Friday, March 25, 2005

Wireless Sensor Networks on the Verge

I recently had the opportunity of visiting David Nagel at George Washington University – a leading researcher in the field of wireless sensor networks. For those not familiar with Wireless Sensor Networks, Dr. Nagel defined it as follows: “Wireless Sensor Systems use pre-determined, single-hop radio frequency links to transfer information from the originating sensor node through other nodes, routers, and gateways to users. Networks with many sensors and communication nodes provide situation-determined multi-hop paths for the information to flow from the source to the destination. These networks can be ad hoc, that is self-organizing which simplifies both their deployment and employment. Also, their configuration in two-dimensional arrays called meshes enables information to travel over alternative available paths which greatly increases the reliability of information delivery. “

Simple networks of 20-50 nodes in a handful of specific applications such as building automation, home networking, and even parking meter monitoring are being used today. Applications with few variables (those parking meters aren’t going to move and if the node stops working the risk is low) are being installed today by industry while university and government groups are researching how to solve the more difficult ones such as how to make each node work autonomously but cooperatively. In other words the node must be able to control its own functioning (e.g. waking up to make a measurement) as well as work in a cohesive manner with other nodes (e.g. sending or receiving data or commands to other nodes in the network). Also, dealing with 20 nodes is not so difficult but how do you handle 2000 nodes or 20,000 nodes. At some point the old way of programming must give way to new techniques to support such large numbers.

Some of the more challenging applications include swarming networks. Researchers envision autonomous vehicles operating under computer control with little or no human control. This will require processing large amounts of data and extensive communications among the vehicles. The work is driven by the government seeking to build autonomous vehicles for work in dangerous situations such as combat zones.

The second application involves the use of sensors for biological detection. In the event of a biological attack, a series of sensors positioned over a wide area could determine the direction and speed of cloud of poisonous chemicals, which authorities could then use to for evacuation purposes.

Finally, a third application involves the use of wireless sensors for “drive by wireless” applications in which the wireless nodes measure and control the various systems in the vehicle and communication to the central processor is done through wireless connectivity. By reducing the cabling, the vehicle runs more efficiently due to the decreased weight which can be quite substantial for vehicles such as airplanes.
Recently, I had the opportunity of visiting several universities doing pioneer work in this area. At Virginia Tech University, I met with the team putting together the
JOUSTER -- Joint Unmanned System Test Experimentation and Research program. The program provides a facility for supporting unmanned system experimentation. They have a 1200 acre site which includes on-road and off-road terrain, site instrumentation, and wireless networking.

I also met with at Dr. David Nagel at George Washington University who researches the impact of security procedures (e.g. encryption) on the performance of a wireless sensor network.
In particular he focuses on ship container and home network applications. He believes there will be a convergence of RFID (What is it?), GPS (Where is it?) and Sensor networks (What is its condition?) in these networks.

If you are working in the area of Ad Hoc Networks or Wireless Sensor Networks, I would like to hear from you. Please email me at hall.martin@ni.com.

Best regards,
Hall T. Martin

Thursday, March 17, 2005

Bioinformatics, Microarrays, and Understanding the Functioning of the Cell

Recently, I had the opportunity to meet with Orly Alter, a researcher at the University of Texas in Austin, to discuss the issues and challenges in the world of Bioinformatics. According to Orly, the primary challenge in Bioinformatics is the visualization of data and clustering data into meaningful sets. The visualization of data needs to be interactive with multiple views and multiple queries as there are many unknowns and thus cannot be automated yet. She also raised the issue of data integration and analysis since data comes from microarrays stations locally as well as publicly available databases. Finally, large data sets can require grid computing and other CPU-intensive resources. Orly indicated that Bioinformatics today is in the same place astronomy was in the 17th century. Lots of data points and efforts are underway but the broad-based principles remain to be discovered.

