Unit 4 Multiplexing, Framing, and Some Solutions

Introduction to Communication Networks Spring 2007

Unit 4 Multiplexing, Framing, and some solutions...

©EECS 122 SPRING 2007 Acknowledgements – slides comming from: • Data and Computer Communication by Wiliam Stallings (our supplementary textbook).

• Data Communications and Networking by B. Forouzan, Mc Graw Hill, 2004 ( a very nice-to-read book!)

• Some figures have been used form the earlier issues of the EECS 122 tought by Prof Jean Walrand.

• Introduction to Telephones & Telephone Systems by A. Michael Noll, Artech House, 1986

• Megabit Data Communication, John T. Powers, Henry H. Stair II, Prentice Hall

• Digital Telephony by J. Bellamy: “”, J. Wiley & Sons, 2rd edition, 2000

Prof. Adam Wolisz 2of 63 MULTIPLEXING

Prof. Adam Wolisz 3of 63 Multiplexing

• General Problem: Several - n- different channels (voice, TV- channels) should be supported between a pair of locations. We would like to avoid usage of n physical links (cables).

• Looking at the features of media you will easily see that the supported bandwidth exceeds by far the bandwidth needed for each channel...

Prof. Adam Wolisz 4of 63 Variants of multiplexing • The dimensions of multiplexing – time (t) – frequency (f) – code (c)

– space (si) – sometimes…

• Care for separation: guard spaces, code orthognonality

• Multiplexing can be – Synchronous (constant allocation) – Statistical (variable allocation)

Prof. Adam Wolisz 5of 63 Frequency Multiplex • Separation of the whole spectrum into smaller frequency bands

• A channel gets a certain band of the spectrum (in the synchronous case – for the whole time!)

• Note: guard zones in frequency are needed!!

Special case: Wave division Mux

Prof. Adam Wolisz 6of 63 Schema for FDM [Forouzan] mod

Prof. Adam Wolisz 7of 63 FDM of Three Voiceband Signals

Prof. Adam Wolisz 8of 63 Example - Community Antenna TV (CATV)

Currently used systems require about 6MHz /TV Channel

Prof. Adam Wolisz 9of 63 Time Multiplex

• The whole bandwidth is used all the time, but – alternatively – by different channels!

Prof. Adam Wolisz 10 of 63 Time Multiplex: Interleaving of data segments [Forouzan]

Prof. Adam Wolisz 11 of 63 Time and Frequency Multiplex [Schiller] • Combination of both methods

• A channel gets a certain frequency band for a certain amount of time

• Example GSM cellular telephony: FDM with TDD (8 bi- directional channels per frequency band) is used...

k1 k2 k3 k4 k5 k6

c f

2.18.1 Prof. Adam Wolisz 12 of 63 Time Division Duplex (TDD) and FDD

Similarly a Frequency Division Duplex - FDD with two frequency Channels: for up-link and down-link respectively, can be defined

Prof. Adam Wolisz 13 of 63 Bursty Data

• Burstiness of data

– In many data communication applications, data occur in bursts separated by idle periods

– This type of data can often be transmitted more economically by statistical (or asynchronous) multiplexing...

Prof. Adam Wolisz 14 of 63 Synchronous vs. Statistical TDM

Note: Data slots must be addressed!

Prof. Adam Wolisz 15 of 63 Statistical Multiplexing Gain [mod.from N.Mc Keown, Stanford] Comment: Synchronous Multiplexing would use 2C bits/s – statistical uses R<2C. But: how to define R? – The queue helps „smooth“ the load but there might be losses !!! A+B Rate 2C R < 2C A C R B

time Statistical multiplexing gain = 2C/R

Other definitions of SMG: The ratio of rates that give rise to a particular queue occupancy, or particular loss probability.

Prof. Adam Wolisz 16 of 63 Example of Statistical Multiplexer Performance

Prof. Adam Wolisz 17 of 63 Probability of Overflow and Buffer Size ρ- is the ratio of the offered load to the nominal service rate of the system – see Queuing (later)

Prof. Adam Wolisz 18 of 63 FHSS (Frequency Hopping Spread Spectrum) I

• Discrete changes of carrier frequency – sequence of frequency changes determined via pseudo random number sequence • Two versions – Fast Hopping: several frequencies per user bit – Slow Hopping: several user bits per frequency • Advantages – ROBUSTNESS: impact of frequency selective fading and interference limited to short period !

2.32.1 Prof. Adam Wolisz 19 of 63 FHSS : Schema of the operation

Fast hopping

Slow hopping

Prof. Adam Wolisz 20 of 63 FHSS – System overview

narrowband spread signal transmit user data signal modulator modulator

frequency hopping synthesizer sequence transmitter

narrowband received signal signal data demodulator demodulator

hopping frequency sequence synthesizer receiver

2.34.1 Prof. Adam Wolisz 21 of 63 CDMA: Code Division Multiple Access A Channel: a unique code in the same spectrum at the same time

Prof. Adam Wolisz 22 of 63 Space division multiplexing...wireless... • Assume a sectorized antenna

• Transmisison/receive in one of the sectors does not limit the usage of other sectors...

