AIMS 2014: Workshop on Active Internet Measurements : Talk Abstracts

This page contains names, talk abstracts (if presenting), and topics the the participants are interested in discussing, as well as any related URLs. Participants are encouraged to read these ahead of time to anticipate workshop discussion.

Participant Abstracts

Jim Warner (UCSC) Talk Title: Wireless carrier claims v customer performance

Talk Abstract: Cellular carriers put a lot of effort and advertising dollars into promoting the bona fides of their networks. All carriers offer coverage maps -- most of which come with disclaimers about their accuracy. Carriers also submit performance data to both the FCC and the the State PUCs. Some of the coverage claims are based on models that use topography, antenna location and antenna height to predict expected radio performance.

I did a survey to test Verizon's coverage claims for data performance using the Ookla application to measure delivered speed at approximately 700 points within a 5 square mile section of Santa Cruz containing six known cell sites. The results indicate carriers have unwarranted optimism about network performance. I will show Verizon's claims and my empirical results.

Interested in Discussing: indoor data coverage.

Sanae Rosen (University of Michigan) Talk Title: Measuring performance impacts of RRC state transitions in cellular networks with user devices

Talk Abstract: Devices on cellular networks transition between different RRC states in response to network traffic, with different performance and power trade-offs. When transmitting data, devices are in a high-power state, and then demote to lower- power, higher-latency states according to carrier-specific timers. These timers can have a substantial impact on application performance and power consumption, and there has been a great deal of interest in recent years in understanding how applications can design their network communication to account for these timers.

Previous work has explored measuring RRC states in a controlled setting. However, to understand the impact of these states on user-perceived performance in a wide range of contexts --- such as with a variety of device types, carriers, geographic locations, or network conditions --- a more general approach is needed. We have developed a tool for automatically inferring RRC state configurations on end-user devices, without requiring any special application privileges or expert knowledge. This allows us to deploy our RRC inference test on devices around the world, as well as easily track changes and trends over time. In this talk, I will discuss the challenges of extracting this information from arbitrary devices, as well as extending this technique to measure very short timers that have not yet been addressed, such as those related to Discontinuous Reception (DRX) in LTE.

Interested in Discussing: Mobile network measurements, collecting measurements on user smartphones

William Lehr (MIT) Talk Title: Mobile Broadband and Wireless Measurement Infrastructure: Policy Challenges and Opportunities

Talk Abstract: The growth of mobile broadband poses dual challenges for policymakers seeking to promote the health and quality of BB services (increasingly viewed as essential infrastructure) and the efficient use of scarce radio frequency spectrum (demand for which is intensified by the growth of mobile BB). For informed policymaking with respect to both mandates, the FCC and other market participants need expanded access to real-time measurement data relating to the performance of mobile applications which typically are provided over wireless access. To address this challenge, we will need to create a multitiered measurement infrastructure, a significant component of which will involve highly distributed, end- user/edge-based measurement infrastructure (e.g., crowd-sourced, likely resident on handset or other mobile devices). This will be complemented by fixed and centralized control measurement infrastructure (e.g., SamKnows-like initiatives). The structure and relationship of these complementary measurement infrastructures will offer opportunities for supporting new regulatory models (e.g., enforcement of Mobile BB SLAs, enforcement of spectrum sharing regimes) and will induce strategic behavior (e.g., tussles over metric definitions, public reporting/data access, and reliance of measurements in standardization and regulatory decision-making). At an R&D level, it will require integrating perspectives from the heretofore divergent engineering communities engaged in Internet measurement and (wireless) spectrum sensing.

The goal of this talk will be to provoke discussion regarding the policy/institutional challenges associated with building/deploying/managing the requisite measurement infrastructure, which requires integrating Internet and wireless measurement tools and platforms. I am envisioning this as a framing talk for a session that would focus on higher-level economic/policy goals of implementing a public mobile measurement infrastructure.

Interested in Discussing: Multidisciplinary economic/policy challenges of deploying/managing/using (for market and policy decision making) a distributed mobile/wireless measurement platform.

Walter Johnston (FCC) Talk Title: FCC Mobile Measuring Broadband America Program

Talk Abstract: In 2010, the Federal Communications Commission initiated an effort to collect broadband performance data on fixed broadband services for the major Internet Service Providers in the United States. It has published four reports since then on the performance of broadband ISPS serving over eighty percent of broadband subscribers in the United States. In addition to a published report, the FCC provides a spreadsheet containing a set of relevant statistical data as well as all the underlying data used in the report. In addition, the FCC periodically posts all data collected for each month of the program both in the interests of transparency and for further use by the research community.

