Workshop Summary Report

Workshop summary now posted! You can choose the full version or the brief version.

NSF Workshop on Mobile Community Measurement Infrastructure Agenda



8:45-10:15 (90min)

Session A: Measurements and application needs



11:00-12:15 (75min)

Panel A: Data Collection & Privacy



1:30-2:45 (90min)

Panel B: Operational challenges



3:15-4:30 (75min)

Session B: Testbeds



SESSION A: Measurements and application needs
This session focuses on (among many topics) identifying key network functionality to monitor and instrument, how to support measurement from all layers of the stack, from spectrum to base stations, protocol behavior to application behavior, and how to support programmable measurements in the mobile network infrastructure, e.g., eNodeB.


  • Rangam Subramanian (NTIA)
  • Federal Initiatives for Wireless Innovation and Measurements (Slides)
  • Ranveer Chandra (MSR)
  • Spectrum observatory (Slides)
  • Aaron Schulman (Stanford)
  • Energy measurement, programmable cellular base stations to support physical layer measurements. (Slides)
  • Lili Qiu (UT)
  • Robust network compressive sensing (Slides)
  • A major challenge to enable effective network analytics is the presence of missing data, measurement errors, and anomalies. To address these issues, we develop LENS decomposition, a novel technique to accurately decompose a network matrix into a low-rank matrix, a sparse anomaly matrix, an error matrix, and a small noise matrix. LENS has the following nice properties: (i) it is general: it can effectively support matrices with or without anomalies, and having low-rank or not, (ii) its parameters are self tuned so that it can adapt to different types of data, (iii) it is accurate by incorporating domain knowledge, such as temporal locality, spatial locality, and initial estimate (e.g., obtained from models), (iv) it is versatile and can support many applications including missing value interpolation, prediction, and anomaly detection. We apply LENS to a wide range of network matrices from 3G, WiFi, mesh, sensor networks, and the Internet. Our results show that LENS significantly out-performs state-of-the-art compressive sensing schemes.
  • Guoliang Xing (MSU)
  • High-fidelity Noninvasive Measurement of Mobile Systems (Slides)
  • Andreas Terzis (Google)
  • Accelerate YouTube/Search (Slides)
  • Chunyi Peng (OSU)
  • Protocol verification (Slides)
  • Srini Seshan (CMU)
  • Title: Towards more timely measurements
  • Abstract: Managing these networks to ensure that they perform well and provide connectivity to all its users remains a complex and labor-intensive task. This complexity arises from the fact that wireless signal environment is difficult to model and predict accurately for a variety of reasons including: 1) the workload is hard to predict, 2) building material and layout are not well documented, 3) wireless propagation is hard to predict accurately even with accurate environment data, and 4) many properties of the environment changes over time (e.g. doors open/close and people move). Unfortunately, characterizing the wireless environment by performing detailed measurements is tedious and error-prone. In addition, since it is such an arduous process, these measurements are usually incomplete and rarely updated. This leaves network administrators configuring their networks based too much on educated guesswork. In this talk, we describe a range of techniques to simplify the process of measuring and monitoring our networks by leveraging both the activities of the network’s users and autonomous robots. (Slides)
  • Romit Roy Choudhury (UIUC)
  • Robotic Wireless Networks
  • Deepak Ganesan (UMass)
  • Title: Ultra-low power wireless backscatter
  • Abstract: Backscatter communication can support megabits/second, while consuming micro-watts of power. This opens the door for cloud offload from a range of high-rate sensors (e.g. wearable cameras, ECG, microphones, etc), while simultaneously making sensor devices smaller, longer-lived, and cheaper. We discuss some of our recent results in addressing challenges in backscatter communication, as well as the road-blocks that lie ahead. (Slides)

PANEL A: Data Collection and Privacy
This panel focuses on security and privacy. Security as a research goal, and privacy as both
a research goal and also a challenge. In particular, how to measure such systems when those systems are in use by humans, and the human use of the system (or their behavior around the system) is sensitive information whose privacy should be respected by the measurement
infrastructure and the researchers using it.


  • David Kotz (Dartmouth)
  • CRAWDAD, privacy/security in mobile and wireless systems (Slides)
  • Gyan Ranjan (Narus)
  • Issues with looking into HTTPS traffic. The real vs. perceived challenges for Network management posed by the rapid adoption of HTTPS. Prospective solutions and limitations.   (Slides)
  • Anoop Gupta (Microsoft)
  • Data formats
  • Lin Zhong (Rice)

PANEL B: Operational challenges for instrumentation and evolution: an industry perspective


  • Haiyang Qian (China Mobile)
  • SDN for Mobile Network Mobility
  • Changbo Wen (T-Mobile)
  • crowdsourcing data collection
  • User experience KPI
  • User experience impact by network upgrade
  • Customer feedback via crowdsourcing data (Slides)
  • Ying Zhang (Ericsson)
  • SDN based measurement for network anomaly detection (Slides)
  • propose an adaptive method that improves the anomaly detection accuracy while reducing the overhead
  • Erran Li (Bell Labs)

Title: Making Cellular Networks Scalable and Flexible

Abstract: The exponential growth of mobile data has put tremendous stress on the cellular

network infrastructure. As a result, mobile providers are scrambling to

aggressively build up their networks. Unfortunately, with today's designs, as

providers add network capacity, the growth of capital and operational costs

greatly outpace the growth in revenues.  To allow the network to scale

cost-effectively, a radically new design of the cellular network infrastructure

is necessary.

I will present CellSDN, a design based on software defined networking

principles. Departing from the current proprietary hardware based data plane

such as base stations, S-GW, and P-GW in LTE, CellSDN data plane consists of

simple SDN switches, radio elements (simplified base stations), and virtualized

network functions. To enable flexible control, we abstract the radio resources

in the network as a 3D grid (radio elements, time and frequency), and abstract

both the data plane and control plane with a set of logical and reconfigurable

data plane components such as virtual base stations and switches. Leveraging the

abstractions, I will present scalable control plane for network wide

applications such as interference management, mobility management, and flexible

sharing of radio access networks.   (Slides)


SESSION B: Testbeds
In this session, we will discuss existing measurement systems and research infrastructure, and how to integrate them, facilitating experiment deployment and analysis.


  • Walter Johnston (FCC)
  • Integrating research efforts
  • Kobus van der Merwe (Utah)
  • Geoff Challen (University at Buffalo)
  • My Platform Knows More Than Your App: Android Platform Experimentation on PhoneLab (Slides)
  • Dave Choffnes (NEU)
  • Mike P. Wittie (Montana State University)
  • TITLE: Survey of End-to-End Mobile Network Measurement Testbeds
  • ABSTRACT: 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, research, development, and regulatory communities have taken an increased interest in measuring mobile networkperformance and its impact on user experience. 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 researchers, developers, regulators, and the public. 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. (Slides)
  • Suman Banerjee (University of Wisconsin)
  • City-scale mobile network testbed (WiNEST)