New Generation intelligent metropolitan Area Optical Network

(Fan Zhongli Nanjing University of Posts and Telecommunications, Nanjing 210003)

A standardized optical control plane is introduced in this paper. The optical control plane is decomposed into a few basic processes such as Neighbor Discovery, service discovery, Connection control and Topology/resource discovery. Different network organization and segmentation result in several different automatic configuration models: Soft persistent link model, user interface model and Peer-to-peer model. Finally, this paper introduces the intelligent Optical Network of the new Generation Ciena company based on SDH Optical Exchange, which has the functions and characteristics of large capacity optical switching capability and network topology automatic discovery, End-to-end circuit configuration and bandwidth dynamic allocation, which will greatly improve the service quality of data and circuit services.

Keyword control plane automatic configuration routing protocol distributed Network Intelligent DWDM Metropolitan Area Network

1 Introduction

In the cluster based NGN (next Generation Network), the crisis of circuit-switched network is obvious. For the major operators, the expectation of NGN is not to tear down the existing network to create an ideal NGN model, but how to evolve from the existing network to NGN, and strive to maintain the dominant position in the increasingly competitive business market. Obviously a standardized optical control plane is the basis of the control plane of Ason.

A well-designed control plane can quickly and accurately establish circuit connections, enabling service providers to better control their networks. The control plane itself must be reliable, scalable, and efficient. The control plane structure should be universally adaptable to support different technical tools, different business requirements, and the capabilities provided by different device providers.

The control plane should be applied to a variety of different transmission network technologies (e.g. SONET/SDH, OTN, PXC). In order to achieve this goal, it is necessary to isolate technology-related aspects from technology. The control plane should be flexible enough to apply to different network applications. To this end, the control plane can be divided into different parts, the equipment manufacturers and service operators can determine the specific location of these components, but also allow service operators to determine the security and policy control of these components.

The control plane should be able to support the basic connectivity functions of the switching connection (SC) or soft permanent connection (SPC) in the transport network. The types of connection features include one-way point-to-point connections, bidirectional point-to-point connections, and bidirectional point-to-point connections. Different network organization and segmentation result in several different automatic configuration models.

2 automatic configuration

Telecom industry has recognized the need for high bandwidth link automatic configuration, based on the existing infrastructure of operators, the potential of developing new products and future strategies, three different models can be selected.

(1) Soft persistent link model

In this model, there is no network management or control interoperability between the terminal system (client) and the network. The network management system at the top of the control plane is used to connect nodes at both ends of communication. Therefore, the SPC model is important for connecting legacy devices to the optical core, as shown in Figure 1, where ATM and FR switching interfaces are connected to the optical core through a network management system. This model has been used in the permanent virtual circuit (SPVC) service of ATM and is recommended by the MPLS network.

Fig. 1 different configuration models in optical networks

(2) User network interface model

The user network interface model (the username network Interface model) is similar to ISDN. In these networks, services are initiated by a terminal system. Figure 1 depicts a router network requesting high-bandwidth connections from the optical network via uni. In the UNI model, the terminal system does not understand the topology and resource conditions of optical networks, but simply requires the establishment or deletion of connections. In some network applications, clients request different routes for different connections [1]. Because the network and the terminal system do not share the topological information, in order to satisfy the diversity demand of the terminal system, Uni must support the "diversification route".

(3) Peer-to-peer model

In the Peer-to-peer model, the initiator's connection request is always for the peer network element, that is, the requester needs to fully understand the topology information. With this information, the connection initiator can select routes through the optical network according to a series of rules, such as the multiplicity of routes, minimum delay, maximum reliability, or minimum hop count.

The peer model is greatly affected by the IP network. In IP networks, routers can be viewed as peer entities of the optical-layer cross Connection (OLXC), sharing all information between OLXC and routers. This is consistent with the mplambdas of the IETF [2]. Routers in separate subnets in the Peer-to-peer model described in Figure 1 Act as peer entities of the optical network. However, for all nodes, not all of the information is required, such as the IP routing table, which range of information needs to be shared is still in the study.

3 Signaling and routing protocols and distributed network intelligence

The essence of signaling systems is the actions that can be requested, the characteristics associated with the connection, the protocols used to deliver actions on the network, and the channels that carry signaling messages.

Establish or delete connections as required, state queries, and attribute modifications [3], which are the four basic actions for identifying optical networks. These features are required for connection requests, as well as customer and connection authentication, source and destination addresses and ports, and security objects.

Fig. 2 Signaling and routing protocols and distributed network intelligence

Many device/service providers recognize the importance of intelligent optical routing, and have jointly developed signaling and routing criteria, such as the Gmpls of the IETF (Internet Engineering Task Force) (Generalized multiprotocol Label switching), it mainly completes the discovery of the neighboring nodes, the broadcast of the link state, the calculation and maintenance of the topology of the whole network, the management and control of the path, the calculation of the route index value, the protection and the recovery, etc. ITU-T introduced the g.7713.1 based on Pnni in February 2002, which is the first draft of Ason. The distributed intelligence of Optical network relies entirely on optical routing and Signaling Protocol to replace the traditional intelligent network management, network topology discovery, circuit automatic configuration and so on is the main embodiment of distributed intelligence. Unlike IP routing, optical routing is not routing and forwarding packets, it mainly plays the role of circuit configuration, when the circuit is formed, only the path management and control.

