Tutorial Schedule:

T1 -  Monday, 9:00 AM – 12:30 PM

MAC Protocols for Wireless Sensor Networks

Koen Langendoen, Delft University of Technology, Netherlands

T2 –  Monday, 1:30 PM – 5:00 PM

Location Discovery in Sensor Networks

Prof. Andreas Savvides, Yale University

T3 – Monday, 9:00 AM – 5:00 PM

Low Power Sensor Network Development with IEEE 802.15.4 and TinyOS 

Joe Polastre, Cory Sharp and Robert Szewczyk, Moteiv Corporation

T4 - Monday, 9:00 AM – 5:00 PM

Networks of Mobile Sensors

Greg Pottie, UCLA, Guohong Cao, Penn State University

Lunch Break on Monday: 12:30 PM – 1:30 PM

For further information, contact:

T1 -  Monday, 9:00 AM – 12:30 PM

MAC Protocols for Wireless Sensor Networks

Koen Langendoen, Delft University of Technology, Netherlands

The effective deployment of Wireless Sensor Networks (WSN) calls for new, unorthodox solutions to traditional distributed-computing problems to handle the nodes' scarce resources: energy and memory. Since radio communication is expensive in terms of energy consumption, yet collaboration between nodes is essential for providing emerging services, managing the communication protocol stack is the key to success. The performance of the Medium Access Control (MAC) layer determines the energy consumption to a large extent, since it decides when to switch the radio on/off.

In contrast to typical WLAN designs, which optimize for latency, throughput and fairness, WSN-specific MAC protocols focus on energy consumption and memory footprint. Impressive energy savings can be obtained by putting the radio into sleep mode for long periods of time, and a wide range of new MAC protocols have been proposed in the last few years.

This tutorial surveys about 20 WSN-specific MAC protocols and classifies them according to three key issues: number of used channels, degree of organization, and notification mechanism. Four MAC protocols (S-MAC, T-MAC, Low-power listening and LMAC) will be studied in depth, including a head-to-head comparison on a common simulation framework. The tutorial concludes with a number of generic guidelines for energy-efficient MAC design.

INTENDED AUDIENCE:

The tutorial targets people that have a basic understanding of MAC protocols for Wireless LANs (e.g., IEEE 802.11), and want to extend their knowledge to the domain of wireless sensor networks. It is important for protocol designers, as well as for application developers to understand the basic trade-offs involved (performance vs. energy-consumption).

OUTLINE:

* Introduction
- Contention-based medium access
- Schedule-based medium access
* Requirements for sensor networks
- Hardware characteristics
- Communication patterns
- Miscellaneous services
* Energy efficiency
- Sources of overhead
- Trade-offs
* Contention-based protocols
- IEEE 802.11
- Low Power Listening and Preamble Sampling
- WiseMAC
* Slotted protocols
- S-MAC
- T-MAC
- DMAC
* TDMA-based protocols
- LMAC
* Comparison
- Simulation framework
- Micro benchmarks
- Homogeneous unicast and broadcast
- Local gossip
- Convergecast, periodic reporting
- Discussion
* Conclusions

BIOGRAPHY:

Koen Langendoen is an associate professor in the Parallel and Distributed Systems group at Delft University of Technology, The Netherlands. He earned an M.Sc. in computer science from the Vrije Universiteit, Amsterdam in 1988 and a Ph.D. in computer science from the Universiteit van Amsterdam in 1993. His research interests include system software for parallel processing, wearable computing, embedded systems, and wireless sensor networks. Koen Langendoen has been actively involved in the area of Wireless Sensor Networks since 2001 when he worked as a visiting researcher atthe Berkeley Wireless Research Center in the PicoRadio group of Prof. J. Rabaey. Currently, he heads a research group working on localization, MAC protocols, link-layer performance, and various demo applications.

T2 –  Monday, 1:30 PM – 5:00 PM

Location Discovery in Sensor Networks

Prof. Andreas Savvides, Yale University

The plethora of applications and the fundamental nature of location awareness in wireless devices have attracted considerable interest in the research and commercial worlds during the recent years. Despite the increased research activity, several challenges and manufacturing costs have so far prevented the widespread deployment of location capable systems, especially in multihop configurations. This tutorial provides a detailed introduction to the localization problem. It provides an overview of the state-of-the-art technologies and describes the key algorithms and localization techniques developed over the last 2-3 years. The tutorial will also cover some of the design issues by examining the interplay between algorithms and existing technologies in the context of existing experimental systems.
 

