Evolution Of The Current Telecommunications Networks And The Next

Evolution of the Current Telecommunications Networks and the Next Generation Optical Networks Hayrettin EVøRGEN2 Hüseyin EKøZ3 Ali Yavuz ÇAKIR 4 Cebrail TAùKIN 1 1 Türk Telekom Genel Müdürlü÷ü, 06103 Aydınlıkevler-Ankara / TURKIYE 2,3 Sakarya Üniversitesi, 4 Osmangazi Üniversitesi [email protected]

[email protected]

Abstract During the years between 1995 and 2000, there had been a strong competition between IP and ATM. MPLS that was approved by IETF in 1997, seems to end this competition with combination of best features of IP and ATM. MPLmS that is the equivalent of MPLS in optical networks, was designed in 1999. Afterwards, a new version of MPLmS, that is to say GMPLS, was developed in 2000 by IETF. Routing and signalling protocols that are supported by MPLS were extended to support GMPLS. The broadband boom, specifically growing data demand in high speed, and increasing bandwidth requirements has caused the evolution of the current transport networks. As a consequence, the IP over WDM networks, that integrate the traditional WDM networks having bandwidths of terabits and IP networks on which all applications run, has been developed. This article contains information on the MPLS, MPLmS and GMPLS network architectures which started to work on network backbones as a result of the evolution of telecommunications networks and new generation IP over WDM networks. Keywords: Current Telecommunications Networks, MPLS, MPLmS, GMPLS and IP over WDM

1. Introduction During the past 5 years, Asynchronous Transfer Mode (ATM) became the dominant WAN technology for the packet switched wide area networks. ATM which has fixed length cells and combines transmission, switching and multiplexing functions had appeared to be a promising solution not only for core networks, but also for the access networks. The characteristics of being asynchronous as opposed to the synchronous transfer mode (STM) of TDM was considered ideal for burst traffic volumes as the Internet traffic. ATM was not accepted by the LAN environment where Ethernet is a cheap, widely available alternative and thousands of applications can

[email protected]

[email protected]

run on. ATM cards which could be installed only on servers were costly compared to the Ethernet. For this reason neither was ATM used on the PC nor did the applications of it reached the end user. It only became the part of research projects. IP and ATM working together led the widely used IP applications to be available on the ATM dominated WAN networks. Communication networks are divided into two categories according to their geographical structures: Local Area Networks (LAN) and Wide Area Networks (WAN). IP became dominant in LAN’s where the end users reside and all applications are supported. SDH became the popular technology in carrying backbone traffic (e.g. voice) and ATM became the popular technology in carrying data traffic in the WAN side. In many circumstances ATM was running on SDH if the capacity requirement exceeds the minimum traffic rate of STM-1 (155 Mbps). As a result of carrying ATM over SDH, SDH took the possession of ATM traffic. SDH transmission networks are used by the most of the Telecommunication Operators to carry data and voice traffics. TDM based SDH technology has many advantages as high transport rates, automatic protection mechanisms, the add/drop capabilities and easy path provisioning. The integration of multi layered networks in a short time reformatted the IP over ATM over SDH connections. In other words, most of the WAN traffic including IP is transformed into ATM cells and carried over SDH paths. In order to carry data with reasonable costs, IP traffic had to be multiplexed with the ATM traffic or other TDM based traffics. The rapid growing demand for the Internet bandwidth brought many market opportunities. Especially in the WAN environments, Gigabits per seconds were reached and as a result ATM appeared to be an excess layer between IP and SDH [1]. As result of exponentially growing data traffic, the data traffic exceeded the voice traffic. But to carry this exponentially growing traffic, SDH (Synchronous Data

0-7695-2649-7/06/$20.00 (c) IEEE

Hierarchy) is used which is based on TDM (Time Division Multiplexing) technology. SDH transport structure which supports the requirements of voice and leased lines becomes an ineffective and costly solution for the burst traffics because of the non-statistical multiplexing structure. With the exponentially growing traffic, this architecture has the scalability and manageability problems. The necessity to do enhancements of one layer with the other layers in a multilayer environment, results the operators to invest continually. The intention of the industry is through the multilayered structure to the cost effective and simplified structure (IP/WDM) keeping the traffic management and quality of service layers [2]. DWDM wavelength multiplexing technology is the newest technology which minimizes the costs. DWDM generates multiple “virtual fibers” in the current fiber cable infrastructures and makes the capacity in the level of terabits per second. As packet switching in these data rates is not possible, OXC’s (Optical Crossconnects) are used to switch the wavelengths. The capabilities of the routers and the OXC’s make it possible to have high data rates by doing the multiplexing of the data rates in the optical layer, so SDH and ATM layers which traditionally did these become needless [3]. As a result, integration of IP with the simplified and cost effective transport layer makes it possible to carry very high data rates. The important point to take into account here is; in order to remove the SDH and ATM layers, the functionality of these layers should be handled by the IP and DWDM layers.

