Wireless Sensor Networks: Structure and Protocol
A problem with any discussion around wireless sensor networks is that—like so many other bleeding-edge technologies—few people can say with certainty what they really are. Of course, this doesn’t mean people will be discouraged from attempting to use them. In fact, for many of the experts who work with wireless, this is a very exciting stage of development, one rife with opportunities for innovation.
“Currently, wireless sensor networks are generally considered to be at their infancy,” said Seapahn Megerian, a specialist in networked embedded systems and assistant professor in the Electrical and Computer Engineering Department at the University of Wisconsin in Madison. “This means that although you are not very likely to see them being used around you right now, there is a very strong impetus in terms of academic research and industrial support that will undoubtedly change this in the near future. The beauty of forward-looking academic research in a new area such as this is that you are typically not constrained by existing standards and guiding entities. But on the down side, this leads to contributions and results that often cannot be made to coexist without significant efforts. I would say a large number of the challenges now—both in academia as well as industry—deal with trying to come to a consensus on what a ‘sensor network’ is and what the applications are.”
A rough explanation of the structure of wireless sensor networks would simply be an arrangement of RF transceivers, sensors, machine controllers, microcontrollers and user interface devices that communicate with each other via two or more nodes. However, this description leaves something to be desired and doesn’t really explain how these interrelate, nor could it. “Wireless sensor networks define a truly vast multidisciplinary domain that encompasses technologies starting from the physics and device levels to circuits, embedded processors, storage, wireless communication, networking, system integration, middleware, operating systems, security and application software. Consequently, a large number of guiding entities can potentially come in play when all of these components are brought together to form a functioning system,” Megerian said.
Likewise, it’s not easy to pin down universal protocols and best practices for wireless sensor networks. In this way, this sphere is similar to wireless technology in general. “There are a number of existing standards, new and emerging standards, and guiding principles for each of the components that make up sensor networks.” Megerian explained. “Many of these areas are quite mature and have been studied in-depth in the past decades. At the same time, there are also a number of newer general principles that have started to guide the research in the area of sensor networks. Various consortia are emerging to try to bring some order, but exactly how this will evolve in the future is unclear.”
Two of the main identified problems with the operation of wireless sensor networks are constraints in the energy that powers the nodes, Megerian said. Generally, these nodes, which can number in the thousands, are battery-operated. Because they often monitor austere and inaccessible locations and there are so many of them, the people using them today have to adopt a kind of “spray-and-pray” approach. “Generally, we do not assume that individual sensor nodes will be reliable. Thus, the common algorithms and protocols of sensor networks often have features to automatically recover from node failures, either due to energy exhaustion, hardware failures or other unforeseen events. Furthermore, the performance of the wireless communication links between the sensor nodes may be quite unpredictable. This adds another dimension of uncertainty in how we design, maintain, troubleshoot and fix wireless sensor networks.”
Megerian used a somewhat transcendental description to explain wireless sensor networks and their (for now) latent capabilities. “I think it helps to think of wireless sensor networks as the missing link between our informational (computational) worlds and our physical reality,” he said. “Sensor networks enable us to see and learn things in our daily environments in ways and at granularities that have never been possible before. Virtually all scientific branches can now have access to the kinds of data that are fundamental in making significant experimental advances in our understanding of the world. These advances will then undoubtedly lead to further, and potentially profound, theoretical advances, much like we have seen in the past: the synergy between experimental and theoretical progress and how one almost always seems to feed off of the other.
“But along with this newfound power to observe come a number of pitfalls that will definitely need to be addressed,” he added. “First and foremost, privacy issues stand out any time sensors are placed in any way that can observe human activities. Furthermore, security and protection of the information presents a number of non-trivial challenges. Sensor nodes in such system are, by design, exposed to a wide array of elements that can include eavesdroppers, malicious attackers and hackers.”
But will there ever be a substantial market for wireless sensor networks? In Megerian’s opinion, almost certainly yes. The watchword for this sector is “potential,” he said. The applications of these systems in the future could include monitoring of houses and offices, observation of the environment for advanced warnings of natural disasters, military surveillance and even exploration of the planets and moons in our solar system. “Basically, with sensor networks, the technologies have been advancing at a much faster pace than the applications and the potential users have. So given time, it seems quite likely that the applications (and the potential users) will catch up and wireless sensor networks will find their way in all sorts of public, consumer, commercial and other environments.”
–Brian Summerfield, firstname.lastname@example.org