1. Preface

The focus of attention of communication carrier in market competition is being shifted from backbone to metro network. As a bridge linking backbone network and service aspect, optical metro network plays more and more important role in entire telecommunication network. The mainstream service in network is also changing profoundly. Since in most of current metro networks SDH/SONET dominates, so in order to carry and support IP related services, most packet-based data transmission and switching predominantly makes use of SDH/SONET network. However, in such approach there are two major drawbacks: Firstly, existing SDH/SONET system is physically and functionally layered into regenerator section, multiplex section, higher order path and lower order path, mapping packed data onto higher order or lower order path (or their concatenation) is rather costly and inefficient; Secondly, since VC concatenation or virtual concatenated path is adopted in physical layer, when contiguous concatenation is used, bandwidth cannot be changed; while for virtual concatenation, bandwidth can only be changed by network management provision or configuration. Due to burst nature of IP service, difference between channel bandwidth and service bandwidth some times is very considerable, and it is hard to make them consistent.

On top of this, in currently prevailing optical metro network construction, Ethernet Layer-2 and Layer-3 switches and routers are commonly widely applied. In pure technical perspective, such application mode has some shortcomings:

1) Real-time support is strictly limited: although various network traffic flow congestion control technique and quality of service (QoS) guarantees are developed, Layer-2, Layer-3 switch and router still bring significant time delay and delay jitter;

2) Carrier-class path protection can only be implemented in primary ring or backbone network, while tributary-based protection (including 1+1, 1:1, 1:N) cannot be accomplished within 50ms. Tributary-based broadcast, binding and mirroring also cannot be realized;

3) Accounting nodes and accounting unit message points (packet-based, flow-based, time length-based) are dispersed in whole metro network, therefore, both maintenance cost and workload are high. The accounting processing based on packets, flow and time length takes up significant part of available resource;

4) Coexistence of SNMP-based information model and CMIP (common management information protocol)-based information model brings about confused network management system.

Form viewpoint of idea “Combining networks three-in-one”, IP technique is recognized by public as unified platform for future networking, and “IP over everything” and “Everything on IP” is just a manifestation of such consideration. Undoubtedly, regarding IP as footstone of future public network is appropriate. However, because IP protocol is in the higher layer, hence “Combining networks three-in-one” can only be implemented in higher service layer, and this will bring difficulty to network design and deployment especially in R&D of equipment. For instance, for transporting voice and video service on pure IP network, only “Voice over IP” and “Video over IP” can be considered. But because of inherent shortage of IP protocol, how to ensure QoS of these real-time services is a cumbersome task. Although various network traffic flow congestion control technique and quality of service (QoS) guarantees are developed, but implementation of these techniques is costly. 

In order to overcome abovementioned drawbacks and to lift limitation to networking under current optical metro network architecture, FiberHome Networks put forward a concept on next-generation optical metro network (NGN), based on its own understanding and experience. FiberHome Networks thinks that the next-generation optical metro network should be an integrated network combining data, voice and video together in best way. Network integration should cover all network aspects including data network, transmission network, broadband access, integrated service, and management service and network operations platform etc. Such integration requires creation of new type, high quality, high reliability, intelligent and cost-effective network architecture so as to provide user with most suitable (cheapest and most effective) network and service. 

This new type, high quality, high reliability, intelligent and cost-effective network architecture was proposed by Dr. Shaohua Yu of FiberHome Networks on behalf of Chinese administration. The new network architecture concept has been accepted by ITU-T and IEEE in a draft Multiple Services Ring (MSR) optical metro network standard, designated as ITU-T Rec. X.msr. 

2. General introduction to MSR

Figure 1 Basic protocol stack of MSR

1) MSR protocol

MSR is a new type networking and application approach for optical metro network. MSR adopts MSR-LEP (Link Encapsulation Protocol Used for Multiple Services Ring, see Figure 1) as its communication protocol, and uses cheap Wide Area Interface Sublayer (WIS) of Gigabit Ethernet and 10G Ethernet as its optical transmission mechanism. In data link layer, MSR aggregates different services ? data (borne by Ethernet and GE etc.), voice and video (borne by TCE (TDM Circuit Emulation) or direct encapsulation), different tributary form (FR, ATM, ISDN, DDN, PDH, SDH, DSL, Ethernet etc.) and different bit stream (service data stream, network management (NM) data stream, CS data stream etc.) together coherently, thereby to accomplish aggregation and unified encapsulation of various data information in baseline layer. MSR-LEP is a data link protocol located between MAC/TCE (or PPP/IPv4/IPv6) frame (packet) and physical layer. This protocol is mainly used for communication among different nodes in MSR ring. 

