[an error occurred while processing this directive]
|[an error occurred while processing this directive]||
Ethernet Passive Optical Networks
Passive optical networks (PONs) have been considered as a solution for the subscriber access network for quite some time, even before the Internet spurred bandwidth demand. One of the first papers describing telephony PON was published by British Telecom researchers in 1988. However, only during the past decade has the optical technology matured enough to make PON-based subscriber access network a reality.
Several alternative architectures for PON-based access networks have been standardized by several standards bodies, one of the main differentiating factors being the choice of the bearer protocol. Currently, standardized specifications exist for ATM-based PON (APON and BPON), gigabit-capable PON utilizing generic framing procedure (GPON), and Ethernet PON (EPON).
Among these PON specifications, EPON is the only access network architecture that traces its ancestry not to public communications, but to private enterprise data networks. EPON is standardized by the IEEE 802.3 work group – a group that rules in the LAN world, but has not ventured much beyond it. The Ethernet in the First Mile project was its first such attempt – a reconnaissance mission into yet unknown territory. EPON took industry by storm, advancing from raw idea to an industry-wide standard in five short years. But will EPON enjoy the same proliferation in access networks that its predecessors have had in corporate LANs? Time will tell.
Motivation for This book
The technical and economic merits of EPON were recognized by the IEEE Standards Association when, in 2001, it authorized formation of the IEEE Ethernet in the First Mile (EFM) task force and approved EPON architecture as one of the objectives of the task force. The EFM task force completed its charter in June 2004, culminating in ratification of the IEEE 802.3ah standard.
In looking at the completed standard, most task force members realize now that EPON architecture is the most significant departure from traditional Ethernet architecture in entire 30-year Ethernet history. The architectural decisions made by the task force may not be obvious to readers of the standard unless alternative solutions are reviewed or some background of task force discussions is presented. Of course, the formal and concise text of the standard does not provide either of these. As a result, many task force members suggested that a book explaining EPON architecture would be of benefit to telecommunications professionals.
An additional motivation for writing this book was the fact that, unlike the areas of responsibility for other standards bodies, IEEE 802.3 only covers a small portion of a communications system, encompassing only physical and media access control layers. Such issues as security, protection, dynamic bandwidth allocation, quality of service, and utilization, are considered out-of-scope for the standard. This book will investigate several such "left-out" issues and provide a comprehensive analysis of state-of-the-art solutions.
Organization of the Book
This book is divided into four parts.
Part I provides an overview of PON-enabling technologies, introduces history of EPON development, and compares EPON to alternative solutions, such as ATM-based PONs and PONs utilizing the generic framing procedure.
Part II provides an overview of a portion of IEEE 802.3ah standard relevant to EPONs.
It is important to emphasize that this book is not intended as a substitution for the standard published by IEEE. Even though, with permission from IEEE, this book reproduces several of the most important state machines, we do not adhere to the paramount level of formality bestowed by the standard. Instead, the book focuses on explaining the ideas behind the EPON specification, where necessary giving specific examples, looking at alternative solutions, or divulging the task force discussions that have led to a particular decision.
The IEEE 802.3ah overview in this book is based on the version of the standard as it was approved by the IEEE Standards Association in June 2004. However, several problems with the specification were identified after the official approval of the standard. Whenever we describe a state machine or a part of the spec with such a mistake, we alert the reader by using the following icon.
The text in this box points reader′s attention to a serious error in the current version of the IEEE 802.3ah standard. It is expected that future revisions of the standard will fix this problem.
The standard is an evolving document. The IEEE 802.3 periodically produces new revisions of the standard by integrating approved amendments and corrigenda.
In Part III we investigate several system-level issues which were considered out-of-scope by the IEEE 802.3ah standard. In Chapter 10 we describe an encryption method optimized for EPON. In Chapter 11 we investigate effectiveness of various path protection schemes in EPON.
Part IV investigates performance of EPON. It is, to a significant degree, based on author′s research conducted at UC Davis Networks Research Lab.
Chapter 12 investigates the various bandwidth overheads associated with EPON, such as optical overhead, framing (encapsulation) overhead, scheduling overhead, and error correction overhead. In Chapter 13 we look at the efficiency of the discovery process in EPON. Then, we turn our attention to various scheduling and bandwidth allocation schemes and their impact on EPON′s ability to support applications and services with diverse requirements. In Chapter 14 we consider the simple scheme with a statically allocated bandwidth. When EPONs first emerged, several companies were advocating for and intended to build EPONs based on static allocation of bandwidth.
Chapter 15 introduces the reader to a simple dynamic bandwidth allocation scheme called interleaved polling with adaptive cycle time (IPACT). Comparing the performance of IPACT to a static scheme presented in the previous chapter illustrates the severe penalties imposed by the static resource allocation on bursty network traffic.
EPON is expected to be a truly converged network, supporting voice communications, standard and high-definition video, video-conferencing, real-time and near-real-time transactions, and data traffic. To support this multitude of applications, EPON must guarantee appropriate performance for each such application. In Chapter 16, we examine how EPON can provide differentiated services by combining IPACT with strict (exhaustive) priority scheduling which is a default scheduling algorithm specified in IEEE 802.1D.
Even though providing differentiated services is what most vendors have in mind when claiming that their EPON equipment supports quality of service, this, in many respects, is not sufficient. In Chapter 17 we take a more formal look at EPON scheduling objectives. We argue that providing service guarantees and fairness, scalability, and isolation requires much more than what priority-based schemes can achieve.
We also find that the above objectives present a conflicting set of requirements. On one hand, the scheduling algorithm should be able to allocate resources to each subscriber individually in order to guarantee SLA and maintain fairness among all the consumers. Hierarchical schedulers cannot be used because they don′t guarantee fairness to the consumers located in different groups. On the other hand, sending control messages to each end consumer is not feasible due to excessive bandwidth consumption and so a hierarchical structure is required.
In Chapter 18, we present fair queuing with service envelopes (FQSE) – an algorithm that successfully achieves both conflicting goals: it uses hierarchical control (i.e., each node receives control messages only from its immediate children), yet is maintains fairness among all the consumers located across different groups.
|[an error occurred while processing this directive]|