions, which are fragile in high churn and failure scenarios. We present a solution to this problem in GLive that provides a more robust overlay by replacing the tree structure with a mesh. We show in simulation, that the mesh-based overlay outperforms the multiple-tree overlay. Moreover, we compare the performance of all our systems with the state-of-the-art NewCoolstreaming, and observe that they provide better playback continuity and lower playback latency than that of NewCoolstreaming under a variety of experimental scenarios. Although our distributed market model can be run against a random sample of nodes, we improve its convergence time by executing it against a sample of nodes taken from the Gradient overlay. The Gradient overlay organizes nodes in a topology using a local utility value at each node, such that nodes are ordered in descending utility values away from a core of the highest utility nodes. The evaluations show that the streaming overlays converge faster when our market model works on top of the Gradient overlay. We use a gossip-based peer sampling service in our streaming systems to provide each node with a small list of live nodes. However, in the Internet, where a high percentage of nodes are behind NATs, existing gossiping protocols break down. To solve this problem, we present Gozar , a NAT-friendly gossip-based peer sampling service that: (i) provides uniform random samples in the presence of NATs, and (ii) enables direct connectivity to sampled nodes using a fully distributed NAT traversal service. We compare Gozar with the state-of-the-art NAT-friendly gossip-based peer sampling service, Nylon, and show that only Gozar supports one-hop NAT traversal, and its overhead is roughly half of Nylon’s.   / QC 20110517