2) MSR topology structure

Typical topological structure of MSR is a ring network consisting of a bi-directional, symmetrical, counter-rotating fiber pair (see Figure 2). There are at least two nodes and possibly maximum 32 nodes in the ring. One or more independent tributary traffic can be added/dropped in every node (such as Ethernet, Gigabit Ethernet, TCE etc.) in MSR ring. Just like router, every node is also able to transmit and receive Layer-3 data packets (IPv4/IPv6), control signaling (CS) packets, and network management (NM) packets. MSR supports unicast, multicast and broadcast of these tributary services. Any node in the MSR ring can be dynamically added or deleted in-line. When the second ring is designated in standby mode, in-line adding or deleting a node will neither affect other nodes that are working, nor cause loss of data packets or service interruption. The opposite situation is also true.

Figure 2 Main topology structure and system composition of MSR

3) Composition of MSR system and main elements

The main elements in a MSR ring system include aggregate pipe, tributary, the first working ring (FWR), the second working ring (SWR), and MSR node equipment.

MSR aggregate pipe is defined as two symmetrical, counter-rotating fiber channels linking MSR node equipment. Traffic in the channel can be any one of the following: STM-1/OC-3, STM-4/OC-12, STM-16/OC-48, STM-64/OC-192, VC4, VC4-4c, VC4-16c, VC4-64c, virtual concatenation of VC4 or VC3, and Gigabit Ethernet or 10G Ethernet. Bandwidth in the pipe depends on service requirement. If bandwidth protection is required, the total bandwidth (sum of all tributary bandwidth) in a pipe should be less than a half of bandwidth capacity of the aggregate pipe. In some particular cases, it is allowed to have sum of all tributary bandwidth exceeding a half of the aggregate pipe capacity. Normally, it is recommended that aggregate pipe of different span that have same total bandwidth take up a same working ring. But it is also allowed that aggregate pipe of different span that have different total bandwidth (for instance, GE and STM-16/OC-48) take up a same working ring.

MSR tributary is independent service add/drop channel accessing MSR node. Multiple, diversified tributaries together act as nerve tips of a MSR system used to sense, accept or deliver various multimedia services. Different tributary can be assigned with different priority value depending on service type it carries. The tributary type, frame type and service capability supported by MSR node are listed in Table 1.

Table 1 Tributary type, frame type and service capability supported by MSR node

The default setting of FWR is the outer MSR ring. FWR and SWR can alternate and can be mutual standby. In case of fiber cut in SWR or in case node equipment is in failure, FWR can become the bypass of SWR so as to ensure uninterrupted service running.

MSR node equipment consists of MSR filtering unit, MSR schedule unit and MSR framer. Its main function is to accept, deliver and forward NM frame, CS frame and data frame. Framer is the key element of MSR node equipment, it is the embodiment of basic thought and philosophy of MSR. 

Figure 3 Generic frame format of MSR

4) MSR framer and generic frame format

MSR framer performs framing function according to MSR generic frame format. The MSR generic frame format is shown in Figure 3. Every MSR-LEP packet adopts fixed overhead and consists of following fields:

a) DNA (Destination Node Address) field: it takes up 32 bits and is used to identify node address in MSR ring. DNA is a local address and only has local meaning in MSR ring. 

b) TTL (Time to Live) field: it takes up 5 bits and is used to calculate the jump count number ? each time when the node in the ring forwards data packet, the count number is reduced by 1. 

c) U/M/B (Unicast/Multicast/Broadcast) field: it takes up 2 bits, 01 represents unicast, 10 represents multicast, 11 represents broadcast, 00 is reserved. 

d) FWR/SWR field: it takes up 1 bit and is used to identify in which ringlet data packet is transported, 0 and 1 represents FWR and SWR respectively. 

e) Priority field: it takes up 3 bits and is used to reflect priority of MSR data packet: range of priority level is 0?7, the large the value, the high the service priority. The priority level of service is set manually in sending end of the node using network management interface before equipment installation, based on service level agreement of carrier. The priority level can be changed in-line. 

f) Reserved field: it takes up 5 bits, reserved for future use. 

