On the Privacy and Performance of Mobile Anonymous Microblogging

On the Privacy and Performance of Mobile
Anonymous Microblogging

[pdf-embedder url=”http://wellapets.com/wp-content/uploads/2019/06/On-the-Privacy-and-Performance-of-Mobile.pdf” title=”On the Privacy and Performance of Mobile”]

Abstract—Microblogging is a popular form of Online Social Networking (OSN) activity. It allows users to send messages in a one-to-many publish-subscribe manner. Most current service providers are centralized and deploy a client-server model with unencrypted message content. As a consequence, all user behavior can, by default, be monitored, and censoring based on message content can easily be enforced on the server side. A distributed, peer-to-peer microblogging system consisting of mobile smartphone-equipped users that exchange group encrypted messages in an anonymous and censorship-resistant manner can alleviate privacy and censorship issues. We experimentally evaluate message spread of such systems with simulations that run on a range of synthetic and real-world mobility inputs, thus extending previous work. We show that such systems are feasible for a range of mobility and network settings, both under normal and under adversarial conditions, e.g., under the presence of nodes which jam the network or send spam. Index Terms—Microblogging, anonymity, censorshipresistance, peer-2-peer, mobile networking, simulation.

MICROBLOGGING is a popular form of Online Social Networking (OSN) activity. It is part of an area of OSN known as micromedia, where users share small snippets of content media such as text, images or video. Specifically, we focus on microblogging of small text messages. Users can subscribe to another user’s channel or create message channels themselves. A service broadly used to this end is Twitter. Client-server architectures back most of the existing microblogging services, and hence message content is visible to service operators. Confidentiality is thus not addressed and service providers might analyze all user-generated content. Moreover, not only message content but the whole user activity is transparent to providers, for instance, received and sent messages as well as group subscriptions. Another issue is the susceptibility of the servers for censorship.
Marius Senftleben, Ana Barroso, Matthias Hollick, Stefan Katzenbeisser and Erik Tews are with the department of computer science at Technical University of Darmstadt and the Center of Advanced Security Research Darmstadt (CASED), Darmstadt, Germany (email:fsenftlebenjkatzenbeisserje tewsg@seceng.informatik.tu-darmstadt.de, fabarrosojmhollickg@seemoo.tu-darmstadt.de). Mihai Bucicoiu is with the department of computer science at Politehnica University of Bucharest, Bucharest, Romania (e-mail: mihai.bucicoiu@cs.pub.ro). can thus be stricken out depending on their content, sender, or receiver, hence interfering content spread according to some censorship policies. Furthermore, centralized systems are vulnerable to local Internet outages, either unintentionally, e.g., power grid failures, or intentionally, e.g., Internet shutdowns via ‘kill-switches’. In both cases such systems cease to function.
To overcome the aforementioned issues, and based on the proposals presented in [1]–[3], we thoroughly evaluate and analyze the feasibility of a mobile distributed microblogging system consisting of users carrying their mobile devices. We extend the work of [1], whose system goals are anonymity, message confidentiality, and censorship-resistance. All messages
are encrypted under a group key, and they are stored across all nodes in a decentralized fashion. In particular, we evaluate a system where message propagation is carried out using local, point-to-point communications taking place between close pairs of nodes; stressing that this is a middleware-free scenario without orchestrating entities.
Our contributions are as follows:
 We experimentally show the feasibility of our mobile microblogging system using simulations with both synthetic and empirical mobility datasets.
 We simulate different networking setups to gain insights on how they affect message spread.
 We evaluate the system under adversarial conditions, discuss its privacy, and state lessons learned.
The rest of the article is structured as follows: an overview of the system is given in Section II. Section III details the different synthetic and empirical mobility datasets used to conduct the simulations, followed by the evaluation of the simulation results in Section IV. Section V deals with the system’s performance under adversarial conditions. In Section VI we discuss our findings. Section VII treats related work, and Section VIII concludes the paper.