Showing papers presented at "International Conference on Peer-to-Peer Computing in 2014"

Bypassing the cloud: Peer-assisted event dissemination for augmented reality games

Abstract: The rising number of mobile devices and their increasing computational capabilities enable new interactive context-sensitive applications. Popular examples are augmented reality games such as Google's Ingress, where users interact with each other in the real world while being part of the game at the same time. This local interaction pattern in the real world as well as in the game is not reflected in the underlying communication pattern. Every locally generated game event is first transferred to a backend server via a cellular connection, from where it is then further disseminated to all players within the given area of interest. This communiation pattern introduces significant delays and limits the interactivity of the game. In this work, we propose an event dissemination system that exploits the locality characteristics of mobile augmented reality games to (i) enable and configure local peer-to-peer dissemination of events when appropriate and (ii) reconfigure or replace the utilized peer-to-peer protocol to adapt to a wide range of requirements. Through extensive evaluation we show that the proposed system decreases the delivery delay by a factor of eight compared to the existing communication pattern, leading to significantly increased information accuracy.

Replicated erasure codes for storage and repair-traffic efficiency

Abstract: This paper introduces a new family of redundancy schemes for distributed storage systems, called replicated erasure codes (REC), which combine the storage-space efficiency of erasure codes and the repair-traffic efficiency of replication. A formal model for analyzing the storage and repair-traffic costs under availability and persistency constraints is also developed. It is shown that under parameters that characterize common P2P environments, REC generally achieves better results than each of the two methods separately.

Decentralizing the cloud: How can small data centers cooperate?

Abstract: Cloud computing has become pervasive due to attractive features such as on-demand resource provisioning and elasticity. Most cloud providers are centralized entities that employ massive data centers. However, in recent times, due to increasing concerns about privacy and data control, many small data centers (SDCs) established by different providers are emerging in an attempt to meet demand locally. However, SDCs can suffer from resource in-elasticity due to their relatively scarce resources, resulting in a loss of performance and revenue. In this paper we propose a decentralized cloud model in which a group of SDCs can cooperate with each other to improve performance. Moreover, we design a general strategy function for the SDCs to evaluate the performance of cooperation based on different dimensions of resource sharing. Through extensive simulations using a realistic data center model, we show that the strategies based on reciprocity are more effective than other involved strategies, e.g., those using prediction on historical data. Our results show that the reciprocity-based strategy can thrive in a heterogeneous environment with competing strategies.

Defining and understanding smartphone churn over the internet: A measurement study

Abstract: Smart portable devices form the largest personal computing platform so far in human history, yet the adoption of P2P techniques has been very slow. One reason is the lack of a detailed understanding of the dynamic patterns of network connectivity and battery usage. For example, we know that when a smartphone is on a charger connected to a WiFi network behind a friendly NAT device, it can act as an unrestricted P2P node. However, we do not know how to model these “good” intervals and so we do not know what P2P applications are possible at all if we restrict participation to only such intervals. This raises a problem similar to modeling churn in classical P2P research. We are not aware of any suitable and available measurement data sets or models. To address this problem, we developed a publicly available smart phone app that provides the user with information about the current network connection such as NAT type, public IP, and so on. The app also collects data about network connectivity and battery status in the background. The app has been downloaded by several hundred users from all over the world. Based on this data we identify and model the sessions during which a phone can participate in a P2P protocol. We also demonstrate through the simulation of gossip protocols that it is feasible to develop smartphone-friendly P2P applications. The raw data is also available for research purposes in an anonymized form upon request.

The power of indirect ties in friend-to-friend storage systems

Abstract: Friend-to-Friend (F2F) storage systems were shown to suffer from two significant limitations. First, users with a small set of friends are penalized by lack of available storage for their needs, while users with many friends get overloaded with resource requests. Second, friends are typically in close geographical proximity to each other, and thus their online times are synchronized, leading to low data availability when they are offline. This paper addresses these concerns by expanding the set of storage resources while still using a measure of social incentives. It proposes an indirect tie measurement to compute the social strength between possibly distant nodes in a social network. Using datasets from co-authorship networks and a video gaming community, we show that the social strength-based mechanism more than doubles the set of storage candidates motivated by social incentives, invites socially low connected users to contribute more resources and improves data availability by up to 6.5 times. We also show that our method complemented with simple load balancing strategies significantly improves peer engagement and workload distribution in F2F storage systems.

