©1997-2005 R. Levine Page 1 Digital Switching EETS8320 SMU/NTU Lecture 9a Switching Software (print slides only, no notes pages)

©1997-2005 R. Levine Page 1

Digital Switching

EETS8320

SMU/NTU

Lecture 9a

Switching Software

(print slides only, no notes pages)


Overview• Switching software is real-time event-driven:

– The driving events are end-user actions such as dialing digits, lifting or replacing handset, etc.

• Circuit-switched voice telephone software mimics the human interface behavior of historical electro-mechanical switches

– Including incidental items like intentional post-dialing delay and non-symmetrical treatment of origin/destination vis-à-vis disconnect (wireline switches)

• Telephone switching software is often described or designed using finite state machine (FSM) formalism– Three isomorphic (equivalent) descriptions:

• Graphical linked points diagram• Graphical flow-chart-like (SDL= specification and

description language)• Tabular row-column lists


Historical Switching• Original 1876 A.G.Bell installations were point-to-point hard

wired. Examples:– Office to warehouse of same firm (like a modern

intercom circuit)– Palace to beach-house of the King of Hawaii

• Manual cord-board switching introduced in Hartford, CT in 1880s.– Teen-age boys pulled electric wires across the room and

temporarily connected them in response to verbal instructions from subscribers

– Later developments led to standard cord-board: a desk-like panel with a retractable cord from each voice connection unit, and a panel in front of the human operator with a socket for each subscriber (and historically later, a socket for each trunk line to another switching center)

– Parallel historical development of common battery power and supervision technology also facilitated the cord switchboard


Other 19th Century Improvements

• Carbon Microphone (Edison and Berliner)– Permitted loops of up to ~5 mi (8 km) due to greater

transmitted electrical audio power level• 2-wire “loop,” instead of single wire using earth

conductivity for current return path– Earth return was previous standard in telegraph systems, but

produced tremendous “cross-talk” for telephones– Loop greatly improved voice quality and reduced audio noise– Invented by J.J.Carty, later chief engineer of AT&T

• Alternating current ringer (low maintenance) instead of previous buzzer devices with vibrating electric contacts subject to sparking, corrosion and deterioration

• Common (central office) battery for dc loop current using transformer to couple audio voice signal between two telephones in a conversation


Switchboard Plug• Same dimensions used today for 1/4 in

(6.35 mm) diameter stereo headset plug

Tip (green positive wire)Ring (red

negative wire)

Sleeve (only in electro-mechanicalswitches,no standardoutside-plant color)

Insulators

Tip

Tip

Ring

Sleeve

SleeveRing

Plug Assembly Graphic Symbol

Socket Assembly Graphic Symbol

Note: use of red insulation for neg-ative polarity is unique to the telephone industry. Other electricalstandards (power,electronics, auto-motive) use redfor positive.


Historical Cord Circuit

_

+

Primitive Telephone set (dial, ringer, cradle switchnot shown). No directionalcoupler here as in latertechnology.

Primitive central office cord circuit. Positive battery terminal grounded to minimize electrolytic corrosion. Audio frequency voice signals coupled via transformer. Does not show ringing power, sleeve wires, signal lamps and buzzer, operator exclusion switches, etc.

telephone set andsubscriber loop

Common battery feedand voice coupling

Note useof sameCO battery(with audiobypass capacitor)for all loops.

Earphone

Microphone

Other telephoneset not shown.Operator headsetalso in parallel withvoice wires tempo-rarily, not shown.


Supervision Methods• In traditional telephone jargon, “supervision” describes only the

aspects of signaling which relate to busy/idle status– Dialed digit information was historically distinct (called “signaling”)– In modern cellular/PCS software both things are often described by the

word “supervision”• therefore, be careful about jargon!

• Historical method to get attention of the operator was a small hand-cranked AC generator or “magneto” at subscriber end

– Produced about 90 V ac, at 20 Hz frequency. – Still standard ringing waveform for North America today

• Then the common-battery circuit was introduced– Subscriber “switch-hook” closed a current loop and operated a light

and/or buzzer near that subscriber’s socket on the switchboard panel– Operator lifted a retractable cord from the desk-top, connecting her*

headset to the subscriber via a voice-frequency transformer– Operator then asked “Number Please?”

* Boys were replaced by more polite ladies in 1890’s; operator corps was exclusively female until 1960s .


Call Connection• Operator plugged other end of cord circuit into callèd subscriber

socket (the second syllable of callèd is artificially stressed in telephone jargon to emphasize the spoken distinction with “call”)

– Outer part of socket and “sleeve” (called “C” wire in European jargon) of plug carried a voltage when that line was busy. (No C wire in modern electronic switches.)