In 2013, the FCC extended its program to include the collection of mobile data using a crowd sourced approach based on smartphones. The FCC released in November of 2013, its FCC Speed Test App to Google Play for Android phones. The app allows both manual and automatic testing. The automatic generation of tests is important to the integrity of both the fixed and mobile programs as it eliminates user bias and a common concern with manual tests is that they are often used as a diagnostic and therefore skew towards abnormal network conditions. Test data collected includes performance data such as upload and download speeds, latency and packet loss. In addition, the Android platform allows the collection of network data including cell tower ID, serving carrier, RF signal strength and bearer channel (3G, 4G, Wi-Fi, etc.) as well as phone data including OS version and phone model type. Since testing is automatic and focused on the data connection currently in use, WiFi performance data is also collected as part of the program. An iPhone app is available shortly. Both apps are released as open source as part of our Open Data program.

The collection of mobile data presented distinct challenges from that of our fixed broadband program. Foremost was a concern for consumer privacy. It was a stated goal at the outset of the program not to collect any personally identifiable information (PII). However, as many researchers have shown, the ability to generate a trace for a unique phone can allow one to infer identity. The FCC worked with the Federal Trade Commission, external academic experts, the major wireless carriers cooperating in this study and other interested stakeholders to develop a privacy policy protecting the interests of the consumer. An outcome of this policy was to ensure the phones are fully anonymized, with no uniquely identifying information collected. Other procedures were developed to further protect the consumer.

In addition consistent with best practices, Consumer Terms and Conditions were developed to fully inform the consumer regarding the type of information collected and how it will be used. The mobile MBA program, like the fixed MBA program is founded on principles of Open Data. However, data that is released will be conformant with our privacy policy meaning that specific data elements may be aggregated (e.g. location) or otherwise obfuscated (rounding time) to ensure consumer protection. However, data released will be consistent with any reports published by the FCC which will be based on the transformed data. While mobile broadband performance data is commonly collected for commercial purposes, we believe this may be the first open data initiative in this area and we believe it will provide a useful and evolving data set for the consumer, government and researchers. Furthermore our work in this area leads us to believe that there should be an effort to standardize on not only performance measurements but also include discussions of privacy and informed consumer consent. The talk will describe the evolution of the program, issues and its future evolution.

Interested in Discussing: What responsibilities should exist in terms of privacy protection and informed consumer consent in the collection of mobile broadband data? How should implementable policy be best developed in this area?

David Clark (MIT) Talk Title: N/A

Talk Abstract: The prior work of my group on edge measurement was concerned with calibrating tools to measure wireline access systems. My goal for this meeting is to understand the state of the art in methods that can be used to measure the performance of wireless access systems, given the increased complexity of the technology, including signalling and access protocols. Do we need to do tests that mimic specific applications to understand how the details of the technology translate into relevant QoE outcomes?

Interested in Discussing: Do we know enough about measurement approaches that we can tease apart issues in the air protocols from the provisioning of the access backhaul and other aspects of the access system? Do we have models of the access systems that allow us to understand how measurements tell us about the components of the system, or are we measuring the system as a black box?

David Choffnes (Northeastern University) Talk Title: Measuring Mobile Networks with Mobilyzer and Meddle

Talk Abstract: I can give talks that give an overview of two separate relevant projects.

1) Mobilyzer is an open platform for network measurement from mobile devices. It currently provides researchers and developers a single, comprehensive codebase for issuing measurements and gathering contextual information (e.g., GPS location, signal strength and radio state) to interpret measurements. As part of our future work, we will provide a unified experiment management system that uses coordination, prediction and fairness principles to efficiently and evenly distribute measurement requests to devices running measurement code, and a public datastore for publishing measurement results. We argue that this approach not only supports existing mobile network measurement systems, but provides an environment that enables novel, impactful experiments. (Joint work with Morley Mao)

2) To understand and address the problems in mobile systems as they evolve over time, we need an approach that supports long-term studies of mobile Internet traffic and the ability to interpose on these flows. Ideally, such a system would be easy to deploy and use for a typical smartphone user running any operating system anywhere in the world. I present Meddle: a platform for measuring and interposing on all mobile-device Internet traffic (e.g., from smartphones and tablets). Meddle combines software middleboxes with VPN proxying, enabling both visibility and control over network flows. These features combine powerfully to enable new measurement tools. As examples, we have made available a visualization tool, ReCon, for users to track and control how they are being monitored by ad and analytics services. We also develop tools for detecting content manipulation by ISPs (Web Tripnets) and service differentiation (Mobile Replay, joint work with Phillipa Gill) in the mobile environment, building upon previous work in these areas.