* Introduction to node localization
* Technologies overview
+ Commercial Systems
+ Experimental/Research Systems
* Taxonomy of applications
* Basic concepts
+ Localization using distances, angles, connectivity, proximity
+ Deterministic Methods: Least Squares and Singular Value Decomposition Methods
+ Multihop-localization requirements
* Advanced Algorithms
+ Deterministic algorithms
+ Probabilistic methods
* Evaluation Metrics and Bounds
* Sensor Nodes and Localization System Architectures

The tutorial will appeal to a wide audience of researchers and practitioners in ubiquitous computing and sensor networks. Practitioners will gain a greater insight on existing systems and will acquire a basic understanding of the underlying processes and capabilities of existing systems. For researchers, this tutorial will provide a review the state-of-of-the-art solutions in terms of algorithms and existing solutions and will outline a set of challenges. No prior knowledge of localization systems is assumed but familiarity of basic wireless networking concepts will be useful.

BIOGRAPHY:
Andreas Savvides is an assistant professor in the Electrical Engineering and Computer Science Departments at Yale University. Andreas is the founder and director of the Embedded Networks and Applications Lab (ENALAB) at Yale. Andreas obtained his Ph.D in 2003 from the University of California, Los Angeles, where he worked on the design and implementation of ad-hoc localization systems for sensor networks. Andreas's research interests are in wireless sensor networks and smart environments. More information about his research can be obtained from his group website at http://www.eng.yale.edu/enalab
 

T3 – Monday, 9:00 AM – 5:00 PM

Low Power Sensor Network Development with IEEE 802.15.4 and TinyOS 

Robert Szewczyk, Joe Polastre, Cory Sharp, and Kristen Wright,  Moteiv Corporation

A family of low power wireless sensor network devices has been built to enable research and deployments. The devices have featured commercial off the shelf (COTS) components integrated together on a platform commonly referred to as a "mote". In parallel to building motes, UC Berkeley has developed an open source operating system for embedded wireless sensors called TinyOS. In use at over 500 universities and companies, TinyOS has become the de-facto standard for experimentation and research in the field of wireless sensor networks. This tutorial describes the basic concepts of TinyOS and their usage. Participants will start
developing and running TinyOS applications on the newest and lowest power mote to date, Telos. Featuring an IEEE 802.15.4 radio at 250kbps and a TI MSP430 microcontroller, Telos uses the latest hardware technology available. Low power operation on real world systems will be explored. Participants will be introduced to IEEE 802.15.4 and its use in TinyOS and will build a mesh networking application with network reprogramming and network management on the Telos platform. Existing TinyOS protocols will be introduced so that researchers may compare and benchmark their designs against existing solutions.

Outline (first 3 in the morning, last 2 in afternoon):

TinyOS Basics
* Characteristics of networked sensing systems
* Execution model/context and split phase processing
* TinyOS simulation environment
Get acquainted with Telos
* IEEE 802.15.4 compliant sensor module
* Basics of hardware low power operation
* Sensor interfaces and power considerations
Hands on: Install TinyOS
* "Hello World" for wireless sensor networks (and simulation)
* Build and deploy a single hop network
* Packet format, composition, and processing
* Exchange RSSI/LQI or build applications from scratch
Data Collection and Visualization
* Build and deploy a multihop/mesh data collection network
* Explore Network Reprogramming and Sensor Network Management
* Explore the IEEE 802.15.4 standard, its applicability to sensor networks
* Build Power Aware Applications and Enable Low Power Operation
Testbeds and Experimentation
* Techniques for quickly and effectively testing new research ideas
* Where services reside in TinyOS for comparison to new algorithms
- link, network, routing, dissemination, etc protocols

Audience:

This tutorial is intended for researchers that have never used TinyOS before or are TinyOS novices. Participants will be able to deploy sensors with power management, mesh networking, and network reprogramming by the end of the tutorial, enabling them to start building new applications and algorithms using TinyOS. This tutorial is hands-on and participants should plan on writing code and installing applications.