I also met with Jim Collett, who is a graduate student at the University of Texas. He is using BASE, an open source tool for microarray analysis. He has a background in lab automation and is familiar with the challenge of automating the reading of microplates. Jim is interested in linking databases, open source bioinformatics tools, and mathematical software into a cohesive whole the better models the functioning of a cell. He pointed out Sydney Brenner , a Nobel prize winner of 2002, as a leader in the field. Brenner described the ideal environment for modeling the functioning of a cell over time and under various conditions as one which includes an object oriented, graphical environment with mathematical tools, simulation capability, and a visualization environment. Finally, Jim pointed out an effort at MIT called Synthetic Biology which focuses in this area. You can see their take on it at this site.

Microsoft had an interesting take on the subject with this site. This is all food for thought as we take Virtual Instrumentation into new fields. It appears the Bioinformatics may be a fertile field.

If you are working in the Bioinformatics area, I would like to share more thoughts with you. You can reach me at hall.martin@ni.com.

Best regards,
Hall T. Martin

Monday, March 14, 2005

Wireless Sensor Networks Research in the Academic World

This past week gave me the opportunity to visit numerous universities working on the Ad Hoc Network/Wireless Sensor Network challenge. The challenge is how to deal with a network of nodes that may be in unknown positions or move around so that the position of the nodes may change. The other challenge is how to deal with a large number of these nodes – say over 1000 of them. At that level, it would be difficult to program and track each node.

In Wireless Sensor Networks, areas of investigation included developing algorithms for locationing, communications, energy management, and dealing with defective or non-responding nodes (perhaps the battery ran down on a particular node). This last one is part of a larger effort to develop algorithms for propagating information through a network of nodes in which the network could be changing or unknown. Many of the researchers used autonomous vehicles such as ground-based robots and aero vehicles to test out concepts and apply algorithms. Their goal is to develop strategies and algorithms for communicating and sharing information among the robots and also some facility for each robot to work autonomously.

You can see what work is being done at Stanford , at Berkeley, and Caltech.
In most cases, the researcher used simulation tools to test out their concepts for dealing with large numbers of nodes. At one university they wanted an application generator – tell the software what you want to measure and then the system determines how many nodes and what position in order to provide the application measurement.

Working with over a 1000+ nodes in a network presents several challenges. How can one keep track of 1000+ nodes and program them individually? There needs to be a way to aggregate the nodes into meaningful groups (e.g. All nodes on floor 1 of a building, or all nodes that read temperature and currently have a temperature > 40C, etc). One researcher described the requirements for a working with a wireless network system as the following:
1. All sensors placed on the node.
2. Communication aspects must be reconfigurable.
3. Distributed computation.
4. Nodes can be heterogeneous.
5. Synchronization – for scheduling strategy.
6. Several sensors controlled by a computer (not one to one)

They pointed me to several conferences that are key to this area including Mobicom, Globecom, and Infocom.

If you are working in the area of Ad Hoc Networks or Wireless Sensor Networks, I would like to hear from you. Please email me at hall.martin@ni.com.

Best regards,
Hall T. Martin

Friday, March 04, 2005

Smart Phones Appear to be the Gadget of Choice

I recently read that Sony is dropping their PDA line completely. The article is here. Worldwide PDA sales have dropped for the 3rd year in a row to a figure below 10 million units. Smart Phones now appear to be the gadget of choice. The range of offerings include not only the PDA functionality such as address book and To-do lists, but also video, and more. The camera on the original mobile phones appeared to be toys which attracted primarily the teen market but now sophisticated imaging tools are coming to the mobile phone. To see an example click here.

Microsoft announced recently a framework for providing an abstraction layer that allows one to use the mobile phone to connect to various ERP systems. This will allow additional web services capabilities. The announcement is somewhat cryptic but it does open the way for more sophisticated web-based applications to run over the mobile phone network.

What does that mean for Virtual Instrumentation? It means a wider range of choices. Not only does the mobile phone take over the PDA functionality, it also brings additional capabilities in communications, web services, vision, and audio, to name a few.

If you see possibilities in using the next generation smart phone in your application, please contact me at hall.martin@ni.com.

Best regards,
Hall T. Martin