Clubbing with the peers: A measurement study of BitTorrent live

Abstract: The peer-to-peer approach can greatly help to cope with highly dynamic live streaming workload by using idle client resources. Yet, P2P streaming typically comes at the cost of increased streaming delays caused by the inevitable multi-hop forwarding of content by peers within the overlay. Various P2P streaming approaches have been proposed aiming at a good tradeoff between flexibility, streaming delay, and costs in terms of traffic overhead for both content providers and clients. Recently, BitTorrent Inc. released a new P2P live streaming system termed BTLive, specifically targeted at low delay and low overhead. For content providers investigating the applicability of BTLive's approach, it is essential to understand its properties as well as its limitations. So far, no publicly available study exists that quantitatively analyzes BTLive's performance. To this end, this paper presents a measurement study of the official beta version of BTLive. The study aims to answer the following key questions: How peer-to-peer is BTLive? How delay optimized is BTLive? What is the overhead of BTLive? To answer these questions, traces of real BTLive traffic between a broadcast server and a number of peers deployed across Europe have been analyzed.

HSkip+: A self-stabilizing overlay network for nodes with heterogeneous bandwidths

Abstract: In this paper we present and analyze HSkip+, a self-stabilizing overlay network for nodes with arbitrary heterogeneous bandwidths. HSkip+ has the same topology as the Skip+ graph proposed by Jacob et al. [1] but its self-stabilization mechanism significantly outperforms the self-stabilization mechanism proposed for Skip+. Also, the nodes are now ordered according to their bandwidths and not according to their identifiers. Various other solutions have already been proposed for overlay networks with heterogeneous bandwidths, but they are not self-stabilizing. In addition to HSkip+ being self-stabilizing, its performance is on par with the best previous bounds on the time and work for joining or leaving a network of peers of logarithmic diameter and degree and arbitrary bandwidths. Also, the dilation and congestion for routing messages is on par with the best previous bounds for such networks, so that HSkip+ combines the advantages of both worlds. Our theoretical investigations are backed by simulations demonstrating that HSkip+ is indeed performing much better than Skip+ and working correctly under high churn rates.

Fully distributed robust singular value decomposition

Abstract: Low-rank matrix approximation is an important tool in data mining with a wide range of applications including recommender systems, clustering, and identifying topics in documents. The problem we tackle is implementing singular value decomposition (SVD)-a popular method for low rank approximationin large fully distributed P2P systems in a robust and scalable manner. We assume that the matrix to be approximated is stored in a large network where each node knows one row of the matrix (personal attributes, documents, media ratings, etc). In our P2P model, we do not allow this personal information to leave the node, yet we want the nodes to collaboratively compute the SVD. Methods applied in large scale distributed systems such as synchronized parallel gradient search or distributed iterative methods are not preferable in our system model due to their requirements of synchronized rounds or their inherent issues with load balancing. Our approach overcomes these limitations with the help of a distributed stochastic gradient search in which the personal part of the decomposition remains local, and the global part (e.g., movie features) converges at all nodes to the correct value. We present a theoretical derivation of our algorithm, as well as a thorough experimental evaluation of real and synthetic data as well. We demonstrate that the convergence speed of our method is competitive while not relying on synchronization and being robust to extreme failure scenarios.

Dataset on usage of a live & VoD P2P IPTV service

Abstract: This paper presents a dataset of user statistics collected from a P2P multimedia service infrastructure that delivers both live and on-demand content in high quality to users via different platforms: PC/Mac, and set top boxes. The dataset covers a period of seven months starting from October 2011, exposing a total of over 94k system statistic reports from thousands of user devices at a fine granularity. Such rich data source is made available to fellow researchers to aid in developing better understanding of video delivery mechanisms, user behaviour, and programme popularity evolution.

A peer-to-peer service architecture for the Smart Grid

Abstract: Important challenges in interoperability, reliability, and scalability need to be addressed before the Smart Grid vision can be fulfilled. The sheer scale of the electric grid and the criticality of the communication among its subsystems for proper management, demands a scalable and reliable communication framework able to work in an heterogeneous and dynamic environment. Moreover, the need to provide full interoperability between diverse current and future energy and non-energy systems, along with seamless discovery and configuration of a large variety of networked devices, ranging from the resource constrained sensing devices to servers in data centers, requires an implementation-agnostic Service Oriented Architecture. In this position paper we propose that this challenge can be addressed with a generic framework that reconciles the reliability and scalability of Peer-to-Peer systems, with the industrial standard interoperability of Web Services. We illustrate the flexibility of the proposed framework by showing how it can be used in two specific scenarios.