The optical routing signaling Protocol is an extension of the OSPF protocol in IP network, which keeps the topology map of the whole network on each network element, which provides the foundation for the optical network to realize distributed intelligence, and can provide the network intelligence and function as follows:

* Through a single network element can see the whole network topology, can monitor the situation of the network;

* The network element and the network element may establish the circuit through the Protocol, also may by the configuration individual network element, realizes the End-to-end circuit configuration;

* In the End-to-end circuit recovery to achieve path lookup, once the need for End-to-end circuit recovery, the network element based on the topology and bandwidth situation to find the path to achieve recovery;

* Provides virtual capacity, through topology and computation, can achieve any cascade, wavelength bundle, the formation of non-standard bandwidth, not continuous, not even in the same fiber or light bandwidth can also cascade, when the capacity exceeds the bandwidth capacity of light wave, can also be used to provide more bandwidth capacity of optical bundles (such as 40gbit/ capacity of S).

Distributed intelligence distributes the network intelligence to the network element, instead of using the intelligent of the network element configuration. Compared with the intelligence of Network management, distributed intelligence has the following advantages:

* Network elements can directly know the network physics, distributed intelligent implementation speed, rapid, strong network viability;

* When in-band, Out-of-band network management failures, network-based intelligence can not be implemented, and distributed intelligence is not affected.

4 neighbour finds

All models have a very similar requirement to understand at least what terminal systems are connected to the network, which network element (such as OLXC) is neighbors, and how the network element is connected when the port is interconnected. We call this process a neighbor's discovery, and it should be automatically implemented. In Figure 2, we describe the neighbor discovery process in a simple example.

The Neighbor discovery process is used to determine node and port identities. The node identity is used to unify the identification of nodes in the network, typically some type of address, such as an IP address. The port identification is used to unify the transport ports that identify the ends of the adjacent interfaces. For example, in Figure 2, node 200 should know his node/port pair (200, 3 is connected to node 2112 of node/port pair (2112, 1), similar, (200, 4) connected to (2112, 5), (200, 62) connected to (1701, 3).

Figure 3 SONET/WDM Neighbor Discovery Sample

Here are a few ways to discover your neighbors.

(1) Discovery of the same layer

When a neighboring device shares the common level of a reusable structure, such as a SONET access multiplexer connected to the SONET path switch interface, the automatic Neighbor Discovery option is determined by the functionality of the reuse structure layer.

Suppose we have a sonet line (SDH multiplexing segment) termination device, and both ends of the link support the line DCC channel High level Data link control (high-level the Data Link CONTROL,HDLC) package process. On the Internet, the PPP protocol provides a common communication protocol. PPP requires FULL-DUPLEX communication and therefore cannot be used in one-way links. However, the data transferred on PPP is not necessarily symmetric. ODSI's neighbor Discovery and address registration Draft [8] details the usage and expansion of the PPP application. The further PPP Link Control Protocol (link controls PROTOCOL,LCP) expands and authenticates information that can be used to debug connection errors in the input/output fiber.

(2) wrong layer and one-way discovery

If both ends of the link run at different levels of the reuse hierarchy, such as performing a multiplexing function at one end or providing a transport service, this is essentially the same problem as a one-way neighbor.

An example of a SONET device (user) connected to a uni WDM device (network) is given in Figure 3. In this case, the WDM device plays the role of the physical layer regenerator, that is, performing the photoelectric conversion, regenerating the waveform, and then performing the electro-optic conversion. WDM devices are transparent to sonet overhead, but can passively monitor sdh/sonet segment overhead. Not all overhead can be inserted into information, such as J0, B1. This allows the topology information from the SONET system to the WDM device to be one-time only.

In the example in Figure 3, the Topology Information (node number, port number) can be transferred in-band between each SONET and WDM device's links. Mainly by segment overhead bit J0. After the information is transmitted, the uni side of the network has subsequent connection mappings: (1701,1) * (2112, 3), (1701,3) (2112,7), (1701,4), (2112,1) and (1701,12)? (2112,2).

In the opposite direction, that is, from the network to the user, the only option is to establish an out-of-band communication channel. If the user's topology information contains an IP address, the network can then start a program to establish an Out-of-band communication channel.

(3) Service discovery

The concept of service discovery is very close to the discovery of neighbors. Through the service discovery, the neighboring network element can understand the "service" provided by each network element and determine the optional interface. For example, a OC-48 connection was established between the two SONET/SDH network elements, and neighbors were "found". As suggested in ODSI service discovery and address registration [8], service discovery can be used to determine whether a signal interface is provided by one of the network elements. Note that this message is also used for network element communication in the UNI model and Peer-to-peer model (such as OLXC to OLXC).

Another important function of service discovery is to get the details of the interface restrictions. Consider the example of OC-48 again, assuming that one network element is a router and the other is a SONET/SDH switch. The router's interface now supports only sts-48c signals, but future-enabled interfaces may support more, for example, a sts-48c or four sts-12c, which makes it important for adjacent network elements to know the limitations or capacities.