2. Current Telecommunications Network Current Data networks are in the four layered structure. These layers are IP, ATM, SDH and WDM (figure 1). IP- Carrying services and applications, ATM (Asynchronous Transfer Mode)- Traffic engineering and Quality of service, SONET/SDH – Transport and protection, DWDM (Dense Wavelength Division Multiplexing)- Supplying the capacity [4].

Figure 1. Current telecom network’s structure

owing to protocol headers used. In this scenario, IP packets firstly are capsulated according to a ATM AAL5 frame by using LLC/SNAP capsules. Then AAL5 protocol data unit (PDU) is divided into ATM cells having a 48 byte data payload. After that, ATM cells are transferred to SDH frame in order to be conveyed at optical layer.

3. Elimination of Intermediate Layers in Backbone By the explosion in the IP traffic within the last years; to overcome this growing IP traffic, it is needed to build a new network structure. IP backbone connections reached to 10 Gbps of (OC-192/STM-64) data rate that is the highest rate in SONET/SDH. Beside this, efficient less transmission of ATM with approximately %25 overhead cause to interrogation of existing structure. Current multi-layer protocol stack has some disadvantage. These are; • In IP over ATM over SONET over WDM architecture, only 76% of bandwidth can be used because of the protocol overhead • Layers often do not work in concert. Every layer now runs at its own speed. As low speed devices cannot fill the wavelength the whole bandwidth can not be used. • In case of a failure each layer wants to restore the fault itself. The detection in optical layer is rapid. • Transmission speeds of optic devices exceed the electronic devices. It is not approved to use multiple layer network structure any more since the dominant traffic is getting IP. There is no need to have an ATM layer between IP packet layer and WDM circuit [5]. Aim is to minimize the transmission overhead to maximize transport bandwidth. To maximize profit it is needed to minimize layered network structure, network design, engineering and operational costs (OPEX) [6]. Figure 2 shows various layers that are used IP forwarding.

Figure 2. Some layers that using transmitting to IP

Internet networks were set up on the basis of carrying Internet Protocol (IP) over SDH and ATM based networks. While IP packets are carried over ATM and SDH infrastructure, their load are about %25 weightier

0-7695-2649-7/06/$20.00 (c) IEEE

Figure 3. A Connection of IP and WDM IP/WDM is the most effective and active solution among the other possibilities. It also requires a simplified frame format for error processing. Two alternative frame formats are available that can be supported on IP. The first method is connecting IP diretly to the WDM using the PPP/HDLC encapsulation. The second method is IP-WDM integration using the GE frames. IP over WDM is named to send IP traffic over the WDM optic networks in order to ensure the universal IP connectivity and the bandwidth of the WDM networks. The integration of IP and WDM; means assigning WDM optic network resources in order to carry the IP traffic fast and efficiently in the IP layer.

MPLS

Control

Rapid growth in data traffic and the predominance of Internet Protocol (IP) in data communication have led researchers investigate the IP over WDM integration. In such architecture, network nodes employ WDMs and IP routers. Today’s IP over ATM over SONET/SDH over optical approach reduces efficiency as well as the effective bandwidth provided by WDM technology. The trend is to integrate the IP and WDM layers by eliminating one or two layers of protocol stack [7]. In this architecture, both the SDH and ATM layers are removed and using the IP and WDM layers, a two layered network is built. Transport efficiency is optimized as ATM layer is removed. As SDH equipment is not used in any way, the first investment rate is reduced. A simple IP over WDM network designed for point to point connection can be seen in Figure 3.