Please bring a laptop with Windows 2000 or Windows XP with a USB port. You must have administrative privileges to your laptop. Please arrive at 8am on the morning of the tutorial to install TinyOS if you have not done so already.

Telos motes will be used for the tutorial. If you have Telos motes, you may bring them to use. If you do not have Telos motes, some will be provided by the presenters and will be available first come first served. If you wish to purchase a mote prior to the conference, they will be offered at a 15% discount to conference attendees by calling Moteiv at 510-965-1312 or emailing info@moteiv.com.


Speaker Bios

Rob Szewczyk is founder and CEO of Moteiv Corporation. While at UC Berkeley, Rob's work has focused on methods for increasing reliability in wireless sensor networks; specifically looking at how to diagnose node failures and use that information in algorithms and system maintenance. Rob and Joe deployed a 150 node wireless sensor network on Great Duck Island, Maine in 2002 and 2003 (website http://www.greatduckisland.net). Our work on Great Duck Island was featured in Wired Magazine (http://www.wired.com/wired/archive/11.12/network.html), IEEE Spectrum, CNN, Business Week, BBC, Slashdot, The San Francisco Chronicle, and MIT Technical Review. Rob was a founding member of the TinyOS software project (http://www.tinyos.net). TinyOS is a small, open source, operating system that runs on embedded devices. Each release of TinyOS has over 10,000 downloads with more than 500 universities and corporations using the software worldwide.

Joe Polastre is founder and CTO of Moteiv Corporation as well as a PhD student at UC Berkeley. Joe's work has focused on enabling long term operation of wireless networks. He invented a low power wireless module, Telos, that features the new IEEE 802.15.4 wireless specification and is the lowest power wireless sensor network device to date. He wrote much of the low power software and algorithms for the Great Duck Island deployment that ran for over 5 months. His work is motivated by real world applications including environmental and building monitoring.

Cory Sharp is founder and CIO of Moteiv Corporation. Cory received his B.S. in Computer Engineering from the University of Oklahoma in 1997 and M.S. in
Electrical Engineering from the University of California, Berkeley in 2001. His work has focused on both real deployments of and programming abstractions for robotics and sensor networks. In 2003 he led a team that deployed a 100 node wireless sensor network for pursuit and interception -- the system coordinates with an autonomous pursuing vehicle to intercept evading vehicles within the sensor field. Cory has designed, built, and published many of the abstractions that are now a part of wireless sensor network applications in TinyOS. He is an active contributor to TinyOS, and he is currently leading a team at to deploy a 1000 node network in a large outdoor and urban area for pursuit and interception.

Kristin Wright has worked as UC Berkeley's user support liason for TinyOS since 2001. She received her MS from Berkeley in 1998 under Steven McCanne. She filled the interim time between Berkeley stints as a Research Associate at the University of Utah. Prior to graduate school, Kristin worked with four other engineers to design, implement, and deploy a cutting-edge OS for PDAs at Geoworks. Although an ardent fan, Kristin had nothing to do with Moteiv's founding and holds no position there.

 

T4 - Monday, 9:00 AM – 5:00 PM

Networks of Mobile Sensors

Greg Pottie, UCLA, Guohong Cao, Penn State University

Scaling up sensor networks to meet the challenges of energy constraints, the complexity of signal processing in natural environments, and the presence of obstacles to sensing, is a significant challenge. Since the sensing channel depends on physical properties of the environment and sensor elements, then in general, only a physical reconfiguration can change this distortion or occlusion. Therefore, a fixed sensor network has limitations in gathering data and disseminating the data. Mobile elements can be used to address these problems by reconfiguring the sensor network. Mobility can significantly increase the capability of the sensor network by making it resilient to failures, reactive to events, and be able to support disparate missions with a common set of sensors.

In this tutorial, we provide an overview of mobility assisted sensor networks. The morning session explores the use of infrastructure-assisted mobility, while the afternoon session focuses upon the use of untethered robots for mobility-assisted sensing and data dissemination. Networked Infomechanical Systems (NIMS) are composed of physical infrastructure (cables) along which robotic nodes carrying sensors and samplers can move, and untethered sensor nodes spread across the environment. The untethered elements of the system communicate using low-power radios. NIMS allows us to introduce the physical reconfiguration that is necessary for adapting physical sensors, recalibration, and logistical support over sustained periods. Various aspects of the NIMS system including its design, coordination mechanisms, adaptive sampling algorithms, statistical foundations, information theoretic foundations and applications will be discussed.