g) TT (Tributary Type) field: it takes up 16 bits representing type of added/dropped tributary channel signal to/from MSR data nodes including Layer-3 forward data packet, control signaling packet and network management packet. At present, following types of tributary channel are supported:

• G.707 SDH circuit
• G.702 PDH circuit
• Video signal
• Voice-band signal
• Digital channel supported by 64 kbit/s-based ISDN
• TCE
• Ethernet (10/100Mb/s, specified in IEEE802.3)
• GE (specified in IEEE802.3)
• L3 Forwarding Packet
• CS & NM Frame

h) TN (Tributary Number) field: it takes up 16 bits representing number of same type of tributary ports within a MSR data node. 

i) CS (Control Signaling) and NM (Network Management) field: it takes up 8 bits and is used to indicate type of overhead and signaling packet and network management packet including network topology discovery, L2PS (Layer-2 Protection Switching) request and response (indication), error report, performance inquiry and report, synchronization request and acknowledge, connect request and acknowledge etc. 

j) FSN (Frame Sequence Number) field: it takes up 8 bits and is used to determine transport order number and count number in frame sequence of Ethernet data packet, Gigabit Ethernet data packet, TCE data packet or IP packet related to L3 forwarding. Its main purpose is to provide performance monitoring function for packet loss or duplication of TCE based tributary. 

k) Payload field: all the services mentioned above can be encapsulated in this payload field (applied to L3 forward packet based tributaries and nodes). It embodies the core thought and philosophy of integrating transmission and switching and realizing “combining networks three-in-one” of MSR ring. The payload can be Layer-3 data packets or can be control signaling and network management data packets. 

l) FCS (Frame Check Sequence) field: it takes up 32 bits (4 octets). FCS cyclic redundancy check is calculated for all bits starting from current DNA field through fields of TTL, U/M/B, Priority, TT, TN, CS & NM, to the Payload field (or parameters corresponding to CS and NM packet data). It is used to check correctness of framing.

3. Market orientation and MSR solution

Market orientation of MSR is clear, that is, it is oriented to optical metro network that can provide integrated services of carrier-class quality. Major domestic and overseas carriers have focused network construction on metro network. Market share in semiconductor chips and equipment related to metro network may account for tens of billions of dollars globally in next several years. Due to the fact that MSR is able to integrate data, voice and video services in two lower OSI layers, and is compatible to old network architecture and existing infrastructure thereby to protect investment already deposited, and is also able to accomplish unified network management in cost-effective manner, therefore, the network architecture and application solution provides by MSR represents future development trend of optical metro network.

As for network architecture and topological structure, MSR can be used in core layer and backbone ring of optical metro network to provide large capacity, high-speed carrier-class backbone channel of metro network. It can be equally used in aggregation layer of metro network acting as aggregation equipment for multiple services and to provide link converged channel. It can also be used in access layer of metro network to directly support various matured access methods including Ethernet, SDH/SONET, ATM, RPR, FR, DDN, and DSL etc. (see Figure 4). In future, it will support direct access of multiple services terminal equipment of MSR. MSR provides flexible network routing: besides common ring topology, MSR can be used to form linear chain topology according to practical fiber route situation. By flexibly using linear chain MSR + ADM, it can support access and aggregation of multiple services more effectively. MSR access ring and backbone ring can be nested mutually. The two symmetrical counter-direction and counter-rotating rings both can be used to transport data frame, CS frame and NM frame simultaneously. MSR has automatic topology discovery function and performance monitoring function, and can realize dynamic node adding and deleting.

As for service application, MSR can directly carry data, video and TDM signals, “Combining networks three-in-one” can be implemented in chip level. Since services in MSR ring are transparent, MSR can integrate transmission equipment and switching equipment in one entity. Thus, complexity of transmission equipment and data switch equipment is greatly reduces, user investment and operation cost is much lowered, and equipment operation and maintenance is simplified. All tributary services, control signaling frame and network management frame can provide priority queuing and QoS classification function.

Future optical metro network will enter a new era of personalized service and intelligent management. MSR just caters for such trend. MSR ring and services running on the ring are resilient: bandwidth is scalable, data speed can be high and low, service variety can be changed and can be automatically expanded in real-time according to user or service demand. Therefore, user confidence and trust in MSR can be built and maintained.

Figure 4 Application combination of MSR in optical metro network