In military defense and medicine, in industry and the environment, and in a host of other applications, wireless sensor networks (WSNs) have quietly become an indispensable part of modern life.
WSNs monitor the temperature and depth of permafrost in the Swiss Alps, and measure water levels and leachate accumulation in landfills. On battlefields, they detect the presence of enemy troops, and on roads, they help regulate traffic. WSNs also make it possible to track a person’s pulse rate and other vital signals remotely.
In a wireless sensor network, battery-powered nodes containing radio transceivers and CPUs (central processing units) are embedded in the environment to sense and process data and transmit it to a base station. That transmission can be accomplished either directly from each node to the base station, in what is called a "star" topology, or indirectly, along a chain of nodes to the base station, in what is called a "multi-hop" topology. At the base station, data is interpreted and an appropriate response is determined.
The batteries that power WSN nodes, says Liang Cheng, consume a growing amount of energy, much of which could be saved through a smarter configuration of the networks.
Cheng, an associate professor of computer science and engineering, is the principal investigator for a grant recently awarded by the National Science Foundation (NSF) to develop smarter, more energy-efficient topologies for multi-hop wireless networks, including WSNs. He is also the co-principal investigator in two other recently awarded NSF grants involving WSNs.
The optimal configuration of nodes
"Topology control," says Cheng, "involves placing and connecting WSN nodes in strategic locations with optimal powers so that total energy consumption—the energy required to do the computation plus the energy required to transmit data—is minimized without affecting the network’s overall performance.
"Topology control in wireless data networks started as a field in the 1970s. The challenges then included where to locate the tower and the sending and receiving nodes. The challenges today include something new—how to make topology control more energy-efficient and environmentally independent."
As a general observation, says Cheng, 80 percent of the energy consumed by a WSN is used for data transmission and the remaining 20 percent for data processing by the nodes in the network.
"We want to see if we can reduce the amount of data transmitted to the base station by determining which data is vital and needs to be transmitted, and which is not. If you are trying to transmit too much data, that consumes too much energy.
"The tradeoff is that, in order to make this determination, more data processing has to be done by the nodes. The overall goal is to transmit as much data using as little energy as possible in a multi-hop network with reduced interference and higher capacity in the presence of multipath fading, link failures, high error rates and many other radio irregularities."
Improving the energy efficiency of WSNs, says Cheng, will also enable nodes and batteries to last longer and thus reduce the number of times workers must replace batteries.
Another way to boost energy efficiency that Cheng is investigating is to design a multi-hop WSN so the roles played by each node in the network can be dynamically changed. If one node is using up too much energy, for example, its tasks can be assigned to other nodes.
The advantages of going wireless
Two other recent NSF awards for which Cheng is co-principal investigator also involve multi-hop WSNs. In one project he is working with Shamim Pakzad, the P.C. Rossin Assistant Professor of civil and environmental engineering, to develop WSNs that monitor a bridge’s response to traffic stresses as well as its responses to earthquakes and explosions.
One objective of this project is to develop WSNs that can preempt regularly scheduled activities, such as the ongoing monitoring of traffic, and switch in a second or less to the monitoring of unscheduled events such as earthquakes. This "preemptive media access control" allows a "trigger" message from an earthquake or explosion to be received, processed and propagated throughout the network. Timely propagation in turn allows authorities to determine the extent to which a bridge has been damaged and to decide if it must be closed or if it can be safely used by emergency and rescue vehicles.
"Previously, without preemptive media access control, the normal transmission of previously recorded data would not allow this trigger message to propagate through the whole network in real time," says Cheng.
Cheng and Pakzad hope to conduct research on a local bridge, perhaps the Route 33 bridge spanning the Lehigh River, to measure the speed at which WSNs can propagate a trigger message.
The overall goal of the project is to make WSNs competitive with wired sensor networks.
"Wired systems are very expensive when you consider the cost of installing miles of wires on a structure as well as the labor costs of maintaining the systems," says Cheng. "So it’s critical to determine if we can make wireless systems that are as accurate, flexible and responsive as wired systems—or better."
In the second project, Cheng is collaborating with Sibel Pamukcu, professor of civil and environmental engineering, to develop WSNs that identify the properties of sand, soil and other subsurface media while monitoring such indiscriminate geo-events as landslides and chemical spills, and the direction and flow rate of those spills.
Cheng and Pamukcu are seeking to obtain both "local" and "global" snapshots of subterranean phenomena by evaluating in real time the changes in the quality of the signals transmitted along the nodes of a multi-hop wireless sensor network.
A wired sensor network can sense only those events occurring in the areas local to the fibers connecting the network, says Cheng, but a wireless sensor network is not similarly constrained.
"A wireless network is not constrained by the location of deployed sensors because the wireless signal is broadcast among the nodes. This allows us to interpret what is happening globally as well as what is happening locally. One of the questions we’re hoping to answer is how far apart the sensors can be and still yield useful data."
The three NSF-supported projects are providing timely research opportunities for Cheng’s graduate students in Lehigh’s Laboratory Of Networking Group (LONGLAB), which Cheng directs. For example, Eric Xu Li is a Ph.D. student in computer science and Lisa Frye, an assistant professor of computer science at Kutztown University, is a Ph.D. candidate at Lehigh.
Jun Peng, who worked with Cheng as a postdoctoral research scientist, co-invented the idea of preemptive media access control. Peng is now an assistant professor of electrical engineering at the University of Texas-Pan American.
Story by Kurt Pfitzer
Posted on Friday, October 16, 2009