Prof. Adam Wolisz 23 of 63 FRAMING

Prof. Adam Wolisz 24 of 63 Framing • WHY:

– The physical layer supports bit-synchronization.

– Data units bigger than a single bit must be recognized...

• HOW:

– Time gaps (not good - might be squeezed within the physical layer),

– Physical signaling - the physical layer has to support some control symbols, besides of a 0 and a 1, say a J and K. Example: the Manchester extension of the IEEE 802.5 - token ring.

– Field Lenght marker at the beginnig of the field: Whole notion of unit lost if this lenght marker would get corrupted!

– Specific symbols: Character oriented, bit oriented variants.

– Clock based

Prof. Adam Wolisz 25 of 63 Framing - delimiting symbols • Delimiting characters in character based transmission, i.e. the case when the transmitted information is composed of symbols - c.f.the ASCII code table.

• SYN SYN - used for the synchronization

• SOH...... STX...... ETX the framing sequence for the header and for the text. – Transmission of binary information: • DLE STX ...... binary information...... DLE ETX – What about a DLE inside the binary information?? • input binary information: ...... DLE...... • transmitted binary information: .... DLE DLE...... • extracting the information: ...... DLE ...... – A) This scheme is closely tied to an 8 bit character representation. – B) A single error can cause a misinterpretation.

Prof. Adam Wolisz 26 of 63 Framing - delimiting symbols (2) Example

DLE STX A DLE BDLEETX (a)

DLE STX A DLE DLE B DLE ETX (b)

Stuffed DLE

(a) Data sent by the network layer. (b) Data after being character stuffed by the data link layer. (c) Data passed to the network layer on the receiving side.

Prof. Adam Wolisz 27 of 63 Bit oriented transmission-delimiting flag

• Delimiting flags in bit oriented transmission, i.e. the case when the transmitted information is represented as a string of bits (the concept of octets is sometimes used to support the bit manipulation).

– The usual flag pattern: 01111110

• Transmission transparency is assured via bit stuffing:

– The transmitter always stuffs a 0 after 11111

– The receiver removes a 0 following 11111

• User data 01111110 are transmitted as 011111010

Combinations of solutions discussed above are frequently used to increase the power of framing - e.g. delimiting flags together with byte (symbol) count. See additional reading after bit error unit! Prof. Adam Wolisz 28 of 63 Bit-oriented Transmission

The packet framed by two special bit patterns called flags. Bit stuffing is used to prevent a flag from occurring in the middle of a packet. The bit-stuffing procedure is illustrated in (b): a bit 0 is inserted after each pattern 011111 that appears in the packet. The reverse procedure (bit destuffing) is performed at the receiver to restore the original packet.

Prof. Adam Wolisz 29 of 63 Clock based

Idea: to have a sequence of markers in predefined positions – here the sequence 101 in proper distance all the elements are assumed to have equal length! Problem: This sequence, with the proper spacing, MIGHT appear just by chance in the content!!! Solution: Look several times – the random appearance will not be repeatable over numerous frame series! Prof. Adam Wolisz 30 of 63 Examples of transmission systems

Prof. Adam Wolisz 31 of 63 Phone Backbone - FDM Carrier Standard – OLD! [Forouzan]

Prof. Adam Wolisz 32 of 63 FDM Carrier Standards - OLD – some numbers

Prof. Adam Wolisz 33 of 63 American Digital Hierarchy

Each channel carries data (voice) digitized at a rate of 8000 samples per second with 8 bit per sample. A frame contains 24 channels plus one framing bit per frame. Thus, the required transmission rate for DS-1 is 8000 x (24 x 8 + 1) bits per second = 1.544 Mbit/s.

Prof. Adam Wolisz 34 of 63 American Digital Hierarchy –synchronization [bellamy]

• T1/D4 Superframe and D4 Channel Slots. – A super frame combines 12 frames of 193 bits each. – The framing bits of these frames produce the T1/D4 superframe pattern of 100011011100

Prof. Adam Wolisz 35 of 63 Extended Superframe Format Framing (1) • Extension of the super frame from 12 repetitions to 24 repetitions. • Framing bit positions take new functions and meanings (24 bits). • Framing (6 bits) • Error Checking (6 bits) • Maintenance Communications (12 bits)

193rd Bit 193rd Bit 125 µs

123456789101112131415161718192021222324

Bipolar format

10101101Binary code 5,2 µs

Basic North American PCM framing and signaling format.

