电子工程系系列学术报告
——无线传感器网络研究系列报告
 
系列报告题目
1. Link Quality and Medium Access
    Time: 4月28日上午　10:00-12:00
    地点：东主楼10区309
   
2. Data-Centric Storage and Querying
    Time: 4月29日上午　10:00—12:00
    地点：东主楼10区309
 
3. Opportunistic Spectrum Access
    Time: 5月5日上午　10:00-12:00
    地点：东主楼10区309
 
4. Joint Routing and Compression
    Time: 5月6日上午　10:00-12:00
    地点：东主楼10区309
 
5. Multi-Hop Scheduling
    Time: 5月7日上午　10:00-12:00
    地点：东主楼10区309
 
6. Optimization-based Rate Control
    Time: 5月12日上午　10:00-12:00
    地点：东主楼10区309
 
7. Vehicular Networking
    Time: 5月13日上午　10:00-12:00
    地点：东主楼10区309
 
8. Mobile Social Networks
    Time: 5月14日上午　10:00-12:00
    地点：东主楼10区309
 
报告人：
Dr. Bhaskar Krishnamachari
Associate Professor 
Department of Electrical Engineering-Systems
Department of Computer Science (Joint Appointment)
USC Viterbi School of Engineering
http://ceng.usc.edu/~bkrishna/

报告摘要：

Annoucement by Professor:  These are exciting times for research into next-generation wireless networks. This series of eight lectures spans a broad range of research topics concerning theoretical and practical issues for such systems, particularly sensor networks. In each talk, I hope to give some insight into the state of the art for that topic, describe some interesting experimental and theoretical findings from my group's research, and point out open problems and directions for future research.

All the talks will be given in room 10-309 of the East Main Building between 10:00am and 12:00am.

1. Link Quality and Medium Access (4/28, Tue)
I begin the series by exploring link quality and single-hop medium access issues in wireless sensor networks. In this talk, I will describe the first-ever large-scale multi-hop wireless network experiment that I was involved with eight years ago, and talk about the gap between idealized simulations and real measurements. I will discuss the relationship between link quality and physical distance and how the resulting "gray area" affects the reliability of network protocols. I will then discuss packet receptions in the presence of interfering transmissions, and show experimentally that today's CSMA-CA protocols are overly conservative because they do not sufficiently exploit power-control. Finally, I will present a study showing how the IEEE 802.15.4 protocol for low-power wireless networks can be enhanced through a feedback mechanism to substantially improve its scalability in terms of both throughput and energy.

2. Data-Centric Storage and Querying  (4/29 Wed)
The central paradigm of data-centric networking in wireless sensor networks allows for in-network storage of named events and their retrieval through on-demand queries. In this talk, we will focus on the mathematical modeling of storage and querying mechanisms through several case studies. In one case study, I will present random walk queries and consider how their performance can be improved. In another, we mathematically compare the performance of different hybrid push-pull techniques. Finally, we present an interesting analysis that reveals fundamental scaling laws for data-centric querying mechanisms.
This analysis shows that there are limiting query and event growth conditions beyond which data-centric storage and querying cannot be sustained in arbitrarily large sensor networks.

3. Opportunistic Spectrum Access  (5/5 Tue)
In this talk, I digress slightly from the principal theme of this series (sensor networks), to talk about a different kind of sensing - the sensing of spectrum opportunities by a cognitive radio. After providing a broad introduction to cognitive radio and dynamic spectrum access, I will describe some recent theoretical work where we explore ideas from Markov decision processes and game theory to optimize the use of spectrum opportunities.

4. Joint Routing and Compression  (5/6 Wed)
In order to obtain data from environmental-monitoring sensor networks in an energy-efficient manner, it is helpful to remove spatial correlations through in-network compression. I first present a theoretical result that shows that while obtaining optimal energy performance requires correlation-aware routing, near-optimal performance can be obtained with a simple cluster-based routing strategy. I then present a more practical perspective on this problem by describing our ongoing efforts to develop a software architecture for compression and routing in sensor networks, including an implementation of a novel distributed-wavelet based compression scheme that works over collection trees.

5. Multi-Hop Scheduling (5/7 Thur)
In this talk, we will consider different variants of the following fundamental question. Assuming global time-synchronization and some traffic pattern in a given multi-hop wireless sensor network, what time and frequency scheduling policy provides the best performance with respect to different energy and latency metrics? While these problems are generally NP-hard, we develop some good heuristics, where possible with provable approximation guarantees.

