EE 228 a - Lecture 1 - Spring 2006 Introduction

In this course we survey some of the key recent ideas in networking, with an emphasis on quantitative methods. We also attempt to get a sense of where the field is going. In this introduction, we review historically how the major ideas appeared and their relationships. We then propose a partial panorama of networking research. I. KEY NETWORKING IDEAS Table I lists a number of key ideas in networking and proposes a classification of the contributions. In that table, A, T, S, C refer to Analogy, Technical, Scientific, and Creation, respectively. An analogy is an idea similar to one used in another field, not after the fact, but as a source of inspiration for the contributor. A technical contribution is an elaboration of another idea performed by adapting it to a different situation. Implementation details and variations of operating rules belong to this category. A scientific contribution is one that requires the discovery of new scientific principles. Finally, an idea is a creation if it has no analog but does not require a new scientific breakthrough. TABLE I CLASSIFICATION OF KEY NETWORKING IDEAS Idea Contributor Type Sampling Nyquist S (sampling theorem) Digital communication Shannon S (source and channel coding) Coding Hamming, RS, BCH, Gallager, Berrou, Glavieu S (algebraic, convolution), C (turbo) CDMA Hedy Lamarr A (player piano) Packet Switching Kleinrock A (mail system) Aloha Abrahmson C (multiple access through collision resolution) CSMA/CD Metcalff T (Aloha), C (carrier sense and backoff) WiFi Alex Hills T (CSMA/CD and busy tone scheme) TCP/IP Cerf, Kahn, Jacobson T (packet switching, backoff in CSMA/CD) Shortest Path Dijkstra, Bellman-Ford S (principle of optimality) OSPF, RIP John Moy, C. Hedrick T (shortest path) VoIP, SIP Many T (packet switching, compression) We will complete this table during the course. For now we can note that the table shows that non-specialists could have discovered few of the ideas. To discover most of the ideas, one requires a detailed understanding of the domain, plus a sufficient dose of inventiveness. Obviously, the different types of contribution typically interact and amplify each other. Another lesson from this table is the importance of creations. Although these contributions may not be scientifically or technically deep, they are typically disruptive and have a great impact. II. PANORAMA OF NETWORKING RESEARCH Figure 1 attempts to sketch some major areas of research in networking in 2006. We realize that networking encompasses a wide range of systems and services that correspond to many areas of research. Accordingly, we do not pretend to make an exhaustive list. A. Architecture At the big picture level, one finds a set of questions related to the services that the network provides and how the network components are organized. This level is generally called architecture. Service Model: One first question concerns the service model. The Internet service is “best effort,” which means that the network tries to deliver the packets without any further specification of the delivery service. The telephone service model is the other extreme in terms of tight specification of availability, connection setup time, delay and delay jitter. Figure 2 attempts to illustrate the best effort model of Internet and a differentiated services model. One issue is whether the service should be specified or whether a choice should be offered to the user who then determines which service to pick based on observed performance and price. Structure: The organization of the Internet into components is arranged in layers, as shown in the left part of Figure 3. In this model, layer n implements its services on top of the services of layer n − 1. Each layer specifies its services and its interfaces. Other models are conceivable, such as the middle part of the figure. The right part shows a model of “parameterized” layers where layer n can figure out the characteristics of the services of layer n − 1 and select the ones it prefers. Admittedly, these figures are vague.