Prof. Adam Wolisz 36 of 63 Some specific ways of using it... [Stallings]

Prof. Adam Wolisz 37 of 63 American TDM Carrier Standard [Forouzan]

Prof. Adam Wolisz 38 of 63 TDM Carrier Standards

North American and International TDM Carrier Standards

Prof. Adam Wolisz 39 of 63 Just for info – the International Frame...

Prof. Adam Wolisz 40 of 63 Trunks

Prof. Adam Wolisz 41 of 63 Mhmm... Not quite unified...

Prof. Adam Wolisz 42 of 63 Instability of the timing.. • Cyclic changes of the data rate of sending

• Systematic difference in timing between the sender and receiver.

• What can we do?

– cyclic- or fluctuating differences removed by elastic buffer...

– Systematic difference has to be dealt with

• Systematic differences. How?

– Plesiochronous operation (T-hierarchy)

– Synchronous operation (SONET)

Prof. Adam Wolisz 43 of 63 Synchronization - Definition of terms: – Single discrete signal may be either • Isochronous : Constant frequency of signal changes (e.g. 8 kHz of PCM-coded voice) • Anisochronous – Two discrete signals may be either • Synchronous or

• Asynchronous – Mesochronous (meso = middle, greek) · Same center frequency, limited phase difference · (e.g. signals from a single oscillator being processed in different stages) – Plesiochronous (plesio = near, greek) · Nominal same center frequency (e.g. two independent oscillators) – Heterochronous (hetero = different, greek)

Prof. Adam Wolisz 44 of 63 Changes in Length of Transmission Media • Path length changes as a result of thermal expansion or contraction of guided media, or as a result of atmospheric bending of a radio path. While a path is increasing in length, the effective bit rate at the receiver is reduced because more and more bits are being ‚stored‘ in the medium.

• Example - 500 km long T2 transmission link using 22-gauge copper wires which have a velocity of propagation of 189671 km/s: – Thermal expansion coefficient 16.5 ppm/°C – The temperature of the wire increases by 20 °C in one hour:

• Change in path length: d −6 =⋅⋅⋅=Δ 165.020105.16500 km 6 ⋅⋅ 165.010312.6 • Change in number of bits: B =Δ = 49.5 bits 189671 49.5 • Change of data rate: R ⋅==Δ 10525.1 −3 bps 3600 • Relative change in ΔR ⋅10525.1 −3 = ⋅= 1041.2 −10 received data rate: R ⋅10312.6 6

Prof. Adam Wolisz 45 of 63 Synchronization – Elastic Buffer

Can deal with fluctuating differences, introduces a delay...

Prof. Adam Wolisz 46 of 63 Elastic buffer - implementation

Elastic store operation with a one-frame memory.

Prof. Adam Wolisz 47 of 63 Pulse Stuffing Concept (1)

• Two channel multiplexer showing equal data rates for each input

No pulse stuffing needed since both streams have exactly the same rate

Prof. Adam Wolisz 48 of 63 Pulse Stuffing Concept (2) • Simplified pulse stuffing example – Assume that stream 1 is slightly faster than 2 – but within legally (standards!) defined upper bound.

• Additional information needed to allow adjustments of the information flow within each sub-channel.

– Ci bit specifies if the following Si bit carries tributary data (Ci = 0) or just stuffing. • The output data rate should be equal to DOUBLE the maximum LEGALLY Permitted data rate of inputs.

Prof. Adam Wolisz 49 of 63 American Digital Hierarchy • Every multiplexing stage adds overhead.

Digital Level Signal Type Rate in Mbits/s Number of Channels of Number of kbits in Channels the type overhead 0DS-00.0641DS-0 - 1DS-11.54424DS-0 8 2DS-26.3124DS-1 136 3DS-344.7367DS-2 552 28 DS-1

Prof. Adam Wolisz 50 of 63 Pulse Stuffing Concept (3) • Example of a higher level multiplexing format for 6.312Mbps DS2 signal in North American digital hierarchy.

– A DS2 signal is derived by bit interleaving of four DS1 signals and adding the appropriate overhead bits.

– The C1 –C4 bits are repeated 3 times each, in order to allow for majority voting.

