Backbone network technologies T-110.300 Jouni Karvo, Timo Kiravuo
Backbone network technologies This lecture tells about backbone networks After this lecture, you should know WDM, PDH, SDH and ATM understand how virtual overlay networks can be created in the physical network understand how lower layer techniques make both telephone and data networks possible to coexist in the same network
Multiplexing There are several multiplexing techniques that allow simultaneous transmissions in the same physical medium: TDM (Time Division Multiplexing) FDM (Frequency Division Multiplexing) WDM (Wavelength Division Multiplexing) of these, FDM is not in widespread use in modern backbone networks (but is in radio networks) TDM can be used in any layer, not only on the physical layer
WDM WDM is essentially FDM, but deserves a new name, since it is implemented by using lasers of different colors: Due to the physical properties of light, signals of different wavelengths do not interfere significantly Up to 160 wavelengths (DWDM, Dense Wavelength Division Multiplexing) and growing with UDWDM (Ultra- Dense WDM) Each wavelength can carry e.g. 10Gb/s with SDH leading to 1.6Tbps speeds per optical fiber in DWDM Compare this to telephone calls; if all 5 million Finns would call simultaneously to a foreign destination, it would result only in 0.32Tbps!
WDM switching Switching: electro-optical or purely optical optical equipment; for example Micro-Electro Mechanical System (MEMS) switches, where wavelengths are first separated by prisms and then switched by adjustable mirrors Add-Drop Multiplexers (ADM) are equipment that are used to add or drop specific wavelengths from the line without affecting the other wavelengths. Optical Cross Connects (OXC) are optical switches, able to route incoming wavelengths to specific outgoing ports. Care must be taken to avoid joining similar wavelengths Each wavelength can be modulated and transmitted separately There is no need to use a single protocol
Virtual Topologies Virtual topologies can be created over the physical one. Enables more flexible service offerings Care should be taken to provide sufficient redundancy
A peek in history In order to understand SDH and ATM properly, knowledge on PDH is essential. PDH was the means to make the telephone network digital in the 60s. These techniques are used in the core networks of the operators, to provide high capacity links. These techniques are very similar; the essential difference is in synchronization and framing.
PCM frames (E1/T1) Speech is transmitted as PCM (Pulse Code Modulation) frames. The speech is converted to 8bit (7bit in USA) digital samples 8000 times in second. Can transmit speech up to 4kHz (Nyquist theorem) A-law or?-law coding After that, the network treats it simply as a binary data stream. One speech channel has a data rate of 64 kbps (or 56 kbps in USA). Any other data can be treated similarly E.g. IP packets may be sent consecutively as a data stream of 64 kbps. Since the trunk lines use TDM for transmitting multiple channels simultaneously, the data streams are cut in octets, and framed
Time Slots The European E1 frame contains 32 channels A channel is an endless stream of data octets 2 channels transmit control information and 30 data The frames contain 32 octets, called timeslots (TS) TS0 TS1... TS14 TS15 TS16... TS31 TS0 contains synchronizing code, notifying the receiver of the beginning of the frame. TS1--TS15 contain speech (or data) channels TS16 contains signaling TS17--TS31 contain speech (or data) channels.
PDH Frames One frame thus contains one octet of each of the carried channels. A frame is transmitted 8000 times each second, yielding a 8 * 32 * 8000 = 2 048 000 bps or 2.048 Mbps. The corresponding T1 frame used in USA contains 24 speech channels with 8-bit coding and one framing bit, ``stealing'' payload bits for signaling. The frame length is 193bits, sent 8000 times a second, yielding a 1.544 Mbps data rate. A multi-frame is a series of consecutive frames (16 in Europe. 12 in USA). Multi-frames are used, since for modern equipment, it is not necessary to tell the receiver every time where the frame starts. Instead, space is given for CRC codes and extra signaling.
PDH Synchronization All digital switches have a master clock. All outgoing streams are slaves of the master clock. Incoming bit streams are slaves of their senders' clocks. For proper transmission, clocks need to be synchronized. Consider a sender A sending its stream a bit faster than the relaying node B can forward it to the receiver C. At some point, the buffers at B overflow, causing a lost PCM frame. This is called a slip. The difference of PDH and SDH is their synchronization and frame structure.