Interested in Discussing: Mobile measurement tools, analysis, techniques. Evolution of performance, how to quantify performance. Also interested in turning this data into something that is easy to digest for users and policymakers.

Valerio Luconi (University of Pisa, Department of Information Engineering) Talk Title: Portolan: a Smartphone-based Crowdsourcing System for Network Sensing

Talk Abstract: Portolan is a smartphone-based crowdsourcing system aimed at sensing both wired and wireless networks. Portolan is a multi-purpose measurement system, that is able to characterise multiple properties of large-scale networks. Currently, Portolan includes measurement capabilities for building Internet topology maps at the Autonomous System (AS) level of abstraction, mapping the signal coverage of cellular networks, detecting traffic differentiation, and measuring the maximum instant throughput on a user's network. This presentation describes the Portolan system architecture and capabilities, as well as the motivations that led to Portolan's conception.

Interested in Discussing: Internet AS-level topology discovery

Ann Cox (DHS) Talk Title: What is the appropriate role of the federal government in the mobile measurement space?

Talk Abstract: What is the appropriate role for federal research and development in the mobile measurement space? Please consider the following questions:

  1. What data, adoption issues, availability, or other concerns should the federal government be tracking in the mobile measurement space?
  2. What measurements contain this data?
  3. Privacy concerns?
  4. Should there be a separate research effort directed at mobile measurement?

Craig Partridge (Raytheon BBN Technologies) Talk Title: Wireless Network Measurement in a World of Software Defined Radios

Talk Abstract: Software defined radios (SDRs) are wireless devices with real-time programmable antennas, filters, digital process devices (DSPs or FPGAs) and processors. These radios are designed to utilize available wireless spectrum by changing their behavior (and protocols) in accordance with both the needs of the applications on the radio and the spectrum conditions that currently obtain in the radio's environment.

A challenge with SDRs is how to measure their behavior, particularly from an external vantage point. Consider that the radios can change protocols and frequencies at will. How do you know where to listen? How do you know what protocols to look for? And, given the radio can change its external behavior, how do you know if the radio running Bluetooth at time t is the same or different from the radio running LTE at time t+x?

I don't claim to have the answers to these questions. My goal is to raise them and stimulate some thinking. In that spirit, I provide a couple of initial thoughts below.

Interested in Discussing: I'm most interested in problems of doing useful measurements for spectrum agile devices

Related URLs:

Sharad Agarwal (Microsoft Research) Talk Title: It's not just latency & throughput. Understanding mobile app performance is hard

Talk Abstract: When measuring a wireless network's performance, a number of metrics should be collected, including signal strength, latency, loss, throughput. While this is useful, it is very hard to translate this into what performance the user sees. Mobile apps are very complex. The underlying OS's network stack interacts with the network in complex ways. The mobile network can do unusual things to your traffic, that may make your app faster or slower.

Interested in Discussing: How network performance translates to mobile app and mobile web performance

Related URLs:

  • Switchboard (MobiSys'11) - background on how latency varies across different celltowers and how (un)predictable that is at short timescales
  • AppInsight (OSDI'12) - background on the complexity of modern mobile apps and how hard it is to understand performance and the different factors that affect it
  • Procrastinator (HotNets'13 and MobiSys'14) - background on how much networking activity there is in apps and how that can / cannot impact app behavior
Bendert Zevenbergen (Oxford Internet Institute) Talk Title: Ethical privacy framework for network researchers collecting, measuring, analysing and disseminating mobile connectivity dat a

Talk Abstract: Network researchers measure Internet connections worldwide and generate vast amounts of useful data for consumers, policy makers and academia on the state and structure of the Internet. With the increasing uptake and importance of mobile devices that connect to the Internet, network researchers are looking to expand their measurement capabilities to accurately gather mobile connectivity data. The additional types of information that can be collected via mobile phones, such as location data or more specific user data, enrich the research datasets.

Data on individuals' Internet behaviour will frequently contain sensitive information about the data subjects' lives. On the other hand, data that only reveal an Internet users' connection to a given point on the network is not necessarily privacy invasive. Although computer scientist consider privacy an important value in their research design, we found that the privacy precautions taken are often misleading or meaningless in practice, as many research datasets can be re-identified through various techniques or dissemination restrictions are not enforced.