IP Router IP Routing Protocols

IP Routing Protocols

ATM Control Layer

Forwarding

3.1. IP Over WDM: Integration of IP and WDM

Longestmatch Lookup

Label Swapping

Label Swapping

Figure 4. Basic idea behind MPLS MPLS is the expansion of the existent IP architecture. MPLS allows being supported new properties and applications by new capabilities to added IP architecture. New applications contain traffic engineering, IP VPN (virtual private network), integration with IP routing and layer 2 or optical switching. MPLS is designed for increasing the speed of internet, scalability and required service capabilities. In traditional IP messaging, packets have to be forwarded according to their destination addresses field, at each interface. MPLS implements that process by a new technique, called “label-switching method”. Packets are sent with an additional, fixed length label. Routing process is separated from forwarding process and different routing services may be applied without any changes along forwarding path [8, 9].

4.1. MPLS Components

Architecture

and

Basic

Basic components used in MPLS architecture are MPLS label; Label Exchange, Label Switching Path (LSP), Label Switching Router (LSR) and Forwarding Equivalence Classes (FEC) are explained in this subdivision. A Sample MPLS topology is shown in figure 5 [6].

4. Multi Protocol Label Switching (MPLS) The origin of the MPLS based on cell switching routing of Toshiba, IP switching of Epsilon, Aggregate Route-based IP Switching of IBM and label switching of Cisco. MPLS that refers label switching technology plans on best features of ATM (rapid, easy transmission) to combine with best features of IP (availability, scalability, flexibility) is admitted in 1997 by IETF. The formation of MPLS is shown in figure 4.

ATM Switch

Figure 5. Traditional MPLS topology

0-7695-2649-7/06/$20.00 (c) IEEE

MPLS Label: Label which is located on the header of forwarded packet and carries information about packet belongs to which Forwarding Equivalence Classes is a short and fixed-length value. Label has similarities with link identifier like a VPI/VCI (Virtual Path Identifier/Virtual Circuit Identifier) used in ATM or used in Frame Relay technology a DLCI (Data Link Connection Identifier). MPLS label format is seen in figure 6. MPLS header is composed of 32 bits. 20 bits of 32 bits are used for label, 3 bits for Class of Service (CoS), 8 bits for TTL value and 1 bit for stack which determines the encapsulated packet is in a last stack or not.

5. Multi Protocol Lambda Switching (MPLambdaS) MPLmS is the equivalent term of MPLS in optical area. In 1999, MPLambdaS was designed to adopt MPLS traffic engineering as the control plane for optical cross-connects (OXCs) [10]. MPLmS brings the advances in MPLS Traffic Engineering control plane technology and applies it to the optical layer through the use of optical Cross connects (OXCs) [3]. MPLambdaS is an IP centric control plan protocol (extended from MPLS) designed for wavelengthswitching in WDM network. Control Plane has fixed topology and it is strictly separated from data channels. IP routing protocols (with extension) are used to distribute the WDM-link state information [11] There are some similarities and differences between MPLS and MPLmS. These are [12]; •

Figure 6. General MPLS label format Forwarding Equivalence Classes (FEC): During the transmission level, Forwarding Equivalence Class is used for packet groups which are processed in the same manner. All packets grouped in this style are routed to the destination point in the same way. This shows envisioned traffic having same destiny. Label Switch Router (LSR): LSR is a high speed switching and routing device which is used for Label Switched Paths (LSP) setup in MPLS networks. LSPs mentioned above use datalink layer transfer. Routers in MPLS Protocol Architecture are classified as according to their location e.g. Label Edge Routers and Label Switching Routers. Label Switched Path (LSP): Paths which are set up for providing FEC based packets transmission before the beginning of data transfer are named as Label Switched Paths. Label Distribution Protocol (LDP): Label Distribution Protocol is a set of LSPs, procedures and messages. These LSPs are set up at datalink layer. Procedures are used for carrying new information other LSRs with the use of network layer routing information. LDP sets synchronous sessions between label switches and required label changes with using transmission functions. LDP is the most important part of the control component [9].

•

• • • •

OSPF, CSPF and IS-IS for routing and LDP, CR-LDP, RSVP, and E-RSVP are applied by both MPLS and MPLmS in setting up or disconnecting label or lambda switched path (LmSP). MPLmS uses wavelength (lambda) as labels while MPLS uses labels instead of headers. Other features as LSRs are mapped as OXCs, and LSPs in MPLS are LmSP. Data transfer is simplified in both technologies by integrating the layers. Optical layers used by MPLmS are a circuit switching network but MPLS is a packet switching network. In lambda switching Networks as MPLmS, the blocking problem occurs less. As MPLmS directly controls the physical layer, it is more effective than MPLS in providing line restoration in times of failure.

