Mike P. Wittie

Exploiting locality of interest in online social networks

Online Social Networks (OSN) are fun, popular, and socially significant. An integral part of their success is the immense size of their global user base. To provide a consistent service to all users, Facebook, the worlds largest OSN, is heavily dependent on centralized U.S. data centers, which renders service outside of the U.S. sluggish and wasteful of Internet bandwidth. In this paper, we investigate the detailed causes of these two problems and identify mitigation opportunities. Because details of Facebooks service remain proprietary, we treat the OSN as a black box and reverse engineer its operation from publicly available traces. We find that contrary to current wisdom, OSN state is amenable to partitioning and that its fine grained distribution and processing can significantly improve performance without loss in service consistency. Through simulations of reconstructed Facebook traffic over measured Internet paths, we show that user requests can be processed 79% faster and use 91% less bandwidth. We conclude that the partitioning of OSN state is an attractive scaling strategy for Facebook and other OSN services.

MIST: Cellular data network measurement for mobile applications

The rapid growth in the popularity of cellular networks has led to aggressive deployment and a rapid expansion of mobile services. Services based on the integration of cellular networks into the Internet have only recently become available, but are expected to become very popular. One current limitation to the deployment of many of these services is poor or unknown network performance, particularly in the cellular portion of the network. Our goal in this paper is to motivate and present the Mobile Internet Services Test (MIST) platform, a new distributed architecture to measure and characterize cellular network performance as experienced by mobile devices. We have used MIST to conduct preliminary measurements; evaluate MIST’s effectiveness; and motivate further measurement research.

Assessment of Multi-Hop Interpersonal Trust in Social Networks by Three-Valued Subjective Logic

Assessing multi-hop interpersonal trust in online social networks (OSNs) is critical for many social network applications such as online marketing but challenging due to the difficulties of handling complex OSN topology, in existing models such as subjective logic, and the lack of effective validation methods. To address these challenges, we for the first time properly define trust propagation and combination in arbitrary OSN topologies by proposing 3VSL (Three-Valued Subjective Logic). The 3VSL distinguishes the posteriori and priori uncertainties existing in trust, and the difference between distorting and original opinions, thus be able to compute multi-hop trusts in arbitrary graphs. We theoretically proved the capability based on the Dirichlet distribution. Furthermore, an online survey system is implemented to collect interpersonal trust data and validate the correctness and accuracy of 3VSL in real world. Both experimental and numerical results show that 3VSL is accurate in computing interpersonal trust in OSNs.

Survey of End-to-End Mobile Network Measurement Testbeds, Tools, and Services

Mobile (cellular) networks enable innovation, but can also stifle it and lead to user frustration when network performance falls below expectations. As mobile networks become the predominant method of Internet access, developer, research, network operator, and regulatory communities have taken an increased interest in measuring end-to-end mobile network performance to, among other goals, minimize negative impact on application responsiveness. In this survey, we examine current approaches to end-to-end mobile network performance measurement, diagnosis, and application prototyping. We compare available tools and their shortcomings with respect to the needs of developers, researchers, network operators, and regulators. We intend for this survey to provide a comprehensive view of currently active efforts and some auspicious directions for future work in mobile network measurement and mobile application performance evaluation.

Itrust: interpersonal trust measurements from social interactions

Interpersonal trust is widely cited as an important component in several network systems such as peer-to-peer networks, e-commerce, and semantic web. However, there has been less research on measuring interpersonal trust due to the difficulty of collecting data that accurately reflect interpersonal trust. Currently, friends of a user in almost all OSNs are indistinguishable, that is, there is no explicit indication of the strength of trust between a user and his/her close friends, as opposed to acquaintances. To address this issue, we quantify interpersonal trust by analyzing the social interaction frequencies between users and their friends on Facebook. We consider bidirectional interacting data in OSNs to deconstruct a users social behavior and apply principal component analysis to estimate interpersonal trust. A Facebook app, itrust, is developed to collect interaction data and calculate interpersonal trust. Results show that itrust achieves more accurate interpersonal trust measurements than existing methods.