The Oxford Internet Institute developed an ethical guideline, which aims to help network researchers worldwide to navigate the challenges of preserving adequate privacy of data subjects while ensuring datasets can be published and disseminated to adhere to - and advance - good scientific practice. The guidelines build on previous work done in this area and a one-day discussion workshop attended by lawyers and computer scientists held in Oxford in June 2013.

The potentially global nature of this type of research requires the guidelines to look beyond the current law in the books, because national privacy and data protection laws vary significantly. Starting from the international human rights framework as a base, the guidelines draw inspiration from several existing legal concepts such as data minimisation, purpose limitation, privacy by design and informed consent.

From the technical perspective, the guidelines aim to educate the mobile connectivity researcher how to think about available techniques and best practices for data collection, processing and dissemination in the context of their overall research design. Further, researchers are required to consider what the different data types reveal about data subjects when collected together or linked to other databases and assess the risks involved. More specifically, the guidelines assess the effectiveness of so-called anonymization techniques and give guidance on the appropriateness of use of these tools.

The researcher is taken through the standard steps of a research design based on assistive questions and detailed explanations in an extensive annex. An iterative process is encouraged, whereby the researcher consults colleagues, an ethical board or a legal expert in the research design process and records all the considerations and precautions of the research design. This will form a specialised Privacy Impact Assessment (PIA) for the perceived research project.

Find the guidelines here:

Interested in Discussing: Technology impact assessments in this area.

Related URLs:

Ahmed Elmokashfi (Simula Research Laboratory ) Talk Title: Mobile Broadband reliability: a user perspective

Talk Abstract: Mobile Broadband reliability is a broad notion encompassing several stability and performance related metrics. Hence, it is unclear how we can assess the overall reliability of a network by using user-initiated measurements. These tests may not capture all related metrics risking leaving some aspects of reliability uncovered. In this work, we propose a framework for measuring user- experienced reliability in Mobile Broadband networks. We argue that reliability must be assessed at multiple levels ranging from the ability to connect to the network to the stability of applications. We employ our framework to assess the reliability of five mobile broadband networks in Norway. Our study uses a large number of measurement nodes spread all over Norway and spans a period of six months. More specifically, we assess the reliability at three levels. First, at the connection-level, by measuring the ability of user equipment to attach reliably to the mobile broadband network. Second, at the data plane level, by investigating the ability of user equipment to have well functioning end- to-end connectivity to the Internet. Third, at the application level, where we assess the stability of two types of applications; HTTP download using curl and Voice over IP (VoIP) using SIP/RTP. Our results reveal clear differences between operators in the stability at all levels. They also indicate that connection-level reliability is highly dependent on the Radio Access Network. Further, the inability to get the requested radio resources is the main reason behind poor applications reliability. In addition, our study indicates that independent measurements can help operators identifying problems in their networks that are hard to spot using network-side monitoring. By correlating bad performance in connections from the same operator we reveal many events with a large scope of impact. Interestingly, many of these events were not captured by operators' network monitoring systems.

Interested in Discussing:

  • Measuring mobile broadband networks in general and from the edge in particular.
  • Incentives to deploy tools.
  • Research infrastructure to support mobile/wireless measurement

Xing Xu (University of Southern California) Talk Title: Investigating Performance Enhancing Proxies in Cellular Networks

Talk Abstract: The use of cellular data networks is increasingly popular as network coverage becomes more ubiquitous and many diverse user contributed mobile applications become available. In this talk, we investigate one techniques used by current cellular carriers: Performance Enhancing Proxy (PEP). We find that such split connections will only be used for HTTP traffic. Based on that, we investigate the benefits of such splitting. From HTTP layer point of view, such splitting allows mobile carrier to manipulate the packets and thus to support caching, redirecting, connection reusing, and potentially compression. More interestingly, from the performance perspective, we find that for downloading a relatively large file, split connection downloads faster than traditional non-split connection. However, for smaller files, which are the most common cases of current mobile traffic, we verify that the split connection performs even worse than traditional non-splitting cases, due to all kinds of cost introduced for splitting.

abstract PDF link: or contact author

Interested in Discussing: mobile network, measurement

Ethan Katz-Bassett (University of Southern California) Talk Title: Initial Measurements of the Internet from Mobile Devices

Talk Abstract: Dave Choffnes plans to present his mobile measurement library, which I am a limited collaborator on. The vision is for a common library that can be incorporated into multiple applications, providing easy incorporation of best practices for measurement, standardization of measurements, and coverage across time and users beyond what a single application can capture.