5.1. MPLmS Architectures Element

and Network

An MPLmS network consists of WRs and is surrounded by LSRs. A typical MPLmS network is shown in figure 7 [6]. These LSRs at the edge are also called Edge-LSRs and have two functions. First, traffic flows of the service layer network are aggregated to high bandwidth traffic streams, suitable for efficient use of the limited number of available lightpaths. Second, Edge LSRs request unidirectional lightpaths (also called LSPs) to be set up by WR through the OTN. WR with MPLmS control planes are referred to

0-7695-2649-7/06/$20.00 (c) IEEE

as OXC-LSRs analogous to ATM and Frame Relay (FR) switches, called ATM-LSRs and FR-LSRs.

switching, but also devices that perform switching wavelength domains. GMPLS extends the MPLmS control plane to include domains such as SONET/SDH, ATM and Gigabit Ethernet for control with the optical layer [10, 15]. While MPLS seemed to combine parts of the IP layer to the Network Access Layer, MPLmS seems to bring down the upper layer functions to the physical layer. The clarification to MPLmS, named GMPLS, seeks to combine the IP and optical layers together, by allowing one control plane to control both protocols. Figure 8 describes the aggregation of layers as proposed by MPLS, MPLmS and GMPLS [12].

Figure 7. An MPLmS working backbone MPLS architecture, the term label denotes a fixed length value carried in the cell/packet header. LSRs process the packet header to distinguish between the different LSPs carried over link. In the MPLmS architecture, the term label denotes a certain wavelength on a fiber span or, if the WRs are TDM capable, a certain TDM channel of a wavelength. Thus, WRs process incoming traffic at channel/port level and distinguish between different LSPs, according to their incoming port/channel identifier [6].

6. Generalized Multi Switching (GMPLS)

Protocol

Label

Generalized Multiprotocol Label Switching (GMPLS) is also known as a Multiprotocol Lambda Switching (MPLmS) [4]. The Common Control and Management Plane (CCMP) working group of Internet Engineering Task Force ( IETF) is currently working on extending MPLS protocols to support multiple network layers and new TDM and optical services. This concept, which was originally referred to as Multiprotocol Lambda Switching (MPLmS) is now referred to as Generalized MPLS (GMPLS) architecture and concepts [13]. MPLmS was originally performing wavelength switching with IP and WDM layers’ integration. Afterwards, IETF expanded to MPLmS that limited with lambda, to multiple layers through the generalized MPLS. Thanks to this, a structure was constituted to support more than one switching type like packet switching, Time Division Multiplexer (TDM) switching, lambda switching and fiber switching [2, 7,14]. Generalized Multiprotocol Label Switching (GMPLS) is a multi-platform control plane technology to support not only devices that perform packet-

Figure 8. MPLS, MPLmS and GMPLS with TCP/IP layers shown GMPLS can be summarized as [7]; ƒ It is composed traffic engineering and optical extensions to the routing and signalling protocols of MPLS. ƒ Integrating the next generation data and optical networks, it is an optical networking standard. ƒ For interoperable and scalable networks, it acts as a bridge between IP and optical layers. ƒ It brings the intelligence to the optical Networks and this enables the operators do better restoration and utilize the bandwidth better. ƒ It is designed to allow edge networking devices such as routers and switches to request bandwidth they need from the optical layer. ƒ It offers the support for optical and hierarchical LSPs (or lightpaths) ƒ It can handle handle multiple traffic types simultaneously as it is designed to do. It creates a control plane to support multiple switching layers.

6.1. GMPLS Interfaces GMPLS made some modifications on MPLS as separation of signaling and data channel and support more types of control interface. GMPLS extends the MPLS to support more interfaces other than packet switch. The GMPLS architecture extends MPLS to

0-7695-2649-7/06/$20.00 (c) IEEE

include five different types of interfaces used on label switch routers: (Figure 9) [5,13].

