Review of Wavelets in Intelligent Transportation Systems by Hojjat Adeli and Asim Karim: John Wiley & Sons Ltd., England; 2005; 224 pp.

This book provides a comprehensive coverage of the application of wavelet analysis, neural networks, and fuzzy logic for modeling transportation operations problems. The authors apply these techniques to solve incident detection and construction work zone management problems. “Wavelets” are used to enhance the computational intelligence of fuzzy logic pattern recognition problems. The book demonstrate the use of wavelets for de-noising using both simulated and real world data. Clear evidence is presented on the suitability of different algorithms for using wavelets in incident management problems. Two questions might arise. First, is there a need for such a book? Second, if there is, is this particular book successful? I believe that the answer to both the questions is an unequivocal yes.

Overall, the book does a commendable job in terms of presenting wavelet techniques for transportation operations problems. While there are many incident detection algorithms in literature, many of them suffer from limitations related to false alarm rate and robust detection. A researcher trying to obtain good incident detection algorithms does not have to search for something better; the algorithms presented in this book provide a definitive answer. The authors have presented the data in a well organized and logical fashion. Chapter and table numbers are nicely coordinated and easy to find through a well organized manuscript. Every time I tried to look up something I found it with ease.

Chapter 2 provides a quick review of wavelet analysis. The authors do a good job in terms of laying out the mathematical foundations needed for the rest of the book. Chapter 3 presents an incident detection algorithm using wavelet transform techniques. The wavelet analysis is used to extract a reliable signal of the incident (without false alarms) from different time locations and having different resolutions. Chapter 4 builds upon the previous chapter to develop a neural network algorithm for identifying incidents. The chapter discusses an adaptive conjugate gradient technique to develop a classifier for traffic incident detection.

Chapter 6 presents a robust traffic incident detection algorithm that provides an effective solution to the traffic incident problem. The main motivation is to increase the detection rate (or reduce percentage of false alarms) by using a fuzzy-wavelet radial basis function neural network. An example for training the algorithm is provided and the algorithm is tested using both simulated and real data. Loop detector data from I-880 freeway in Oakland, Calif., is used to test the detection algorithms.

Chapter 7 compares the algorithm developed in Chapter 6 with the California algorithm #8 on typical urban freeway systems. The algorithms are evaluated using different criteria—detection rate, false alarm rate, detection time, and portability. The algorithms are tested with data from two sources: a Cincinnati-Northern Kentucky area freeway system and I-880 freeway corridor between Oakland and San Jose and under different freeway configurations. It was shown that the fuzzy-wavelet algorithm outperformed the California algorithm for both detection rate and false alarm rate. Chapter 8 discusses an improved incident detection algorithm that effectively distinguishes features produced by capacity-reducing incidents and those produced by recurrent congestion. Using data available locally, it detects the presence of an incident very quickly. An algorithm that takes the time series of lane occupancy and lane speed at the upstream or downstream detector is developed. In Chapter 9, a parametric evaluation of the algorithm developed in Chapter 8 is performed using both real and simulated traffic data. The evaluation is performed to account for false alarm performance in the vicinity of on- and off-ramps. In addition, special considerations for rural freeways to overcome the lack of necessity of having closely spaced detectors and low-flow rates are addressed. The book changes direction in Chapters 10 and 11, which deal with work zone traffic management. A simple tool that accurately calculates the work zone costs (comprised of construction costs, road user costs and maintenance costs) is developed in Microsoft Excel. Different parameters related to work zone type, work zone layout, work characteristics, traffic flow characteristics, phases of work, and traffic control measures are considered in the developed tool. The final chapter provides a mesoscopic-wavelet simulation model to estimate the traffic delay and queue lengths when the capacity of the roadway is reduced. A continuum traffic flow model with discrete time approximation is used to model the movement of vehicles. The elegant mathematical treatise of the problem is noteworthy.

A remarkable feature of this book is that the algorithms discussed in this book were tested with real data collected on freeways and compared with the performance of approaches available in literature. Overall, I found the book to be enjoyable, scholarly, and complete. The book provides original and rigorous approaches to some of the most difficult problems encountered in transportation engineering that have been tackled by other researchers over the past few decades. It has considerably contributed to the furthering of transportation engineering to a new height. This book is likely to be of interest to academic researchers and advanced graduate students in transportation systems.