I can present initial results from some studies we have done, showing:

  • some interesting aspects of the Internet that affect mobile performance
  • how currently limited coverage across time and users restricts current results
Together, these observations provide motivation for the measurement library, for future studies, and for longitudinal studies. (However, I don't need to present if there is not sufficient interest)

The 2 studies I can present are:

  1. Our forthcoming PAM study of causes of path inflation from mobile users to Internet services: As mobile Internet becomes more popular, carriers and content providers must engineer their topologies, routing configurations, and server deployments to maintain good performance for users of mobile devices. Understanding the impact of Internet topology and routing on mobile users requires broad, longitudinal network measurements conducted from mobile devices. In this work, we are the first to use such a view to quantify and understand the causes of geographically circuitous routes from mobile clients using 1.5 years of measurements from devices on 4 US carriers. We identify the key elements that can affect the Internet routes taken by traffic from mobile users (client location, server locations, carrier topology, carrier/content- provider peering). We then develop a methodology to diagnose the specific cause for inflated routes. Although we observe that the evolution of some carrier networks improves performance in some regions, we also observe many clients - even in major metropolitan areas - that continue to take geographically circuitous routes to content providers, due to limitations in the current topologies.

    Key takeaways of that paper include:

    • The number of GGSNs/ingresses into the public Internet is expanding, allowing much more direct routes than previously reported by Xu et al. [SIGMETRICS 2011]. The expansion shows that we need ongoing studies to stay up to date.
    • However, various routing inefficiencies still exist.
    • Because our app was only available to Google employees, our coverage was limited in certain areas and degraded over time as employees received new phones. The common measurement library that Dave proposes could help coverage in the future.
    • Our dataset is available to the community.

  2. Our ongoing work on the impact on mobile Web performance of middleboxes in cellular networks. One of the students submitted an abstract on this:

    I hope that he can present that work if there is a student session. However, I could also present a short version of it as part of this broader talk motivating Dave's measurement library and showing what aspects of cellular networks are interesting to measure. I find two aspects of the project interesting:

    • What are middleboxes doing to traffic, and how does it help/hurt performance?
    • How can we take what we learn in tightly controlled experiments, in which we control both the phone and the server (as in the student's extended abstract) and extend the results out to broader coverage, using apps such as the ones that Dave is proposing?

Interested in Discussing: How to measure Internet performance and performance problems on a wide scale, and how to create incentives for users to contribute measurements towards this goal

Abbas Razaghpanah (Stony Brook University) Talk Title: Identifying Traffic Differentiation on Cellular Data Networks

Talk Abstract: As wireless providers offer faster and more widely available cellular data to their customers, more mobile applications emerge that utilize this additional capacity to add more services to the mobile platform or improve on the quality of service for existing services. However, it is often unclear how traffic from mobile services is treated while traversing these networks. The goal of this research is to detect traffic differentiation in cellular data networks. Traffic differentiation is defined as tampering with the performance of the network done by the ISP in any shape or form. Extended abstract available here:

Interested in Discussing: network neutrality, internet censorship in mobile networks

Ashkan Nikravesh (University of Michigan) Talk Title: Toward Meaningful Mobile Network Performance Measurements

Talk Abstract: In cellular networks, end user network performance is not as stable as wired broadband network and a particularly difficult challenge in this environment is to capture a view of network performance that is representative of these conditions. A better understanding of the the impact of factors such as access technology, geographic location, and time can help cellular providers improve performance for their subscribers. To study the characteristics of such networks and identify the factors causing performance degradation, a longitudinal, continuous and large-scale monitoring that samples a variety of devices across carriers, access technologies, locations and over time is needed...

rest of 1 page abstract:

Interested in Discussing: What are the challenges of mobile performance measurement? How to design an effective measurement tool that can help (1) identify anomalies and (2) diagnose the reason behind the observed anomalies and performance degradations.

Hongyi Yao (RobustNet Group, Univerisity of Michigan) Talk Title: Mobilyzer: A Network Measurement Library for Android Platform

Talk Abstract: Mobile Internet availability, performance and reliability have remained stubbornly opaque since the rise of cellular data access. Ideally, we would be able to collect measurement data from any mobile device on any network at any point in time. Using this information, users can evaluate the service they are paying for, carriers can study the factors impacting the performance and detect the problems causing degradation, and application developers can tune and improve their service.