6. Optimization-based Rate Control (5/12 Tue)
Since the pioneering work by Kelly et al. (1998) and Low-Lapsley (1999), there has been a concerted attempt to make congestion control a science rather than an art, by giving it a mathematical basis. While this has found good success in wired networks, I will describe why there is a greater gap between theory and practice in wireless networks. I present a cross-layer bandwidth model for wireless sensor networks called the receiver capacity model. As an application of this model, I will
describe the design and implementation of the wireless rate control protocol (WRCP), the first explicit and precise distributed rate-based congestion control protocol for wireless sensor networks. I will also discuss how this model can be used to develop market-based rate optimization algorithms. Finally, I will discuss work on developing a novel Backpressure-based Rate Control Protocol (BRCP), which performs rate optimization based on queue-backpressure scheduling and Lyaopunov-drift minimization.

7. Vehicular Networking (5/13 Wed)
Vehicular ad hoc networks (VANETs) represent a major frontier in the development of large-scale wireless technologies. They are envisioned for a wide range of applications spanning vehicular safety, traffic and other urban sensing tasks, toll collection, information access and entertainment. I will discuss the state of the art and some of the key challenges and opportunities in this domain from a research perspective. I will present results pertaining to content replication and carrier-based content delivery for vehicular peer-to-peer networks and a proposal for a novel information-centric architecture for this domain.

8. Mobile Social Networks  (5/14 Thur)
There is an incipient school of thought that we will some day (perhaps soon) reach a stage where the principal innovations in the domain of wireless technology will occur in overlay applications that harness the underlying infrastructure rather than enhancements to the infrastructure itself. In this final talk, I will discuss an important emerging class of community based/social networking applications based on mobile devices, which represent a new kind of large-scale sensor network. I will first discuss some fun mobile social games that we have developed and tested in my lab, and describe results from an interesting experimental study to model social contacts. I will then present and describe some experimental and analytical results pertaining to energy-efficient personal activity tracking using mobile devices.  Finally, I will discuss a fundamental privacy-participation trade-off that is common to many mobile-based community applications and show through two concrete examples how it can be handled through utility optimization and game theory.

报告人简历：
Dr. Krishnamachari is Philip and Cayley Early Career Chair Assistant Professor at the Viterbi School of Engineering at the University of Southern California. He has been a faculty member in the Department of Electrical Engineering - Systems since 2002, with a joint appointment in the Department of Computer Science at USC. He directs the Autonomous Networks Research Group at USC.
 
Dr. Krishnamachari's research interests are focused on performance analysis and design of algorithms for information routing and self-configuration in large-scale wireless sensor networks. Sensor networks are large-scale unattended networked embedded systems, with applications ranging from environmental sensing, structural monitoring, and industrial process control, to emergency response and mobile target tracking.
 
Dr. Krishnamachari received the USC Viterbi School of Engineering Junior Faculty Research Award in 2005 and the prestigious National Science Foundation CAREER award in 2004. He is actively involved in the sensor networks community, serving as TPC member and organizer for several research conferences. He is an editor for the Elsevier Journal of Ad Hoc Networks, the ACM Mobile Computing and Communications Review, and the EURASIP Journal of Wireless Communications and Networking. Dr. Krishnamachari has authored a textbook titled Networking Wireless Sensors, recently published by Cambridge University Press.
 
Dr. Krishnamachari obtained his B.E. in Electrical Engineering with a four-year full tuition scholarship at The Cooper Union for the Advancement of Science and Art, in New York City in 1998. He was a co-recipient of the first prize at the 1998 IEEE Region I annual student paper contest, for a paper based on his senior project: Design of Ultra High Speed Microwave Digital Logic Gates. He then pursued his graduate studies at Cornell University, where he was awarded a three-year graduate fellowship and named one of the eight 1998 Olin Presidential Fellows. He completed his M.S. in Electrical Engineering with a thesis titled Global Optimization in the Design of Mobile Communication Systems in 1999. In May 2002, he defended his Ph.D. thesis dissertation at Cornell University on Phase Transitions, Structure, and Complexity in Wireless Networks.
 
Dr. Krishnamachari is a member of the IEEE and ACM and the honor societies Tau Beta Pi and Eta Kappa Nu.
 
 
联系人：张林
Complex Engineered Systems Lab (CESL)
Department of Electronic Engineering, Tsinghua University
Tsinghua (EE)—Cisco Joint Lab for Green Technology
Tel / Fax： +86 (10) 62781382
http://sensor.ee.tsinghua.edu.cn