The hottest new high-capacity man architecture and

  • Detail

A new high-capacity Metro architecture and its key technologies

Abstract This paper introduces a new next-generation high-capacity Metro technology architecture combining high-speed adjustable devices and wavelength routing, but in the development process of hydraulic universal experimental machine, its key technologies such as media access control, fairness protocol, survivability protocol are analyzed

keywords metropolitan wavelength routing media access control

1. Introduction

at present, the metropolitan technology architectures widely used in the world mainly include SDH and Ethernet. When they use wavelength division multiplexing (WDM) technology to expand the capacity to 1tb/s, in addition to the above-mentioned cost problems, they also have their own defects when applied to the metropolitan environment: SDH is a circuit oriented time-division transmission system, but in the era of grouping, The efficiency of data transmission is not high; Ethernet is naturally suitable for packet data transmission, but it is difficult to be applied to ring topology, and it does not make full use of the ability of the ring, and it cannot meet the requirements of carrier level in terms of quality of service (QoS), reliability, and better mobility, processing performance and scalability

resilient packet ring (RPR) technology, which has just been standardized by IEEE 802.17, efficiently supports ring topology and rapid recovery in case of fiber failure and link failure; At the same time, it can provide efficient, simple and low-cost data service transmission like Ethernet. RPR can also use WDM for capacity expansion (RPR over WDM), which is a widely favored Metro technology. However, the design of RPR does not consider the use of wavelength routing, so when the capacity reaches 1tb/s, the number of wavelengths will increase to more than dozens, and each node requires a large number of transmitters and receivers, which is very expensive. And the packet ADM must be designed to receive data packets from more than dozens of wavelengths in both directions, and the transmission queue must be designed to send data packets to more than dozens of wavelengths in both directions. This design is undoubtedly extremely expensive

with the maturity of high-speed adjustable devices, combined with wavelength routing, a new high-capacity Metro architecture has become a very competitive solution

2. New next generation high-capacity metropolitan area architecture

this new high-capacity metropolitan area architecture is shown in Figure 1: it is a 2-fiber bidirectional ring topology. The node uses a fast adjustable packet transmitter to insert the packet into the ring, and the packet uses the wavelength transmission received by the destination node. Each node receives only data packets of unique wavelengths from within the ring. Because the service is layer 1 optical direct when passing through the intermediate node, it does not need to be processed by the intermediate node, so it can greatly reduce the end-to-end delay and be easier to expand to a higher speed

consider the example shown in the figure. Node i wants to transmit two packets, one to node 0 and one to node 1. Suppose node o is at wavelength λ Receive packets on 0, node 1 at wavelength λ Receive packets on 1. Transmission node i adjusts its transmitter to wavelength λ 0, and some users still use origin and other more professional graphics processing software for later data processing. The destination is the grouping of node 0. Then adjust its transmitter to the wavelength λ 1 and inserted into the packet whose destination is node 1. The wavelength of the packet to node o is not consistent with the wavelength that other nodes can receive, so it transmits directly to node o in the form of light on the ring, where it is finally received and processed. The grouping to node 1 is the same process

in RPR over WDM system, these two packets will be received by the passing node, converted into electrical signals, transmitted to the packet ADM, and then forwarded at each node between the source and destination. In this way, RPR over WDM nodes need a lot of devices because they must receive, process and forward all the packets that have passed. In the new architecture, the node only needs to be divided into its groups, which can greatly reduce the equipment, simplify the structure and reduce the cost

Figure 1 ring structure and node composition

3. Key technologies of the new architecture

the new architecture requires new protocols, such as media access control, fairness protocol, survivability protocol, etc., which are the key technologies of this new architecture