Toward this goal, the last few years has seen the growth of a cottage industry of apps that measure mobile Internet performance, reliability and evolution. Each of these apps captures a small sliver of the big picture, resulting in at best patchwork coverage of mobile networks worldwide. The methodologies, data formats and data access proliferate in non-uniform ways, making it difficult to compare, combine and generalize the myriad research efforts. In lieu of this approach, we propose a unified system for issuing network measurements, collecting results and openly publishing them.

As a first step toward this goal, we have built Mobilyzer, a network measurement library developed for Android platform. It provides app developers with a flexible and easy-to-use suite of standard measurement tools that manages network measurements and data collection. Measurements can be scheduled either directly from the app, or from a cloud service. For researchers, the incentive for developing apps with Mobilyzer is that participants can submit measurements to any device running any app that includes the Mobilyzer library (analogous to the RIPE Atlas incentive model). Mobilyzer offers an easy to use API and a scheduler which ensures both efficiency and fairness as well as solving the interference problem when multiple applications with Mobilyzer are running on the same device. The following paragraphs describe our current design and implementation.

Mobilyzer is mainly focused on active measurements. It supports DNS Lookup, HTTP Get, Ping, Traceroute, TCP Throughput and UDP burst test, which cover the basic network delay and bandwidth measurements of general interest. In addition, Mobilyzer also contains complex measurements like RRC state inference. It has been written in a modular way to support easy addition of new measurement tasks. To satisfy common dependencies across measurement, it also offers two kinds of task combination mode: the parallel mode and sequential mode, to generate more user-specific measurement types.

(See for full paper)

Interested in Discussing: mobile/wireless measurement capabilities (tools), incentives to deploy tools

Ranveer Chandra (Microsoft Research) Talk Title: Quantifying Spectrum Utilization Using the Microsoft Spectrum Observatory

Talk Abstract: We have been measuring spectrum utilization over the years, first as part of the TV white spaces project, and now to study utilization of other spectrum bands, so that we can opportunistically use unutilized spectrum for data communication. In this talk, I will describe the Microsoft Spectrum Observatory, and present some research questions that we are answering as part of this effort.

Related URLs:

David Reed (University of Colorado Boulder) Talk Title: Implications of Wi-Fi as a Commercial Service

Talk Abstract: While Wi-Fi has enjoyed explosive growth and deployment for use in residential homes, the rollout of commercial Wi-Fi service has been more limited due in large part to the strategic concern that the commons model to spectrum management lacks the incentives for service providers to invest due to the limited ability to manage interference in the unlicensed band. My hypothesis is that this situation appears to have changed now due to the confidence service providers are placing in the activities of the Wi-Fi Alliance and IEEE 802.11 standards body. The talk will show how these groups essentially replicate many, but not all, of the functions traditionally employed by an effective band manager that is optimizing efficiency on a licensed spectrum block more typically associated with the deployment of commercial services. Consequently, with the Wi-Fi ecosystem functioning as an effective spectrum manager, the service provider investment in Public Wi-Fi networks is rational and the risk posed by saturation is reduced to a manageable level. The strategic implications of this finding is that the requirements from service providers are already significantly influencing the evolution of the Wi-Fi standard. These attempt to address the risks and liabilities associated with the unlicensed spectrum management model and potentially the usage models of consumers. Service providers increasingly need functionality in Wi-Fi technology to manage interference, and monitor and improve network performance. I will discuss some of the current ideas under discussion for the next version of Wi-Fi to support both commercial Wi-Fi requirements. I will address the interference concerns, but only up to a point as the unlicensed model intrinsically leaves some risk to participants of spectrum saturation through overuse.

Interested in Discussing: The proposed talk above is based on a TPRC paper presented this past fall (see co-authored with Jim Lansford. I will present some new material on additional trends in new technologies proposed in Wi-Fi by service providers.

If the above proposed topic is not accepted for the conference, I am interested in the spectrum usage monitoring topic in general and would be happy to participate as a discussant or report on this breakout session is some way.

James Martin (Clemson University) Talk Title: SciWiNet: A large scale wireless testbed for the academic community

Talk Abstract: The academic research community is extremely diverse with respect to wireless infrastructure requirements. Wireless network research has long utilized experimental wireless networks. However, traditionally these systems have been limited to wireless systems such as 802.11, Bluetooth or Zigbee that utilize unlicensed spectrum. Further, these testbeds are generally limited in scale in terms of geographic footprint as well as in terms of diversity of radio access and device technology. Recent commercial adoption of advanced 4G networks now makes commodity 3G/4G wireless networks of interest both to the wireless research community as well as to the broader research community. Widely available broadband wireless data connectivity is enabling research in application domain areas such as intelligent transportation systems, smart structures, the SmartGrid, environmental and agricultural sensing to name a few. Non-engineering research areas such as education, public health, social science can (and are) making use of 3G/4G data connectivity. .