In the afternoon, solutions will be discussed for relocating sensors in response to an event (e.g., a target comes) or network condition changes (e.g., sensor failure).
Sensor deployment, localization, failure (coverage hole) detection, and time-synchronization algorithms will also be described. Mobility assisted data dissemination using short range communication in combination with routing protocols will also be described. Specifically, nodes buffer and carry data during network partitions, and forward the data to other nodes. Techniques of using node mobility to increase network lifetime are also discussed.

Outline of Topics

Morning
1. Foundations of infomechanical networked systems
Greg Pottie (UCLA EE Dept.)
2. Coordinated mobility and marine applications
Gaurav Sukhatme (USC CS Dept.)
3. Adaptive sampling foundations
Mohammed Rahimi (UCLA CS Dept.)
4. Adaptive task allocation
Maxim Batalin (USC CS Dept.)
5. Robo-gaming
Aman Kansal (UCLA EE Dept.)

Afternoon
1. Mobility-assisted sensing
Guohong Cao (Penn State CSE Dept.)
2. Mobility-assisted data dissemination
Guohong Cao (Penn State CSE Dept.)

Intended Audience

This tutorial is designed for a general audience with some interest in extensions of sensor networks to include robotic elements. It is pitched at an introductory level so that practicing engineers (EE or CS) or graduate students beginning their research can benefit. Both mathematical foundations and examples of deployed systems will be discussed.

Biographies

Gregory J Pottie. From 1989 to 1991 he worked in the area of voice-band modems and digital subscriber lines for Motorola. In 1991 he joined the faculty of the UCLA Electrical Engineering Department. He now serves as the Associate Dean for Research and Physical Resources for the engineering school. His research at UCLA has included wireless networking, channel coding and sensor networks, the latter since 1994 in a series of collaborative projects in both academia and private industry. His current focus is upon the information theoretic foundations of such networks. He has received the Allied Signal Faculty Research Award.

Gaurav S. Sukhatme is a faculty member of the Computer Science Department at the University of Southern California (USC). He is the co-director of the USC Robotics Research Laboratory and the director of the Robotic Embedded Systems Laboratory which he founded at USC in 2000. His research interests are in distributed mobile robotics and sensor/actuator networks. He is on the editorial boards of two leading journals in robotics, IEEE Transactions on Robotics and Automation, and Autonomous Robots and a leading magazine in ubiquitous computing, IEEE Pervasive Computing. He has published over 100 technical papers and is a recipient of the NSF CAREER award.

Mohammed Rahimi is a research engineer on staff at the Center for Embedded Networked Sensors and also pursuing graduate studies in the UCLA CS department. He has been lead developer in a sequence of field deployments of several classes of sensor nodes, and is investigating the fundamentals of how nodes should be deployed to optimally sample a natural environment.

Maxim Batalin is a Ph.D. student in the USC CS department, with lead responsibility for the adaptive task allocation project in the Center for Embedded Networked Sensors. He has developed algorithms, software tools, and field experiments that demonstrate the interaction of static and mobile elements in sensor networks.

Aman Kansal is a Ph.D. student in the UCLA EE department. He has developed a laboratory-scale NIMS system, derived fundamental information theoretic limits to the benefits of local cooperation, and investigated at many levels the effects of the introduction of a small number of mobile elements in a static sensor network. He is also the principal developer of the Ragobot, the basis for a set of robotic games a student team is creating.

Guohong Cao received his BS degree from Xian Jiaotong University, Xian, China. He received the MS degree and PhD degree in computer science from the Ohio State University in 1997 and 1999 respectively. He joined the Department of Computer Science and Engineering at the Pennsylvania State University in 1999, and is currently an Associate Professor. His research interests are mobile computing, wireless networks, and distributed fault-tolerant computing. He is currently an editor of the IEEE Transactions on Mobile Computing and IEEE Transactions on Wireless Communications, has served as a co-chair of the workshop on mobile distributed systems, and has served on the program committeeof numerous conferences. He was a recipient of the Presidential Fellowship at the Ohio State University in 1999, and a recipient of the NSF CAREER award in 2001.