3.1 MAC protocol

because the packet ADM processing in wavelength routing is completely different from that of any current commercial network, it is necessary to develop a new suitable data link layer protocol starting from MAC protocol. The main function of the MAC protocol is to prevent the sender from inserting data packets into the node. In order to avoid conflicts, the MAC protocol should monitor the WDM traffic passing through the node, determine the wavelength available information, and notify the transmitter which wavelength is allowed to be used at a specific time, so that the transmitter will not insert the used wavelength carrying other data packets passing through the node. The most difficult part of implementing MAC protocol is to obtain WDM wavelength status information (i.e. available wavelength information) passing through the node

a scheme is to centralize all control and processing to a master node on the, while the master scheduler regards the whole ring as a high-capacity packet switch. Each node requests transmission from the scheduler, which decides the best plan for all requests, and then notifies each node to send the package. In principle, this is an ideal scheme, because it can avoid conflicts and determine the most fair possible plan for transmitting data packets. However, because the total switching service of the network exceeds 1tb/s, the implementation of this scheme is extremely difficult. In addition, due to the huge geographical area of the network expansion, it will also cause many difficulties: to have a 1tb/s capacity packet switch just receive the current information and immediately transmit the plan to the transmitter, which is certainly not what a network with a radius of dozens of kilometers can achieve

the second scheme is that the node to send data transmits a request to the destination node, and then waits for reception. Only when it receives the information that the destination node can receive will it send a data packet. Theoretically, this adjusts the spacing between the main body and the dynamometer according to the shape and foundation map, and adjusts the direction sample to ensure that the conflict will not occur, and the network is fair to all users, but the problem of realizability still exists when the geographical area of the network is large

at present, the more feasible scheme is to use a control channel to transmit wavelength available information in view of the optical management channel (OSC) in the traditional optical network. In WDM network, the control channel uses its own wavelength. This control wavelength is sent and received at each node so that all nodes can process and change the control channel. In each frame, the bit stream carries the available wavelength information of the following frame. In this way, the node can know the situation of the upcoming frame, so as to solve the conflict problem. The concept of control channel has also been mature in commercial mobile communication networks, so the scheme applied to wavelength routing metropolitan area is also completely feasible

3.2 fair protocol

in the traditional ring architecture, such as RPR over WDM, the electric packet ADM can cache the passed data packets, while in the wavelength routing network structure, the data packets are directly connected to the intermediate node at layer 1; In addition, in the wavelength routing ring, nodes are not end users (nodes contain many wavelengths), and fairness to nodes does not mean fairness to end users. Therefore, we should create a new fairness protocol and algorithm to ensure the fairness of all nodes on the ring and improve the total throughput of the network

from this new structure, it can be seen that there is conflict only when multiple nodes send data to the same destination node at the same time, so this is the problem of competing for the destination node, so the ring can be converted into a FCFS queue, and the IEEE 802.6dqdb protocol suitable for this structure can be used to complete fair control

in IEEE 802.6, when a data packet arrives in front of the sender queue, the node sends a request to the opposite direction of the data packet to be transmitted. The request consists of the request bits in the control information field of the current frame and passes through all upstream nodes. These nodes calculate the requests they know. According to the protocol, a node must allow enough unused frames to pass to meet all requests that are known when a data packet arrives in front of the queue. In this way, the network is approximately simulated as a distributed FCFS queue

the corresponding fairness protocol is similar to IEEE 802.6: only one request is allowed for each wavelength in each control channel frame, and a request counter is used for each wavelength to adjust the fairness of WDM ring of wavelength routing

3.3 survivability protocol

the new architecture shown in Figure 1 can learn from the "redirection" protection of RPR to achieve a more efficient elastic recovery mechanism: there are two paths between two nodes. Under normal conditions, when the access node has data packets to transmit, it selects the better of the two paths to send through a simple routing algorithm. When an interruption occurs, there is only one path to each destination, and the node is forced to use the remaining path. In the new system, the impact of interrupts on the transmission capacity of specific nodes is location related. The transmission capacity of nodes far away from the cut-off point is basically not affected. The closer to the cut-off point, the greater the impact. In the extreme case, the node adjacent to the cut-off point can only use one of the two optical fibers to transmit data

4. Conclusion

using high-speed adjustable packet transmitters and wavelength routing architecture, a new high-capacity metropolitan area can greatly eliminate the excessive need for optical/electrical and electrical/optical converters, high-speed line cards and packet switching capacity, and there is no need to use optical demultiplexing equipment, so as to expand to 1tb/s capacity at low cost. (end)

Copyright © 2011 JIN SHI