SciWiNet was inspired by these trends. The goal of SciWiNet is to build wireless infrastructure that supports current and emerging wireless requirements of the academic research community. Through benefits of scale, SciWiNet will enhance the ability of the research community to integrate 3G/4G wireless in their respective research.

In this talk, we overview the SciWiNet project. An important component of the research infrastructure is the performance/coverage data collection support and visualization capabilities. Our talk will demonstrate these features.

Interested in Discussing: As there are many efforts underway that fall under the scope of wireless research, it would be helpful to connect the community in some way. I would guess that a workshop will at best provide a 'sampling' of synergistic efforts. How can we address this?

Aaron Schulman (Stanford University) Talk Title: PlanX: Enabling Innovative Measurements of Operational Wireless Networks

Talk Abstract: In this talk we will describe a programmable wireless network architecture that enables researchers to perform detailed measurements of operational wireless networks. Recent measurement studies with Software Defined Radios (SDR) indicate that programmability enables new detailed measurements of the PHY and MAC layers. These measurements have explained mysterious faults and driven protocol improvement. Currently measurements of operational wireless networks only are limited to the statistics collected by the wireless network's Application Specific Integrated Circuits (ASIC).

As of yet, there is no way to perform these innovative SDR-based measurements in operational wireless networks. To bridge the operational gap, we envision a deployment of wireless base stations built with PlanX -- our software framework for programming multi-core Digital Signal Processors. Multi-core Digital Signal Processors (DSP) are like ASICs, efficient and inexpensive. With PlanX, researchers can deploy the measurements they developed with SDRs on operational wireless networks.

Interested in Discussing: Programmability of wireless networks and smartphone energy consumption

Nick Feamster (Georgia Tech) Talk Title: WTF? Locating Performance Problems in Home Networks

Talk Abstract: Most users of home networks have experienced the intense frustration that comes with diagnosing poor performance. Even determining something as simple as whether a performance problem lies with the ISP or somewhere in the home network is incredibly difficult; this lack of visibility results in unnecessary service calls to ISPs and a general inability to have the network perform as well as it should. In this paper, we design and develop WTF (Where~Rs The Fault?), a system that reliably determines whether a performance problem lies with the user~Rs ISP or inside the home network. The tool can also distinguish these problematic situations from the benign case when the network is simply under-utilized. WTF uses cross-layer techniques to discover signatures of various pathologies. We implemented WTF in an off-the-shelf home router; evaluated the techniques in controlled lab experiments under a variety of operating conditions; validated it in real homes where we can directly observe the home conditions and network setup; and deployed it in 64 home networks across 15 countries. The real-world deployment sheds light on common pathologies that occur in home networks. We find, for instance, that many users purchase fast access links but experience significant (and frequent) performance bottlenecks in their home wireless network. In collaboration with SamKnows and the FCC, we are developing a version of WTF that can be deployed on the SamKnows "whiteboxes" as part of a new in-home study program sponsored by the FCC. In addition to the WTF algorithm and our preliminary results from a deployment of WTF on the BISmark testbed, I will describe the current state of the WTF deployment on the SamKnows hardware and our experiences as part of the FCC's pioneering efforts to support more research on this unique tested. This project is joint work with Srikanth Sundaresan, Renata Teixeira, and others.

Interested in Discussing: Mobile measurement in developing regions.

Mike Wittie (Montana State University) Talk Title: MITATE: Mobile Internet Testbed for Application Traffic Experimentation

Talk Abstract: This talk described a Mobile Internet Testbed for Application Traffic Experimentation (MITATE). MITATE is the first programmable testbed to support the prototyping of application communications between mobiles and cloud datacenters. We describe novel solutions to device security and resource sharing behind MITATE. Finally, we show how MITATE can answer network performance questions crucial to mobile application design.

Interested in Discussing: How can interactive applications prosper in mobile networks with aggressive traffic shaping policies?

Michael Piatek (Google) Talk Title: What should researchers work on?

Talk Abstract: Not provided.

Interested in Discussing: mobile

Narseo Vallina Rodriguez (ICSI, UC Berkeley) Talk Title: Netalyzr for Android: challenges and opportunities

Talk Abstract: The popular ICSI Netalyzr has been used by some 1.1 million users around the world to perform network troubleshooting and diagnose from the edge of the network since its release in 2009. Its comprehensive test suite allowed many users to detect potential problems with their network that may affect and limit their Internet experience.

