Home > People > Vivek Shrivastava

Vivek Shrivastava

Researcher
Palo Alto, CA

I am part of Mobile Computer Systems team working on  the advanced smartphone design and prototyping. Prior to that, I received a Ph.D. from University of Wisconsin-Madison in 2010 under the guidance of Professor Suman Banerjee . At Wisconsin, I worked on improving wireless connectivity and performance, especially in enterprise settings. My work demonstrated that the inherent centralization in enterprise WLANs can be leveraged to enable high performance, predictable and manageable wireless networks. I also received IBM PhD fellowship for 2009-2010 and a best paper award at Mobicom 2009 for my work in hybrid scheduling in enterprise networks.

 

Professional Activities

TPC Member Mobihoc 2011

Reviewer for Transaction on Mobile Computing (TMC), Transaction on Networking (ToN) 

Research Interests

Mobile and wireless networking. Protocol/application/algorithm development, measurement and analysis.

Research Projects

 

Channel Management in Wireless LANs
In this project, we address the joint problem of channel assignment, i.e., assigning channels to access points and load balancing, i.e., assigning clients to access points such that the overall network throughput is maximized according to a client-driven fairness objective. Built practical centralized and distributed solutions separately tailored for both enterprise and hotspot environments. Through carefully studied experiments, we showed the ability to use partially overlapped channels in both 2.4 and 5 GHz band, can improve spectrum utilization by a factor of 3 in dense ap deployments. Our Work was published at conferences such as ACM Mobicom and ACM Sigmetrics.
 
Analysis of city wide commercial mesh network
Wireless Mesh Networks (WMNs) represent a new and promising paradigm that uses multihop communications to extend WiFi networks. Several WMNs are already deployed and operational. Since the clients pay the same flat rate, the throughput sharing should also be fair. In this project we are working closely with Madcity Networks, that has deployed a city wide mesh network comprising of 750 access points in Madison,WI. We are working closely with the company to identify the performance bottlenecks and other practical implementation challenges.
 
Online passive conflict graph generation for enterprise WLANs
Trends in enterprise WLAN usage and deployment point to the need for tools that can capture interference in real time. Such a tool for interference estimation can enable WLAN managers to improve the network performance by dynamically adjusting operating parameters like channel of operation and transmit power of access points, but also diagnose and potentially proactively fix problems. Prior work on interference estimation using active probes impose significant overhead and cannot be employed to continuously obtain interference information as they evolve over time. In this project, we present the design, implementation and detailed evaluation of a Passive Interference Estimator (PIE) that can dynamically generate fine grained interference estimates across an entire WLAN. The most attractive feature of PIE is that it imposes no measurement traffic, and yet provides an accurate estimate of WLAN interference as it changes with client mobility, dynamic traffic loads and varying channel conditions.
 
Centralized Scheduling in Enterprise Wireless LANs
Wireless LANs are commonplace installations in enterprise environments. Their ease of use and deployment, however, are accompanied by a difficulty in their management and security. While centralization of some control plane tasks has been shown to be feasible, centralization mechanisms on the data plane are significantly harder to realize. The reason is that they need to take into account the inherent variability in the wireless medium while offering bounded delay and latency. In this paper, we first, present a comprehensive study of the various problems that arise in data plane centralization in an enterprise WLAN, and then, demonstrate that overall performance is best optimized through a hybrid approach that allows a limited amount of centralization to co-exist with decentralized, randomized channel access methods.

Personal Information

 

Ph.D., Computer Sciences,
Advisor: Dr. Suman Banerjee,
Thesis: Optimizing Enterprise Wireless Networks through Centralization
University of Wisconsin, June 2004 - June 2010.
 
M.S., Computer Science,
University of Wisconsin, May 2006.
GPA 3.93/4.0
 
B.Tech., Computer Sciences and Engineering,
Indian Institute of Technology, May 2004.
Department Rank 2/41, Institute Rank 2/200

Other Information

 

  • Best paper award at Mobicom 2009
  • Awarded 2009-2010 IBM PhD Fellowship
  • Third Prize Cisco Network Programming Contest - UW Madison,2007
  • First Prize ACM Student Research Competition - Mobicom 2006
  • Student Rank One Merit Scholarship 2000-2001
  • Merit Scholarship - All India Talent Search Competition, 1999

Publications

 

 

  1. FLUID: Improving Throughputs in Enterprise Wireless LANs through Flexible Channelization. Shravan Rayanchu, Vivek Shrivastava, Suman Banerjee, and Ranveer Chandra. In Proc. of ACM Mobicom 2011

  2. PIE in the Sky: Online Passive Interference Estimation for Enterprise WLANs. Vivek Shrivastava, Shravan Rayanchu, Suman Banerjee, and Konstantina Papagiannaki. In Proc. of USENIX NSDI 2011.

  3. CENTAUR: Realizing the Full Potential of Centralized WLANs through a Hybrid Data Path. Vivek Shrivastava, Nabeel Ahmed, Shravan Rayanchu, Suman Banerjee, S. Keshav, Konstantina Papagiannaki, and Arunesh Mishra. In Proc. of ACM Mobicom 2009. (Best paper Award).
  4. 802.11n Under the Microscope. Vivek Shrivastava, Shravan Rayanchu, Jongwoon Yoonj, Suman Banerjee. In Proc. ACM/USENIX Internet Measurement Conference 2008.
  5. A Measurement Study of a Commercial-grade Urban WiFi Mesh. Vladimir Brik, Shravan Rayanchu, Sharad Saha, Sayandeep Sen, Vivek Shrivastava, Suman Banerjee. In Proc. ACM/USENIX Internet Measurement Conference 2008.
  6. Understanding the Limitations of Transmit Power Control for Indoor WLANs. Vivek Shrivastava, Dheeraj Agrawal, Arunesh Mishra, Suman Banerjee, Tamer Nadeem. In Proc. ACM/USENIX Internet Measurement Conference 2007.
  7. Interference Mitigation in WLANs with Speculative Scheduling (short paper). Nabeel Ahmed, Vivek Shrivastava, Arunesh Mishra, Suman Banerjee, Srinivasan Keshav, Dina Papagiannaki. In Proc. of ACM Mobicom 2007.
  8. Towards an Architecture for Efficient Spectrum Slicing. Suman Banerjee, Arunesh Mishra, Vladimir Brik, Vivek Shrivastava, and Victor Bahl. In Proc. of IEEE HotMobile, 2007
  9. Load Balancing Large Scale RFID Systems. Quenfeng Dong, Ashutosh Shukla, Vivek Shrivastava, Dheeraj Agrawal, Suman Banerjee and Kaushik Kar. In Proc. of IEEE Infocom Minisymposium, 2007
  10. On the (In)feasibility of Fine Grained Power Control. Vivek Shrivastava, Dheeraj Agrawal, Arunesh Mishra, Suman Banerjee, and Tamer Nadeem. ACM Mobicom SRC, September 2006 (First Prize in Student Research Competition)
  11. Distributed Channel Management in Uncoordinated Wireless Environments. Arunesh Mishra, Vivek Shrivastava, Dheeraj Agrawal and Suman Banerjee. In Proc. of ACM Mobicom, 2006.
  12. Partially Overlapped Channels Not Considered Harmful. Arunesh Mishra, Vivek Shrivastava, Suman Banerjee and William Arbaugh. In Proc. of ACM Sigmetrics, 2006.
  13. Natural Selection in Peer-to-Peer Streaming: From the Cathedral to the Bazaar. Vivek Shrivastava and Suman Banerjee. In Proc. of ACM Nossdav 2005.