Clearwater Skyline
IEEE site
IEEE Computer Society

Monday, November 7, Ruby Room

9:00 AM - 10:30 AM

WNM Session 1:

9:00 Measuring Broadband Access Network Performance in Pakistan: A Comparative Study

M Faheem Awan and Tahir Ahmad (National University of Science and Technology, Pakistan); Saad B. Qaisar (School of Electrical Engineering and Computer Science (SEECS), NUST & National University of Sciences & Technology, Pakistan); Nick Feamster (Georgia Institute of Technology, USA); Srikanth Sundaresan (International Computer Science Institute(ICSI), USA) Although broadband Internet enjoys wide penetration in Pakistan, performance of fixed and wireless broadband networks have not been thoroughly investigated from perspective of end-users. To the best of our knowledge, no independent study exists to date documenting home user broadband experience in Pakistan. This dearth of information is troubling because benchmarking broadband performance is fundamentally important for consumers in Pakistan because of its big Internet market. To address this gap, we conducted a pilot study of both fixed landline and fixed wireless broadband connections in Pakistan through deployment of programmable wireless routers for home users. Our work is inspired by (and compares to) a previous study of fixed and wireless broadband performance in South Africa that was published in 2013, drawing comparisons to previous work where appropriate. We measure the performance of both fixed landline and fixed wireless ISPs. Our findings agree with the findings of this previous work, suggesting that (1) consumers in Pakistan are not getting advertised speeds; (2) wireless broadband satisfies the advertised performance bounds better than wired broadband in a majority of cases under consideration; (3) network latency often contributes to performance bottlenecks moreso than throughput; and (4) ISPs in Pakistan shapes internet traffic according to their profile. Limiting factors such as latency mean that investing in local server infrastructure and rich peering between different ISPs will also make a significant difference in improving the Internet experience for users in Pakistan.

9:30 Remotely Inferring Device Manipulation of Industrial Control Systems Via Network Behavior

Georgios Lontorfos (Johns Hopkins University Information Security Institute, USA); Kevin D. Fairbanks (United States Naval Academy, USA); Lanier Watkins (Johns Hopkins University Information Security Institute, USA); William H. Robinson (Vanderbilt University, USA) This paper presents preliminary findings on a novel method to remotely fingerprint a network of Industrial Control System (ICS) devices, and to use the fingerprint to infer the functionality of the devices. A monitoring node measures the target device's response to network requests and statistically analyzes the collected data to build and classify a profile of the devices functionality via machine learning. As ICSs are used to control critical infrastructure processes such as power generation and distribution, it is vital to develop methods to detect tampering. A system employing our measurement technique could discover if an insider has made unauthorized changes to a device's logic. Our architecture also has advantages because the monitoring node is separate from the measured device. Our results indicate the ability to accurately infer (i.e., using a tunable threshold value) discrete ranges of task cycle periods (i.e., CPU loads) that could correspond to different ICS functions.

10:00 Feature Selection for Robust Backscatter DDoS Detection

Eray Balkanli, Nur Zincir-Heywood and M. I. Heywood (Dalhousie University, Canada) This paper proposes a novel approach to detect Backscatter DDoS traffic using only packet header features. Specifically, we analyze the performances of different learning classifiers trained on the darknet and normal traffic traces captured in different years over the Internet in terms of attack detection rate, false alarm rate and computational cost. To achieve this, we analyzed four different training sets with four different feature set. We employed two well-known feature selection algorithms, namely Chi -Square and Symmetrical Uncertainty, together with the decision tree classifier. All the datasets employed are publicly available and provided by CAIDA. Our experimental results show that it is possible to develop a robust detection system that can generalize well to the changing Backscatter DDoS behaviours over time.

10:30 PM - 11:00 AM

Coffee Break

11:00 AM - 12:30 PM

WNM Session 2:

11:00 An Analysis of the YouNow Live Streaming Platform

Denny Stohr (Technische Universit├Ąt Darmstadt, Germany); Tao Li (TU Dresden, Germany); Stefan Wilk (TU Darmstadt, National University of Singapore, Germany); Silvia Santini (TU Dresden, Germany); Wolfgang Effelsberg (University of Mannheim, Germany) Video streaming platforms like or YouNow have attracted the attention of both users and researchers in the last few years. Users increasingly adopt these platforms to share user-generated videos while researchers study their usage patterns to learn how to provide better and new services. In this paper, we focus on the YouNow platform and show the results of a analysis of its traffic patterns and other characteristics. To perform this analysis, we have collected YouNow usage patterns for 85994 users over a period of about one month. Our results show that YouNow's characteristics are in part equal to and in part different from those of other video streaming platforms. Like on YouTube or, for instance, few YouNow videos attract most of the view requests. On the other side, YouNow sessions are notably shorter than ones. We believe the observation of these similarities and differences to be crucial to inform the design and implementation of better upcoming video streaming services.