Plesiochronous Digital Hierarchy (PDH) In plesiochronous operation the nodes are synchronized on the E1 (T1)-level. Synchronization sources send the E1 multiplex using their high precision clocks and the receivers synchronize themselves according to the input signal Both mutual synchronization and master-slave synchronization modes possible. High precision clock means here a clock with a stability range of about [10^-11... 10^-13], implying use of atomic clocks, possibly UTC or GPS coordinated.
PDH multiplexes A PDH multiplex is means a bunch of primary PCM multiplexes (E1 or T1). In a plesiochronous system, each PCM multiplex is transmitted with a different clock. The E1 level clocks are synchronized, but higher levels synchronize according to the E1 level. Since the actual transmission rates differ a little from the nominal bit rate, some action is needed to compensate the differences in different streams. The solution used is justification (called stuffing in USA). Justification means adding extra bits to the stream, which allows for reading of an input buffer faster than the sender fills it.
PDH hierarchy a) E1 b) E2 c) E3 d) E4 The first PDH multiplex level is E1, containing (as noted before, 30 channels and 2 control channels), 2.048 Mbps. The second PDH multiplex level is E2, containing four E1 multiplexes (120channels). The bit rate is 8.448 Mbps, containing frames of 1056bits, (4 * 256 bits for the multiplexes and 4 * 8 bits for justification and frame alignment). The third PDH multiplex level, E3 contains four E2 multiplexes (called tributaries), yielding 480 channels. The bit rate is 34.368 Mbps. The fourth PDH multiplex level, E4 contains four E3 multiplexes (1920 channels), with bit rate of 139.264 Mbps.
PDH Hierarchy cont. Note that for each multiplex level, the frame rate is 8000 frames/second, i.e. the same as for E1! Sometimes synchronization errors are too big to be corrected and a slip occurs. In these cases a frame is either lost or re-sent. In speech communications this might cause a small rustle for all channels transmitted on the link. For data communications, one octet is lost or duplicated for all channels on the link. This might result a lost packet or even connection on the application level. PDH is being replaced by SDH (and in Finland already mostly gone).
Synchronous Digital Hierarchy SDH is a timing hierarchy Allows higher and more flexible bit-rates than PDH Uses a different frame structure and synchronization STS (Am. called Sonet) or STM (Euro) frame structure SDH is used to synchronize the frames Hierarchical synchronization Nodes on a higher level on hierarchy are used to synchronies the lower levels. One master clock to synchronize whole operator network More accurate synchronization reduces slippage
SDH STM Frame structure The basic SDH frame, STM-1 contains 2430 octets, and is transmitted once in each 125 microseconds, i.e. 8000 times a second. This yields a 155.52Mbit/s gross bit rate. a) 270 octets b) 9 rows c) 9 octets d) 261 octets e) 9 rows f) AU g) SOH AU = Administrative Unit (payload) SOH = Section Overhead
Frame Structure cont. Higher order SDH frames, STM-N contain N interleaved STM-1 payloads, and a N times the transport overhead. Again, 8000 frames are sent each second. Typical STM-N multiplexes include STM-4 at 622.08Mbit/s, STM-16 at 2488.32Mbit/s, STM-64 at 9953.32Mbit/s. Synchronization errors are handled by AU-Pointers. The AU may float within the frames: a) Frame N b) Frame N+1 c) AU-Ptr d) AU (payload)
Frame Structure cont. In addition to AU-Pointers, three octets of STM-1 SOH can be used for justification. The AU in is then further divided to smaller containers capable of accommodating different bandwidth tributaries. (this makes SDH able to carry many types of lower level multiplexes simultaneously) The data rates of the lowest level containers in an AU are: 139 264 kbit/s, 44 736 kbit/s, 34 368 kbit/s, 6 312 kbit/s, 2 048 kbit/s, and 1 544 kbit/s. I.e. both E1 and T1 PCM multiplexes are compatible with SDH.
SDH Synchronization There are four possible synchronization modes in a SDH network: Synchronous --- all elements are synchronized (possibly indirectly) to the PRC. Normal operation. Pseudo-synchronous --- all elements are synchronized to a PRC, but there are several PRC:s. Normal operation on links between operators. Plesiochronous --- Part of the nodes have lost their synchronization reference, and work on hold-over. Asynchronous --- Nodes are on free run. Due to possible float and plesiochronous operation, also SDH needs justification, but is is implemented differently than in PDH.
