Abstract
At first glance, I was slightly dubious about this book. It is a sensor networks book that concentrates neither on sensing nor networks. Although I am well aware that power conservation is a major issue for sensor network implementation, I was concerned that it was a book on a peripheral subject that inserted the term sensor networks in the title solely to boost sales.
After having read the book and thinking long and hard about the issue, I realize that I was wrong. Sensor networks will generally be embedded in their environment. The following alternatives exist:
Limit node placement to areas with an existing power supply. Install a power delivery infrastructure to support the nodes. Resign yourself to the network having a finite lifetime limited by battery power. Occasionally add new nodes to the sensed region, in order to replace nodes that have lost power. Or, find ways of exploiting the environment to scavenge power.
This book is dedicated to research on the final alternative, which may be the only real solution to the problem.
Of course, my initial reservation could also be viewed in another way. Why not insert another application for energy scavenging in the title? Energy scavenging is needed for systems that are embedded in their application environment. The power requirements for actuators will generally be large enough that scavenging will be of limited utility. Sensors on the other hand, may be able to be sustained by this technology. The more I considered the problem, the clearer it was that energy scavenging and sensor networks are technologies that will complement each other in the future. The book presents research results from Department of Energy and DARPA sponsored research programs performed at The University of California Berkeley. In particular, this work was closely tied to the PicoRadio and Motes research agendas.
In reviewing the literature from these projects, it is clear that the ratio of energy consumed by transmitting a bit of data for a reasonable distance to the energy consumption of computing one instruction on the bit is currently on the order of 102 and is likely to stay that way. While this emphasizes the importance of power aware communications technologies, I disagree with many researchers who discount the energy consumed by computation on the local node. Some of the results in the literature are based on applications with little or no local processing. Tests performed by my students indicate that for current applications and prototypes in excess of 70% of the power consumed is often due to local processing. The case for assuming that this percentage will of necessity decrease in the future has also been overstated. This is especially true when smaller feature sizes are tied to greater leakage of energy and faster processor clocks.
This book is an in depth discussion of the energy issues related to sensor networks. As I mentioned previously, I have grown to find this a very relevant aspect of the problem space. That said, many researchers who concentrate on either the networking or signal processing issues are probably not going to find information relative to their work in this book.
The first chapter of the book reviews the current art of power storage and introduces potential sources of energy for sensor networks. I personally feel that it would have been appropriate to include a review of the relevant sensor network literature at this point. While the book contains references to the relevant articles, a ten-page summary of current results would have greatly helped in developing the context of the book. Except for that criticism, this chapter was extremely interesting and informative.
In the first chapter, the authors show that a particular promising source of energy for sensor networks (especially for researchers near a geological fault line) are ambient vibrations.
The rest of the book considers sources of ambient vibrations and various methods for converting vibrations into electronic power. The conversion approaches considered include induction, capacitance, and the use of piezoelectric materials. Prototype implementations are described, implemented, and tested.
Their testing tends to indicate that commercial off the shelf components are currently designed for systems with higher power dissipation than currently possible using these technologies. In spite of that, the future for these concepts seems particularly bright.
Although electric power systems are outside my realm of expertise, I found the book to be clearly written. It is an example of a good engineering study of a specific problem domain. While his book is not necessarily situated in the mainstream of sensor network research, it provides an attractive, elegant solution to one aspect of the problem. Many researchers will gain important insights by reading this book. I recommend the book to researchers seeking an in depth understanding of all aspects of the problem domain. Other readers will probably be less interested in the technical contents of this book.