The Study Of GMPLS Based Intelligent Passive Optical Network

The passive optical network (PON) has been becoming the most promising access network technology because of the features of good performance, low-cost and easy deployment. However, the most important problem is that current PON technology is still a simple layer 2 data pipe; it lacks capabilities of traffic identification and intelligent resources control. It can not satisfy the variety service demands from end users and provide good business mode. From another view point of next generation network (NGN), the future network must have intelligences. Network can identify user traffic and make automatic resource distribution. Therefore, this thesis proposes the concept of intelligent PON. Through the control functions, PON can provide traffic identification and the intelligent resource management.This thesis first introduces the PON technology, and then the intelligent access network - generalized multi-protocol label switching (GMPLS)-based PON (GMPLS PON), is proposed. GMPLS provides a common intelligent control plane which is able to manage the resource allocation for most kinds of networks. We propose to push the GMPLS core functions and part of the protocols into PON to establish an intelligent access network. Using GMPLS, a common control plane which interfaces the NGN seamlessly can be established in PON. By this control plane, PON can intelligently identify and control the user traffic. Therefore, an integrated end-to-end service provision is enabled in our proposal. Before getting serviced, users can use signaling to request services from service provider. Service provider has to make admission control decisions according to the user's request and current network resources. This thesis also gives the designs and solutions of functional block, protocol layers, framing, signaling and system structures of GMPLS PON. The intelligent control plane and connection-oriented service management also help PON deploy dynamic service level agreements (SLAs). With dynamic SLA, users can establish and modify their SLAs with service provider online. Resources are allocated intelligently by dynamic SLA to guarantee the QoS of each service. In order to fulfill the dynamic SLAs, this thesis proposes a two-layer dynamic bandwidth allocation mechanism. In this mechanism, a kind of configurable ONU scheduler and a traffic prediction based DBA algorithms are combined to ensure the dynamic SLAs are accurately executed.This thesis also discussed the call admission control (CAC) algorithm in GMPLS PON. An improved optimal CAC algorithm is studied. At last, the media access control chip design is presented.Access network is the"first mile"and"last mile"between users and the core network. This thesis proposes the concept of the intelligent access network to meet the developing trend of next generation network. The access network is able to intelligently identify user traffic and allocate the resources accordingly. Our solution points out a promising direction of next generation access network technology.