Censorship-resistant and privacy-preserving distributed web search

Abstract: The vast majority of Internet users are relying on centralized search engine providers to conduct their web searches. However, search results can be censored and search queries can be recorded by these providers without the user's knowledge. Distributed web search engines based on peer-to-peer networks have been proposed to mitigate these threats. In this paper we analyze the three most popular real-world distributed web search engines: Faroo, Seeks and Yacy, with respect to their censorship resistance and privacy protection. We show that none of them provides an adequate level of protection against an adversary with modest resources. Recognizing these flaws, we identify security properties a censorship-resistant and privacy-preserving distributed web search engine should provide. We propose two novel defense mechanisms called node density protocol and webpage verification protocol to achieve censorship resistance and show their effectiveness and feasibility with simulations. Finally, we elaborate on how state-of-the-art defense mechanisms achieve privacy protection in distributed web search engines.

LAYSTREAM: Composing standard gossip protocols for live video streaming

Abstract: Gossip-based live streaming is a popular topic, as attested by the vast literature on the subject. Despite the particular merits of each proposal, all need to implement and deal with common challenges such as membership management, topology construction and video packets dissemination. Well-principled gossip-based protocols have been proposed in the literature for each of these aspects. Our goal is to assess the feasibility of building a live streaming system, LAYSTREAM, as a composition of these existing protocols, to deploy the resulting system on real testbeds, and report on lessons learned in the process. Unlike previous evaluations conducted by simulations and considering each protocol independently, we use real deployments. We evaluate protocols both independently and as a layered composition, and unearth specific problems and challenges associated with deployment and composition. We discuss and present solutions for these, such as a novel topology construction mechanism able to cope with the specificities of a large-scale and delay-sensitive environment, but also with requirements from the upper layer. Our implementation and data are openly available to support experimental reproducibility.

Reducing costs in the personal cloud: Is bittorrent a better bet?

Abstract: Lately Personal Cloud storage services, like Drop-box, have emerged as user-centric solutions that provide easy management of the users' data. To meet the requirements of their clients, such services require a huge amount of storage and bandwidth. In an attempt to reduce these costs, we focus on maximizing the benefit that can be driven from the interest of users in the same content by the introduction of the BitTorrent protocol. In general, it is assumed that BitTorrent is only effective for large files and/or large swarms, while the client-server approach is more suited for small files and/or small swarms. However, there is no concrete study on the comparative efficiency of both protocols for small files yet. In this paper, we study the download time and offload ratio in BitTorrent compared to HTTP. Based on this study, we propose an algorithm for the management of these protocols. The choice of the protocol is made based on the prediction of the efficiency of BitTorrent and HTTP for each case. We validate our algorithm on a real trace of the Ubuntu One file service, achieving important savings in the cloud bandwidth without degrading the download time.

Cloudy knapsack problems: An optimization model for distributed cloud-assisted systems

Abstract: Cloud-assisted approaches for both peer-to-peer systems and mobile apps require optimized use of elastic cloud resources. Due to budget constraints, a subset of the tasks has to be selected for offloading considering context parameters like device battery level and task variability. This leads to the challenging problem of context-sensitive task scheduling on elastic resources with a limited global view, which is not addressed by existing works. We identify a new class of formal problems called cloudy knapsack problems to effectively model the same. Abstracting out the problem formally can spur future independent works related to different variants of the problem and corresponding bounds and optimal algorithms. We illustrate the global view related issues through simulations, identify some theoretical bounds for a variant of cloudy knapsack problems and discuss several open problems.

4P: Performant private peer-to-peer file sharing

Abstract: In recent years fully decentralized file sharing systems were developed aimed at improving anonymity among their users. These systems provide typical file sharing features such as searching for and downloading files. However, elaborate schemes originally aimed at improving anonymity cause partial keyword matching to be virtually impossible, or introduce a substantial bandwidth overhead. In this paper we introduce 4P, a system that provides users with anonymous search on top of a semantic overlay. The semantic overlay allows users to efficiently locate files using partial keyword matching, without having to resort to an expensive flooding operation. Included into 4P are a number of privacy enhancing features such as probabilistic query forwarding, path uncertainty, caching, and encrypted links. Moreover, we integrate a content retrieval channel into our protocol allowing users to start downloading a file from multiple sources immediately without requiring all intermediate nodes to cache a complete copy. Using a trace-based dataset, we mimic a real-world query workload and show the cost and performance of search using six overlay configurations, comparing random, semantic, Gnutella, RetroShare, and OneSwarm to 4P. The state-of-the-art flooding based alternatives required approximately 10,000 messages to be sent per query, in contrast 4P only required 313. Showing that while flooding can achieve a high recall (more than 85% in our experiments) it is prohibitively expensive. With 4P we achieve a recall of 76% at a considerable reduction in messages sent.