11:30 An Analysis of Friend Circles of Facebook Users

Esra Erdin, Eric Klukovich and Mehmet Hadi Gunes (University of Nevada, Reno, USA) The Internet has spawned different information exchange systems ranging from file sharing to online social networking. Online social networks (OSN) have gained significant popularity and are among the most popular sites. Users interact through various features of online social networking services such as connecting with friends, establishing new friendships, sharing photos and videos, and commenting on each others posts. As digital interactions supersede over physical interactions, it is important to understand how users interact over such digital platforms. Moreover, understanding OSN traffic is valuable in designing new OSN platforms and adjusting content distribution network. In this study, we provide a deeper analysis of online activities that happen around a user, i.e., circle. To better understand activities in a user's friend circle, we developed a Facebook application that monitored the walls of Facebook users. In particular, we obtained statistics of every post shared with the user by their friends for 16 volunteers that provided continuous measurements for 15 days.

12:00 Understanding Evolution and Adoption of Top-level Domain Names

Thitipong Jarassriwilai, Tiffany Dauber, Nevil Brownlee and Aniket Mahanti (University of Auckland, New Zealand) The Domain Name System (DNS) is used in the Internet to map IP addresses to Fully Qualified Domain Names. As the Internet continuously grows, there have been challenges in keeping up with the demands on the DNS service. Recently, ICANN announced the introduction of new generic Top-level Domains (TLDs). Using packet traces collected from a large edge network, we analyse the usage of TLDs between 2008 and 2015. We observe changes in usage of TLDs, and analyse the adoption of the new generic TLDs announced by ICANN. We find that while there were no changes in the appearance of most frequently used TLDs, the presence of the new generic TLDs in the datasets is growing. The number of different late new gTLDs appearing also doubled or tripled each year, implying that more and more people are starting to use these new gTLDs.

12:30 PM - 1:30 PM

Lunch Break

1:30 PM - 2:30 PM

Keynote Speaker: Nur Zincir-Heywood, Dalhousie University, Canada

Title: "Lessons learned on analyzing one-way network traffic"

Abstract: The rapid growth in size, and complexity of computer systems, and networks make network measurement and monitoring tasks increasingly important, and extremely challenging. The rate at which digital data is accumulated from such systems far exceeds the human ability to make sense of it, i.e., to interpret and attach meaning to it. Major challenges include data volume and diversity, data dependency, and system dynamics. The goal of this talk is to explore some of these challenges in the context of one-way traffic, while presenting research taken to mitigate the impact of these factors using (semi)automatic and intelligent systems.

Nur Zincir-Heywood

Speaker Bio: Dr. Nur Zincir-Heywood is currently a Professor of Computer Science at Dalhousie University, Canada. She received her PhD in 1998 in Computer Science and Engineering from Ege University, Turkey. Prior to moving to Dalhousie in 2000, Dr. Zincir-Heywood had been a researcher at Sussex University, UK and Karlsruhe University, Germany as well as working as an instructor at the Internet Society Network Management workshops. She has published over 150 papers on networks, security and intelligent systems. She has substantial experience of industrial research in autonomous systems, network management and security related topics with partners such as RUAG, Raytheon, Gtech, Sandvine and Public Safety Canada. She is a member of the IEEE and the ACM.

2:30 Hybrid Community-Based Forwarding: A Complete Energy Efficient Algorithm for Pocket Switched Networks

Khadija Rasul, Dwight Makaroff and Kevin G Stanley (University of Saskatchewan, Canada) Sensor devices and the emergent networks that they enable are capable of transmitting information between data sources and a permanent data sink. Since these devices have low-power and intermittent connectivity, latency of delivery for certain classes of data may be tolerated in an effort to save energy. Several previously developed algorithms employ models which considers the popularity of individual nodes within communities and forward messages to nodes with higher probability of delivery according to some heuristic. In previous work, we developed Community-Based-Forwarding (CBF) that considers the interactions between communities as a factor in message forwarding. Using this information, CBF is able to exploit intermediate connections between clusters to route messages with more balanced node participation and higher levels of reliability and efficiency. One disadvantage of CBF was an increased delivery latency for some subset of messages that could not be delivered using other algorithms. In this paper, we extend the semantics of CBF with the Hybrid CBF algorithm (HCBF) by optimizing forwarding inside communities by considering the social diversity (measured by Unique Interactions). We find that all performance metrics are improved with this heuristic on a representative set of human mobility traces, but most significantly the message delivery latency is substantially improved over the other algorithms studied.