SDH synchronization levels US terminology: stratum, pl. strata Primary Reference Clock (PRC, Stratum 1) requirements in ITU-T G.811 relative error <10^-11. For example cesium clocks. free running, but may be co-ordinated by UTC One per operator Transit Synchronization Supply Unit (Transit SSU, ~ Stratum 2) requirements in ITU-T G.812-T relative error in holdover: <10^-9 e.g. rubidium clocks stable enough for short periods of holdover operation (operation without synchronization reference from PRC)
SDH synchronization levels cont. Local Synchronization Supply Unit (~ Stratum 3) requirements in ITU-T G.812-L relative error in holdover: <2 * 10^-8 e.g. temperature compensated crystals reduced holdover capabilities local switches SDH Equipment Clock (SEC) (~ Stratum 4) relative error in holdover: <10^-8 relative error in free run: 4.6 * 10^-6 local switches, digital channel banks and private exchanges
Clock adjustment Clocks on the higher levels on hierarchy give correction terms to adjust the clock speed.? clocks' frequencies float around the frequency given by the PRC.
Topology Rather free topology can be used Typically, SDH networks are organized as dual rings Backbones might be organized as meshes Rings are configured to be self-healing (in milliseconds) Digital Cross Connects (DXC) and Add-Drop Multiplexers used Section-Line-Path: Sections are between devices (such as signal regenerators) Line between multiplexers Path between terminals
Topology cont. SDH (or SONET in USA) is the Layer 2- transmission technology used mostly for telephone traffic, but also for the Internet backbones nowadays. Usage of SDH containers is ideal for creating a logical topology over the physical one. The approach for flexibility and high data rates in SDH works well, but is a bit complex at the frame level. Thus, a third approach, ATM.
Asynchronous Transfer Mode Instead of E1-frames (or T1-frames), ATM uses 53 octet frames, that are called cells. The payload of an ATM cell is 48 octets, i.e. that of 2 T1 multiplexes. Thus, a stream of ATM cells sent 8000 times a second could carry 48 simultaneous speech connections. In practice this is not done, but AAL transport is used instead. In PDH, each channel has its own timeslot which does not change. The same holds for higher order multiplexes. SDH uses pointers inside frames to show the location of the containers. ATM cells have the channel number (Virtual Path/Virtual Circuit identifier -- (VPI/VCI) L2 address) carried in each cell instead.
ATM cont. VPI/VCI can be used to create a logical topology over the physical one When creating higher order ATM multiplexes, there is no need for reframing data, or sending ATM cells a specific order, due to the Layer 2 address. The ATM network needs synchronization just as PDH and SDH networks. It can be synchronized using the same clock hierarchy as the SDH network. There is no justification capability in ATM, but due to the lack of frame structure, it is possible to drop single cells instead of whole higher order frames. The absence of frame structure also allows for variable bit rate virtual circuits (channels).
ATM cont. To take advantage of the increased flexibility, a new set of protocols was developed. The combination of ATM and these protocols is called B-ISDN. ATM will be used in 3rd generation mobile networks, so its importance might rise. Application ATM Adaptation Layer (AAL) ATM cell switching Physical layer
ATM cell switching services: Constant bit rate (CBR) --- stream service Variable bit rate (VBR) --- stream described by peak cell rate, sustainable bit rate, and maximum burst size. Available bit rate (ABR) --- the applications can renegotiate the bit rate with the network devices during the connection Unspecified bit rate (UBR) --- no QoS guarantees (cells are dropped by the network element if necessary)
ATM Adaptation Layer Segmentation and Reassembly (SAR) Different types of AAL: AAL1 supports CBR services (such as E1/T1) AAL2 for low bit rate packet traffic with end-to-end timing requirements AAL3/4 for bursts of data with no timing requirements. Two modes: message mode and stream mode AAL5 is like AAL3/4 but without sub-addressing SAAL (Signaling AAL) for reliable transfer of signaling messages
Summary of trunk links Telephone network is designed for streaming data Data networks have been designed using the experience in telephone network engineering Synchronization is essential PDH, SDH and ATM are a "family'' ATM offers a bit more flexibility than SDH. WDM adds capacity and logical topology SDH is currently the major transmission system, WDM increasingly popular