Stochastic models of pull-based data replication in P2P systems

Abstract: We consider pull-based data synchronization issues between a source and its replicas in P2P networks. Under continuous information change and lazy synchronization, these systems are highly susceptible to serving outdated content, which negatively affects their performance and user satisfaction. To understand these scenarios, we first introduce a novel model of interaction between two stochastic point processes — updates at the source and downloads at the replica — and derive the probability that a random query against the replica retrieves fresh content. Unlike prior work, we assume non-Poisson dynamics and determine statistical properties of the replication process that make it perform better for a given download rate. The second half of the paper applies these results to several more difficult algorithms — cascaded replication, cooperative caching, and redundant querying from the clients. Surprisingly, we discover that optimal cooperation involves just a single peer and that redundant querying can hurt the ability of the system to handle load (i.e., may lead to lower scalability).

Routing table construction method solely based on query flows for structured overlays

Abstract: In structured overlays, nodes forward a query hop by hop to deliver it to the responsible node for the query. Each node maintains its routing table and determines the next hop by referring to the routing table. Each node has its node identifier and determines which other nodes to be on its routing table based on node distance, that is defined by difference of node identifiers or the number of nodes between two nodes in identifier order. In existing structured overlays, routing table construction and maintenance based on node distance enable efficient lookup, that is a small number of hop counts to the responsible node. We found out that efficient lookup does not require node distance in routing table construction and maintenance. As an example, this paper presents Flow-based Flexible Routing Tables (FFRT), a routing table construction method solely based on query flows. In an FFRT-based overlay, a node calculates query flows, that is the amount of queries forwarded to each node on its routing table. And the node maintains its routing table toward a state in which all nodes on the table have equal query flows. FFRT also provides such a practical merit as it performs efficient lookups though node and queries' target identifiers are distributed nonuniformly. The merit enables range query support.

Automatic transport selection and resource allocation for resilient communication in decentralised networks

Abstract: Making communication more resilient is a main focus for modern decentralised networks. A current development to increase connectivity between participants and to be resilient against service degradation attempts is to support different communication protocols, and to switch between these protocols in case degradation or censorship are detected. Supporting multiple protocols with different properties and having to share resources for communication with multiple partners creates new challenges with respect to protocol selection and resource allocation to optimally satisfy the applications' requirements for communication. This paper presents a novel approach for automatic transport selection and resource allocation with a focus on decentralised networks. Our goal is to evaluate the communication mechanisms available for each communication partner and then allocate resources in line with the requirements of the applications. We begin by detailing the overall requirements for an algorithm for transport selection and resource allocation, and then compare three different solutions using (1) a heuristic, (2) linear optimisation, and (3) machine learning. To show the suitability and the specific benefits of each approach, we evaluate their performance with respect to usability, scalability and quality of the solution found in relation to application requirements.

Diversity entails improvement: A new neighbour selection scheme for Kademlia-type systems

Abstract: Discovery of nodes and content in large-scale distributed systems is generally based on Kademlia, today. Understanding Kademlia-type systems to improve their performance is essential for maintaining a high service quality for an increased number of participants, particularly when those systems are adopted by latency-sensitive applications. This paper contributes to the understanding of Kademlia by studying the impact of diversifying neighbours' identifiers within each routing table bucket on the lookup performance. We propose a new, yet backward-compatible, neighbour selection scheme that attempts to maximize the aforementioned diversity. The scheme does not cause additional overhead except negligible computations for comparing the diversity of identifiers. We present a theoretical model for the actual impact of the new scheme on the lookup's hop count and validate it against simulations of three exemplary Kademlia-type systems. We also measure the performance gain enabled by a partial deployment for the scheme in the real KAD system. The results confirm the superiority of the systems that incorporate our scheme.

RankSlicing: A decentralized protocol for supernode selection

Abstract: In peer-to-peer applications deployed on the Internet, it is common to assign greater responsibility to supernodes, which are usually peers with high computational power, large amount of memory, or high network bandwidth capacity. In this paper, we describe a practical solution to the problem of supernode selection, that is the process of discovering the best peers in the network by some application-specific metric. We provide a distributed heuristic that allows to identify the best K nodes in the P2P overlay, by taking into consideration the realities of actual deployments, such as the presence of NATs. Our approach consists of an epidemic protocol which does not require new connections to be established, but rather relies on established connections, such as the ones provided by a NAT-resilient peer sampling framework. We support our claims with a thorough evaluation of our solution in simulation and in a real deployment on thousands of consumer machines.

