David Carey uses Wireless Sensor Networks for Flood Plain Monitoring

One of our lead users for the Wireless Sensor Networks application is David Carey who performs research at the Wilkes University in Pennsylvania. I had a chance to speak with him today about his research.

How are you using Wireless Sensor Networks?

The WSN will be stationed along the flood protection system in Wilkes-Barre, PA to monitor the condition of the levy system. Ultimately this will lead to an early warning system for the flood protection system and a real time levy health monitoring system. The WSN will be configured to monitor water level, vibration and pressure that the river exerts on the levy.

What are current solutions and why do they not work?

Currently the system that monitors the levy is a hardwired system and covers a limited area along the protected zone. The system also only looks at pressure and water level. A more detailed understanding of hydrodynamics is required but the 60,000 foot view follows:

The current system has a limited number of pressure sensors. They are spaced along the levy offering partial coverage. If a breech occurs upstream, the pressure wave as seen by the sensors will indicate this but only minutes prior to the event hitting the system. The current system does not offer enough early warning that a potentially catastrophic event is on its way.

How rugged must the system be to survive an outdoor environmental application?

The sensors need to be mounted in a weather and water tight box. They must be able to withstand the swing in temperature and the elements of rain and snow seen in northeast PA. There is a potential that the WSN will be submerged in water. This is only for a short time and the sensor will not be expected to transmit during this time. It will however be expected to collect data for later transmission. Some WSN nodes will be mounted in subterranean conduits (pipes). For them to transmit data out a drone vehicle has been proposed to patrol the piping system to collect data and service the nodes as required.

What is the biggest advantage in using wireless sensor networks?

The greatest advantage is in the flexibility of placement of the sensor nodes. The network can be customized in placement to the levy system and will not require the laying of long cables.

What is the biggest challenge in using wireless sensor networks?

There are several challenges. The device must survive in the environment. This can be handled through packaging. Transmitter efficiency and reliability in the environment is the second problem. The devices will be mounted above and below ground. The two areas will not be in contact with each other. It must be determined how to link the two environments together. Additional challenges having nothing to do with being a WSN: First is data fusion of the vibration and pressure sensors to detect an event early enough to be of service. The current sensor package might not have enough onboard memory to allow the level of processing required to make this effective. Finally, time is an issue. In the academic environment where teaching of classes is of higher priority than research it will be difficult to get everything completed by the January deadline.

What is your research focused on?

Experimentation has been conducted to examine the throughput and longevity of the sensors through various data sampling rates, loads and configurations. A report on the transmit rates VS battery life is being drafted. Additional work is being done on reliability of transmission. The sensors are being placed at various distances and the signal strength is being monitored by looking at the data throughput. In addition data latency is being examined. All of this is to go towards quantifying the sensors ability to perform and this will help determine the overall structure of the final network.

Signal processing and analysis is being examined. Various levels of onboard sensor processing routines are being developed to reduce the data transmission load. The bulk of the work will be in the feature identification and extraction from the sensor signals. It will then be determined if the fusion process can be embedded into the WSN or will it have to be offloaded.

Additional analysis is being performed through river and hydrodynamic simulation. Wilkes University uses a software package in their Environmental Engineering Program. The simulation is being used to determine the effects of various events on the levy. This is going to help verify if the vibration/pressure package is going to provide the early warning required.

What is the role of software in your application?

The embedded application will sample the sensors and perform some analysis to determine if an early warning event on the levy has occurred. The application that is interfaced to the coordinator will monitor all sensors and provide real-time up-to-the-minute health status of the levy.

What is your next application challenge?

My next big hurdle is in streamlining the embedded analysis routine to fit on the sensor. The features I need to monitor are in the frequency domain. I need to reduce the number of samples I am using for the frequency analysis. The nice thing is the events that need to be identified are in the low frequency range. This means that the sampling rate is not too high.

Best regards,
Hall T.