Prof. Adam Wolisz 51 of 63 A real multiplexer…

Prof. Adam Wolisz 52 of 63 Problems of PDH • The Plesiochronous Digital Hierarchy had a number of problems: – Each part of the world has its own transmission hierarchy (expensive interconnection equipment) – Justification (bit stuffing) spreads data over the frame • add-drop-multiplexers are hard to build • extract a single voice call -> demultiplex all steps down • switching of bundles of calls (n * 64 kbit/s) is difficult • (every switch has to demultiplex down to DS0 level) – The management and monitoring functions were not sufficient in PDH – PDH did not define a standard format on the transmission link • Every vendor used its own line coding, optical interfaces etc. • Very hard to interoperate

Prof. Adam Wolisz 53 of 63 Prof. Adam Wolisz 54 of 63 Prof. Adam Wolisz 55 of 63 SONET / SDH (1) • Synchronous Optical NETwork (SONET) and the Synchronous Digital Hierarchy (SDH) – Started by Bellcore in 1985 as standardization effort for the US telephone carriers (after AT&T was broken up in 1984), later joined by CCITT, which formed SDH in 1987 – Three major goals: • Avoid the problems of PDH • Achieve higher bit rates (Gbit/s) • Better means for Operation, Administration, and Maintenance (OA&M) • SDH is THE standard in telecommunication networks now • Originally designed to transport voice - used for everything

Prof. Adam Wolisz 56 of 63 SONET / SDH (2) • SONET / SDH - Basic concepts – SONET / SDH system consists of switches, multiplexers and repeaters (and the fiber in between) – PATH is the connection between source and destination – LINE runs between two multiplexers (possibly through repeaters) – SECTION is the connection of any two devices (point-to point)

Source Multiplexer Repeater Multiplexer Repeater Multiplexer

Section Section Section Section Line Line Path

Prof. Adam Wolisz 57 of 63 SONET / SDH (3) Level US Europe, Data rate Data rate Data rate Japan (gross) (SPE) (user) 1 OC-1 - 51.84 50.112 49.536 2 OC-3 STM-1 155.52 150.336 148.608 3 OC-9 STM-3 466.56 451.008 445.824 4 OC-12 STM-4 622.08 601.344 594.824 5 OC-18 STM-6 933.12 902.016 891.648

6 OC-24 STM-8 1244.16 1202.688 1188.864

8 OC-36 STM-12 1866.24 1804.032 1783.296 9 OC-48 STM-16 2488.32 2405.376 2377.728 10 OC-192 STM-64 9953.28 9621.504 9510.912

• No overhead bits needed for justification – higher speed link is formed by byte-interleaving data from lower speed links – exact multiples of lower speed data rates so e.g. OC-12 contains 12 byte interleaved OC-1 frames

Prof. Adam Wolisz 58 of 63 SONET [PD] • Clock-based – each frame is 125us long – e.g., SONET: Synchronous Optical Network – STS-n (STS-1 = 51.84 Mbps) STS-1 = OC-1

Overhead Payload r STS-1 STS-1 STS-1 Hdr Hdr Hd

9 rows

Hdr STS-3c 90 columns

Prof. Adam Wolisz 59 of 63 SDH - Clocking • All network elements are totally synchronous

• Still, there are delays in the network

• Hierarchy of clocks, lower levels synchronize to higher levels

Stratum Min. Accuracy Min. Stability Pull-In Range 1±1 in 10-11 Master Reference Master Reference 2 ±1.6 in 10-8 1 in 10-10 Should synchronize with a clock ⇒ ±0.025* accurate to ±1.6 in 10-8 2 ±4.6 in 10-6 ±3.5 in 10-9 Should synchronize with a clock ⇒ ±7.0* (some conditions) accurate to ±4.6 in 10-6 3±32 in 10-6 N/A Should synchronize with a clock ⇒ ±50* accurate to ± 32 in 10-6

* = Minimum accuracy relative to 1,544,000 bits/s.

Prof. Adam Wolisz 60 of 63 What about tributary speed differences [PD] - Frames appear synchronously, and have always the header of fixed lenght (9 rows in STS-1) and position (at the begining of the frame!) - The Payload does NOT have to begin directly after the header – it is fixed by a pointer (part of th eheader). - The Payload has always a constant length – thus might „overflow“ into the next frame - If there are excessive bytes, those are stored in the header – and the moves to the left. If bytes are missing – the „empty“ bytes are marked and the pointer move to the right.

87 columns Frame 0

9 rows

Frame 1

Prof. Adam Wolisz 61 of 63 High Reliability – 60 ms for reconfiguration [LUCENT]

Prof. Adam Wolisz 62 of 63 What SONET/SDH does better (conclusions) • SONET (Synchronous Optical NETwork) and SDH (Synchronous Digital Hierarchy) are almost identical • Interconnection is easy (exists, works) • Justification, if still needed, is performed by pointers • Data from each input is placed in a payload container (Administrative Unit- AU) – it spans multiple SONET/SDH frames – a pointer in the header of the SONET/SDH frame signals the start of the payload container in the frame (in 3-byte increment for SDH) – positive and negative justification through this pointer – slip buffer delay reduces from 193 bit for a T1 signal down to 24 bit – Single 64 kbit/s lines (1 byte in the SONET/SDH frame) can be found and extracted in the frame – HIGH RELIABILITY!!!!

Prof. Adam Wolisz 63 of 63