In order to broaden Netalyzr's reach to cellular networks, we have recently implemented an Android version of the tool. Released in November 2013, some 15,000 users have installed the app to date. The app is available for free in Google Play at

In this talk, we'll describe the system design of the mobile app, stressing how we have adapted and extended the test suite for the mobile environment, and report on related challenges we are facing in its development, deployment, and data analysis. Interesting factors here include handset diversity, network misconfigurations, the best-effort nature of cellular networks, and their overall complexity.

Based on ~3,000 cellular sessions recorded between November and January covering some 300 mobile carriers in 80 countries, we will present preliminary results, including handset capabilities, network deployment of cellular networks, and DNS performance.

Finally, we will report on challenges extending to any active measurement effort on mobile networks, conducted from the handset: result bias, users' incentivization, privacy considerations, app support, mobile app markets coverage, as well as future research directions.

Interested in Discussing: Mobile measurements, privacy, network performance, DNS, DNSSEC, topology

Jacobus (Kobus) Van der Merwe (University of Utah) Talk Title: PhantomNet: An end-to-end mobile network testbed

Talk Abstract: In this talk I will describe our ongoing work to create a testbed, called PhantomNet, to enable research at the intersection of mobile networking, cloud computing and software defined networking. PhantomNet is being developed as an extension of the Emulab testbed at the University of Utah and, like Emulab, will be remotely accessible to researchers. Using long term evolution (LTE) mobile devices and small cell technology for the radio access network (RAN) and a software implementation of evolved packet core (EPC) functionality, PhantomNet will, as a baseline, allow end-to-end LTE/EPC experiments to be created. To enable larger scale experiments, RAN emulation (mobile devices and LTE base stations) will be provided.

Interested in Discussing: Use mobile testbeds.

Justin Cappos (New York University) Talk Title: Programmatically Deploying Code on End User Devices: Seattle and Sensibility Testbed

Talk Abstract: While many researchers have measured networks from the inside out, measurement from end users have been harder to come by. Researchers typically have performed measurements by asking users to run tracing programs on edge devices. However, this loses some of the flexibility of platforms like PlanetLab where researchers can re-deploy their code interactively.

This work talks about our experiences building a large scale sensing platform that allows researchers to programmatically deploy code. Our Seattle platform runs on tens of thousands of laptops, smartphones, tablets, and similar devices. Any experimenter can use Seattle to push safe, sandboxed code to end computers. Seattle has been used by more than three thousands of participants around the world and has been used to map a variety of network effects.

Our Seattle platform is currently being extended in multiple ways by different groups, including as part of the Sensibility Testbed. Sensibility Testbed provides access to sensors on user owned Android devices, assuming the researchers pass IRB approval. This allows researchers to read sensor data (such as wireless network data) from participants around the world. This early stage work has already been used to map wireless network technologies around Vienna.

Interested in Discussing:

  • What sort of facilities do researchers need on end user devices?
  • Given issues like smartphones which do not provide low level network access, what are you willing to do without?
  • If you could only choose one, do you want a huge number of networks with only TCP / UDP or a few with essentially complete measurements?

Related URLs:

Vijayarangam Subramanian (National Telecommunication Information Administration) Talk Title: Wireless usage measurement data: Research needs and support for national policy and rule making

Talk Abstract: Spectrum crunch can have a significant impact on the economic and security aspects of the nation. There have been several initiatives to address this issue for the industry, and to promote efficient spectrum utilization. From a technology perspective spectrum sharing across the government and industry users, among other possibilities, is viewed as important and necessary. This new paradigm has immense complexities involved in technology development, testing, data usage measurement, analytics, deployment, usage enforcement and security. This talk will focus on various government supported R&D program initiatives in addressing some of these aspects, and why spectrum usage measurement and analytics is critical to national policy and rule making development.

Interested in Discussing: All topics in the current agenda are interesting and important.

Related URLs:

Jeffrey Cubbal (DHS)

Interested in Discussing: I'm interested in Dr. Cox's talk about Government role in measurement.

Hiroyuki Ishii (DOCOMO Innovations, Inc.)

Interested in Discussing: Measurements/ Big Data Analysis in Cellular Network

Erin Kenneally (CAIDA/UCSD)

Interested in Discussing: Data sharing policy

Related URLs:

Last Modified