A case for faster mobile web in cellular IPv6 networks

The transition to IPv6 cellular networks creates uncertainty for content providers (CPs) and content delivery networks (CDNs) of whether and how to follow suit. Do CPs that update their CDN contracts to allow IPv6 hosting achieve better, or worse performance in mobile networks? Should CDNs continue to host mobile content over IPv4 networks, or persuade to their CP customers the performance benefits of IPv6 content delivery? In this paper we answer these questions through a comprehensive comparison of IPv4 and IPv6 mobile Web performance in cellular networks in the US from the point of view of Akamais content delivery infrastructure. Our data show that IPv6 hosting outperforms legacy IPv4 paths in mobile Web. Our analysis leads to clear recommendations for CPs to transition to IPv6-hosted mobile Web. Finally, we propose new mechanisms, through which CDNs can safely transition mobile content to IPv6-enabled servers for improved content delivery.

Impact of Network Performance on Cloud Speech Recognition

Interactive real-time communication between people and machine enables innovations in transportation, health care, etc. Using voice or gesture commands improves usability and broad public appeal of such systems. In this paper we experimentally evaluate Google speech recognition and Apple Siri - two of the most popular cloud-based speech recognition systems. Our goal is to evaluate the performance of these systems under different network conditions in terms of command recognition accuracy and round trip delay - two metrics that affect interactive application usability. Our results show that speech recognition systems are affected by loss and jitter, commonly present in cellular and WiFi networks. Finally, we propose and evaluate a network coding transport solution to improve the quality of voice transmission to cloud-based speech recognition systems. Experiments show that our approach improves the accuracy and delay of cloud speech recognizers under different loss and jitter values.

Faster Web through Client-Assisted CDN Server Selection

Modern websites use Content Delivery Networks (CDNs) to speed up the delivery of static content. However, we show that DNS-based selection of CDN servers can be refined to fully deliver on the speedup of CDNs. We propose DNS-Proxy (dp), a client-side process that shares load-balancing functionality with CDNs by choosing from among resolved CDN servers based on last mile network performance. Our measurement study of CDN infrastructure deployed by five major CDN providers shows that dp reduces webpage load time by 29% on average. If dp has already resolved the domain, the reduction in webpage load time is as much as 40%. Finally, dp reduces the load time of individual static Web objects by as much as 43%. We argue that dp enables a more effective use of existing content delivery infrastructure and represents a complementary strategy to a continual increase of geographic content availability.

Uncovering the mystery of trust in an online social network

Trust is a hidden fabric of online social networks (OSNs) that enables online interactions, e.g., online transactions on Ebay. The fundamental properties of trust in OSNs, however, have not been adequately studied yet. In this work, we advance the understanding of trust in OSNs by analyzing the Advogato dataset [1]. We study the properties of direct trust, indirect trust, and trust community detection in Advogato. We found that 1) the trust between users are asymmetric, 2) high degree users are usually associated with high trust, 3) diversity in peoples opinions on the same person will affect indirect trust inference, 4) users live in many separate “small small worlds” from the perspective of trust and it is difficult to identify these “small small worlds” with existing random walk-based community detection algorithms, e.g., ACL [2]. It in fact motivates the need for a new community detection algorithm to identify clusters of user connected by trustful relations. Although our findings are from a specific OSN, they can significantly impact how OSNs are designed and configured in the future, e.g., a better user crowdsourcing setting based on trust information.

Detecting Cellular Middleboxes Using Passive Measurement Techniques

The Transmission Control Protocol (TCP) follows the end-to-end principle – when a client establishes a connection with a server, the connection is only shared by two physical machines, the client and the server. In current cellular networks, a myriad of middleboxes disregard the end-to-end principle to enable network operators to deploy services such as content caching, compression, and protocol optimization to improve end-to-end network performance. If server operators remain unaware of such middleboxes, TCP connections may not be optimized specifically for middleboxes and instead are optimized for mobile devices. We argue that without costly active measurement, it remains challenging for server operators to reliably detect the presence of middleboxes that split TCP connections. In this paper, we present three techniques (based on latency, loss, and characteristics of TCP SYN packets) for server operators to passively identify Connection Terminating Proxies (CTPs) in cellular networks, with the goal to optimize TCP connections for faster content delivery. Using TCP and HTTP logs recorded by Content Delivery Network (CDN) servers, we demonstrate that our passive techniques are as reliable and accurate as active techniques in detecting CTPs deployed in cellular networks worldwide.