Figure 9. Supported interfaces by GMPLS in control plane Packet-switch capable (PSC) interfaces: Can read the packet headers and forward the data based on these. Routers’ and L3 Ethernet switches have these interfaces. Layer 2-switch capable (L2SC) interfaces: Can read the L2 cell or frame headers and forward data according to these. ATM, Frame Relay and L3 Ethernet switches have these interfaces. Time-division multiplexing (TDM) interfaces: Can read the time slots and forward data according to these repeatedly. Digital cross-connects (DACSs), SONET add/drop multiplexers (ADMs) and SONET crossconnects have these interfaces. These interfaces are called “TDM capable”. Lambda-switch capable (LSC) interface: These have the capability to forward data based on the lambda (wavelength) the data was received on. Optical crossconnects (OXCs) that switch traffic on the wavelength level use these interfaces. Fiber-switch capable (FSC) interfaces: Forwards data based on the position of the data in real-world physical spaces. OXCs that switch traffic on the fiber or multiple-fiber level have these interfaces [13].

6. CONCLUSION As a result of ‘the rapid increase in Internet user numbers’, ‘increasing the number of user that are sending and receiving video’, ‘picture or large files’, ‘due to a lot of companies to use internet for intra and inter corporation information exchange’, ‘become very popular of mobile internet access’ and ‘the expansion of IP telephone usage’, data traffic excess voice traffic. Internet traffic is growing about 10% per month. To support this increasing demand with existing infrastructure seems to be very difficult. This growing bandwidth issue can be solved by integrating IP and WDM systems which support terabit level bandwidth

over a pair of fibre. Telecom operators will establish Optical Cross Connect (OXC) WDM equipments that are switching in optical layer in their backbone. The interoperability of IP and WDM layers which will remain after the simplification of existing multiple layers enable GMPLS signaling and routing protocols which provide TDM wavelength switching as well as packet switching in control plane.

7. REFERENCES [1] SERRAT, J., GALIS A., “ Deploying and Managing IP over WDM Networks”, Artech House, Boston, 2003 [2] ARKUT, R. C., ARKUT, I. C., “Etiket Anahtarlama Teknolojisinde Evrimsel Geliúmeler”, i-net.tr 2001 [3] GHANI, N., et al., “ On Ip Over WDM Integration”, IEEE Comms. Mag., Marc 2000

[4] BANERJEE, A., AWDUCHE, D., KOMPELLA, K., “Generalized Multiprotocol Label Switching: An overwiew of Signaling Enhancements and Recovery Techniques”, IEEE Com. Mag., July 2001, pp 144-150 [5] LIU, K., “IP over WDM”, John Wiley & Sons, New Jersey, june 2002 [6] TAMSU, P., SCHMUTZER, C., “Next Genaration Optical Networks”, Prentice Hall, New Jersey 2002 [7] ILYAS, M., MOUFTAH, H.T., “Optical Communication Networks”, CRC Pres, USA, 2003 [8] VISWANATHAN, A., FELDMAN, N., WANG, Z., and CALLON R., “Evolution of Multi-Protocol Label Switching”, IEEE Communications Magazine, V. 36 Issue: 5, pp.164-173, May 1998. [9] ROSEN, E., VISWANATHAN, A., and CALLON, R., “Multiprotocol Label Switching Architecture”, RFC 3031, January 2001. [10] YOO, S.B.J, “ Optical Label Switching, MPLS, MPLmS and GMPLS”, Optical Network Magazine, May/June 2003, pp 17-31 [11] AWDUCHE, D., ‘Multi-Protocol Lambda Switching: Combining MPLS Traffic Engineering Control With Optical Cross-Connects,’ Internet Draft, Work in progress, draftawduche-mpls-te-optical-03.txt, April 2001. [12] CHO, G.Y., CHUNG J.M., “Analysis of MPLS vs. MPLambdaS Next Generation Networking Tecnologies” IEEE 2001, pp 518-522 [13] HALABI, S., “Metro Ethernet”, Cisco Press, Indianapolis, September 2003 [14] CHRISTIANSEN, H., WESSING H., “ Modelling GMPLS and Optical MPLS”, IEEE 2003, pp 288-294 [15] KHAN, H.K, SOO, H.M., REYES, J.S., CHO, G.Y., CHUNG, J.M.,“Analysis of GMPLS Architectures, Topologies and and Algorithms”, IEEE 2002, pp 284-287

0-7695-2649-7/06/$20.00 (c) IEEE