Abstract
In the spirit of full disclosure, it should be noted that the publisher of this book, Morgan Kaufmann, has recently been acquired by Taylor & Francis. In addition to publishing this journal, Taylor & Francis also owns CRC Press, which is printing a competing book that was edited by Dr. Iyengar (Editor-in-Chief of this journal) and myself (associate editor). Given these potential conflicts of interest and incentives for both positive and negative reviews, I hope that I will be able to provide a useful review. Remember that reviews are by nature subjective.
I am not acquainted with Professor Guibas. I am well acquainted with Dr. Zhao from his work in the DARPA SensIT program. In addition to being present at each other's program reviews, we collaborated on some issues as part of an initiative started by the program manager, Sri Kumar. At the time, Dr. Zhao was with PARC (formerly Xerox PARC). Their project developed innovative approaches to target tracking and in-network coordination. They also field tested a number of applications at 29 Palms Marine Base and on their campus in Palo Alto. He now works with Microsoft Research and is Editor-in-Chief of the ACM Transactions on Sensor Networks.
It is hard to imagine a more qualified author for this book. In addition to having a Ph.D. from MIT, Feng has worked in both academia and in industry. My understanding is that part of this book was written to support a sensor networks course he was teaching as an adjunct faculty member at Stanford.
Before reading the book, I expected it to be a research monograph documenting his approach to sensor networks. I thought that many details would be provided on the use of particle filters, and the TinyGALS programming language. I was not disappointed; those topics are handled in sufficient depth. On the other hand, I was very pleasantly surprised that the book provides such an excellent overview of the larger field of sensor networks.
It is written in clear, concise, and engaging English. Discussions of technical details were written in a way that I found understandable and not overly taxing. The book was fairly self-contained. I would think that most upper division students in technical fields could follow it with some effort. I was quickly able to grasp the topics that are not in my area of expertise, so I suspect that the topics that are in my area will also be accessible to others. I think that it would be an excellent textbook for a course on sensor networks. It does not contain exercises, which could be a drawback for those interested in using it to support a course.
The book's structure is very appropriate. It starts with an overview of the field that sets the stage for the rest of the book. The next two chapters give background on sensing and wireless networking. I was particularly impressed with the section on estimation theory, and found that it provided me with some new insights. I suspect that many readers will end up using this chapter as a reference that summarizes many essential results in the field. The concepts presented tend to be based on Bayesian analysis. I found the depth of the chapter appropriate. If I were to change anything about this chapter, I think that it would be useful to add a brief section discussing issues related to the existence of false positives and false negatives in sensor outputs. This could include a brief overview of some of the basic physics behind many sensing systems.
The networking section is well done and provides an overview of many of the more innovative wireless protocols that have been proposed for sensor networks. In my opinion, the chapter managed to hit most of the important and influential concepts, although some researchers may be disappointed that their works are not cited. This chapter may become dated more quickly than the rest of the book, which is always a danger when writing a book on a timely topic. In addition to reviewing the networking advances achieved by sensor network researchers, the authors also provided the basic concepts needed to understand the results.
The infrastructure establishment and sensor tasking sections seemed to me to be the most closely associated with the authors' own research. Both are well written and well motivated. Probably because the sensor tasking section is most closely related to my own research, I found that there were many aspects of the approach that I would do differently. Since this is an area of ongoing research, it is obvious that the final solutions have yet to be found and different approaches will exist. When teaching a course based on this book, it is probably a good idea to emphasize that the solutions presented in the text are not the only approach that could be used.
Chapter 6 discusses sensor network databases and is a good review of the current art. It discusses the Cornell Cougar and Berkeley TinyDB approaches. The write-up is very good. Since TinyDB is available for download, it would be possible to use it for in class projects.
Chapter 7 discusses existing sensor platforms. It is particularly interesting, but oriented very strongly towards the Berkeley mote platforms. This is a drawback, because other platforms do exist. It is also to be expected that sensor platforms will evolve quickly in the near future. I suspect this chapter will be dated before the second edition of the book is printed. (I am assuming that there will be a second issue. The quality of the book warrants it.)
The final chapter describes existing applications and future directions. I found it to be relatively complete and accurate.
This book is well worth the time and effort to read it. I recommend it for both researchers interested in the field, and for use as a sensor network textbook.