Browse Books

Reviewer: Bruce Hajek

Significant new material is included in the second edition of this unique and delightful book on the theory of computer communication networks. The additions include up-to-date descriptions of new networks and additional elements of the theory of performance analysis. The book deals primarily with the physical, data link, and network routing and flow control aspects of networks. The authors explain their view of how networks ought to be constructed, and use logical arguments to justify their reasoning. They then compare their concepts to existing networks, up to the most recently developed ones. The chapters are entitled: Introduction and Layered Network Architecture Point-to-Point Protocols and Links Delay Models in Data Networks Multiaccess Communication Routing in Data Networks Flow Control A summary, a section of notes, sources and suggested reading, and a set of problems follow each chapter. A list of references and an index are given at the end of the book. The book provides an introduction to queueing theory that makes good use of graphical arguments. The standard z-transform method is not introduced, but nevertheless an elegant, fairly complete description of delay systems is given, ranging from MM1 and MG1 queues with vacation to priority queueing systems. An analysis of the popular “leaky bucket” method of traffic regulation is included in this edition, and appears in textbook form for the first time. The authors describe several recent advances in network research. For example, the description of random access methods, including “splitting” or “tree” type algorithms, and the description of the theory of routing and flow control are much more extensive and up to date than in other networking texts. The book includes material on distributed broadcasting with unreliable links or without the use of sequence numbers and material on projection methods for optimal routing. Added in the second edition is a treatment of loss networks (circuit switched networks with blocking), an expanded treatment of closed queueing networks, and material on drift analysis, a useful technique for proving the stability of random systems. The brief section on physical communication channels and modems is excellent. It focuses on communication over limited-bandwidth channels. The book does not cover the components of fiber-optic networks. The second edition includes a good overview of FDDI, the fiber-optic token ring that is destined to replace Ethernet as the most popular local area network in the next few years. Asynchronous transfer mode (ATM), the emerging standard for high-speed integrated data transfer, is described and rationalized. Also covered is the promising wide area network standard DQDB, which incorporates ATM. Another new feature is a good description of spanning tree routing, a method for interconnecting local area networks using intelligent bridges. Other popular networks and standards discussed in the book include basic ISDN, HDLC, X.25, IBMs SNA, TYMNET, ARPANET, and the basic IEEE standards for a bus or ring. The discussion of TCPIP and transport-level protocols has been enhanced in the second edition. All the above systems are used to illustrate ways to solve particular types of problems. Though the book does not go into much detail when it describes standards or existing systems, it often provides a meaningful comparison of different systems by stressing fundamental differences. The book is best used as a text for a first course in networking for students who are able to think clearly, logically, and a bit abstractly. A good background in elementary probability theory would be useful to the student. The problems given in the text do not require much background, but are geared toward thoughtful, mature students. Most students in computer science tend not to acquire the necessary reasoning ability and background in probability at the undergraduate level to appreciate the finer points of the book, while most students in more analytically oriented fields such as systems theory or operations research might not be too interested in networking at the undergraduate level. Thus, for either type of student, I think the book is most suitable at the graduate level. It can provide an excellent introduction to networking for analytically oriented students, and an excellent introduction to analytical techniques for computer engineering students. The book is also a valuable reference for researchers working on the design and evaluation of new networks. The book requires no knowledge of operating system design or any other aspect of computer architecture or programming. Many computer-oriented students seem to learn most efficiently through computer interaction. Hence, having them write and test software for a simple automatic request for repetition (ARQ) scheme or other basic networking task would convince them more firmly than any amount of abstract reasoning could that correct protocols are extremely hard to produce. Thus, if the text is used for an undergraduate computer science course, I recommend that the instructor supplement the book with reading material or projects that go into more detail on a few selected network topics. The book covers a broad range of topics, as it must, since the theory of physical, data link, and network layers includes such diverse topics as Fourier analysis, graph algorithms, coding theory, Markov chains, and nonlinear optimization. The most recent network protocols and standards are also included. Nevertheless, the authors judiciously choose topics to focus on, which allows them to discuss issues more deeply than other introductory texts o n communication networks. It is refreshing to see a textbook on networks aimed at providing a lasting education.