An empirical study of the global behavior of a structured overlay network

Abstract: Distributed applications built on top of Structured Overlay Networks (SONs) operate based on certain assurances from the underlying Peer-to-Peer network. Such applications continue to increase in scale, becoming more complex and difficult to manage. The situation becomes worse if the behavior of underlying SON is non-deterministic or even unknown for a given scenario. This implies non-trivial questions: what behavior should the application layer expect from the underlying SON in a given scenario? Under what conditions should a SON exhibit resiliency against an extremely hostile environment? Ideally, the behavior of a complex system such as a SON needs to be defined for every possible operating condition. Existing literature lacks a systematic and in-depth study of the global behavior of a SON. This work is a step towards answering those questions, which starts by proposing an organization of the global operating space of a SON, also defines the term “behavior”, with respect to a SON. In order to conduct the experimental study, an existing ring-based SON, namely Beernet, is chosen as a representative. As the entire operating space of a SON is extremely large, due to space limitation, this paper presents the first results of our investigation, the behavior of Beernet along the dimension of Churn. The study assesses behaviors like key availability, updates, replica management, and failed transactions, as a function of churn up to 100% node turnover per time unit. The result shows that, continuous injection of extremely high churn causes the ring to be dissolved, creating isolation of peers. However, at such a high node turnover of 100% per 5s, there are instances, where the % of failed transactions didn't reach 100%, especially in cases, when the join events dominate failures during initial period.

100 Million DHT replies

Abstract: Crawling a DHT allows researchers to monitor the behaviour of peers, determine their geographic location, etc. However, it has always been an error-prone process as it is not feasible to capture a full snapshot of the Mainline DHT in a timely manner. Therefore, researchers have developed approximation methods which can run on subsets of the DHT and extrapolate those in order to reason on the size and structure of the complete DHT. However, in this paper we introduce a new method of collecting information on peers connected to a DHT. We exploit the caches present at bootstrap servers to collect information on peers which are currently connected. Originally added to the bootstrap servers in order to be able to withstand more than 20,000 requests for peers each second, we now use the same mechanism as peers bootstrapping into the DHT to discover more than 20 Million peers in less than 2 hours. Using the bootstrap servers, we discover more than twice as many peers as BitMon, which crawls a subset of the DHT and then extrapolates. Moreover, in contrast to related work which often require highly customized/tuned BitTorrent clients, our script consists of 90 lines of Python code and runs on a simple laptop.

On the tradeoff between resilience and degree overload in dynamic P2P graphs

Abstract: Prior work has shown that resilience of random P2P graphs under dynamic node failure can be improved by age-biased neighbor selection that delivers outbound edges to more reliable users with higher probability. However, making the bias too aggressive may cause the group of nodes that receive connections to become small in comparison to graph size. As they become overloaded, these peers are forced to reject future connection requests, which leads to potentially unbounded join delays and high traffic overhead. To investigate these issues, we propose several analytical models for understanding the interplay between resilience and degree. We formulate a Pareto-optimal objective for this tradeoff, introduce new metrics of resilience and degree, analyze them under Pareto lifetimes, and discover that traditional techniques can be highly suboptimal in this setting. We then show evidence that optimization can be solved by a family of step-functions, which connect outgoing edges to uniformly random users whose age exceeds some threshold.

User freedom: To be or not to be a ‘supernode’

Abstract: Peer-to-peer infrastructures in a world where many systems are mobile, or found behind NATs and firewalls that cannot be programmed by users, require significant support from computers with appropriate connectivity, called supernodes. We introduce a new fully decentralized unstructured peer-to-peer (P2P) approach to open-source instant messaging systems (that is employed in the DirectDemocracyP2P system). Here each human owning a peer can control the traffic supported by her system. The control may be based on criteria such as: (a) her desire to help the endpoints of the communication based on friendship, (b) her desire to help a cause based on the content/topic of the communication, (c) reputation and rewards, or (d) her interest in the handled data itself. Providing intrinsic incentives for peers to help with traffic is important in order for an open-source freeware P2P system to eventually be viable. In non-incentive P2P systems like Skype, availability of open-source versions can potentially starve the system of supernodes (once users learn how to disable the resource consuming supernode-function). This paper analyzes key functions of the solution such as integration of incentive management into an extension of the STUN protocol, connection establishment and message transfer under different network setups. We show how to use the results of surveys, in conjunction with simulations, to quantify the effects of various incentives on the survivability of an open approach.