– Voltage (if present) on sleeve produced an audible click in operator earphone, indicating busy line

• If callèd line is idle, cord circuit is plugged in, connecting voice circuit of both telephones

– … and connecting temporarily the operator as well– Operator presses momentary contact switch to apply 20 Hz, 90 V ac ringing

to the callèd loop– When callèd person answers, operator presses a latching switch to

disconnect operator’s headphone from the cord circuit– When either participant hangs up, dc loop current from common central

office battery stops, indirectly operating a distinct buzzer and light on the cord board via a relay.

– Operator then “tears down” the connection by pulling both retractable cord plugs from the callèd and calling part circuit sockets. Cords fall back into desk surface due to weights under the desk.


Cord Switchboard Capacity• The number of simultaneous conversations is limited to the number of cord

circuits installed in a cord switchboard– Each cord circuit is similar to a storage address (byte) in an electronic

switch vis-à-vis capacity– The BHCA (call processing) capacity is limited by the attention and

operational speed available from the human operator• Both were improved by providing more operator positions (cord circuits)

– Each subscriber loop appeared at multiple sockets, each one within reach of an individual operator position

• Thus a historical need for busy status signal (sleeve or C wire)• Early example of switch concentration

• Operator-handled calls were controlled by human intelligence– Computer controlled (stored program controlled - SPC) switches

merely strive to put back into automatic service many of the clever things human operators did historically (example, ring back to originator when initially busy destination finally becomes available)

©1997-2005 R. Levine Page 10

Some Human Operator Features• Call by name (no telephone number required)

– Response to: “Please call the Smith home.”

• Wake up calls (at pre-determined time)• Re-connect calls accidentally disconnected*• Notify busy line of incoming call waiting• Set up 3-way (or more) conference call • Connect call to alternate line when subscriber is

away from home (call forwarding)Note that modern “feature-rich” PBX, small business key systems, and some

PSTN switches now do these things via computer control

• Several experts have calculated that there are not enough people on earth to support the today’s (2001) level of public telephone traffic using operator cord board switching!

*The GSM cellular system can optionally be configured to do this.

©1997-2005 R. Levine Page 11

Strowger Step-by-step Switch• Almon B. Strowger, a mortician (undertaker) in Kansas City

KS, invented the first practical automatic dialing system– Famous story: fearing that the human operator was directing calls for a

mortician to his competitor, he invented an automatic user-controlled switch

– First version (installed in LaPorte, IN, circa 1895) used extra wires and push buttons on each subscriber set

– Rotary dial with impulsive current on the voice wire pair was a later development

• Strowger’s manufacturing firm, Automatic Electric, moved to suburban Chicago, IL, later absorbed by GTE, later moved to Phoenix AZ, now AG Communication Systems (partly owned by Lucent)

– “Stepper” progressive control switches were manufactured world wide for many decades

– Electromechanical common-control switches developed by other manufacturers, such as “panel” and “crossbar” types partially succeeded steppers in the 1930 - 1960 decades

©1997-2005 R. Levine Page 12

Schematic Stepper Diagram

Many details omitted here

Vertical Motion

Rotary Motion

Rank 1

Rank 0

1

23

4 5 6 7 89

0

Tip, Ring, Sleevewires from Rank 8,column 7.

Electromagnets andsprings activate the motionsof the wiper arm in responseto dial impulses.

Ten places oneach circularrank wherea 3-contact assembly islocated -- notillustrated indetail.

Rank 9

Axle

Arm

©1997-2005 R. Levine Page 13

Stepper Switching• Strowger switches evolved into an assembly with a movable wiper

switch “inlet” and 100 “outlets” (wire pairs with “sleeve” wire)– 10 contact pairs arranged in a horizontal arc, selected by rotating the

wiper switch arm. (Also a third “sleeve” wire in addition)– 10 such horizontal arc sub-assemblies stacked and selected via vertical

motion of the axle (actually the first motion is vertical)– Single-motion (rotation only) switch assemblies were also used

• “Line Finder” switch (mostly single motion) acts as input concentrator (“inverse” of selector action)

– Wiper arm contacts act as the single outlet– Line finder single-motion stepper typically wired to 10 subscriber lines,

selects a line when that line goes off-hook• Stepper starts stepping from line to line when any of the 10 lines go off

hook, then stops when correct “off-hook” line is “found”– analogous to operator responding to buzzer and light– Multiple line finders wired in parallel to the same 10 telephone sets

analogous to multiple operator stations with each having access to the same subscriber sockets.

• Number of simultaneous originating conversations limited to the number of line finder switches connected to those lines. Ten line finders wired to ten subscribers is “non-blocking” with regard to line finders. (Overall system may still block at later stages…)

©1997-2005 R. Levine Page 14

Selector Switches• Line finder outlet goes through a transformer “cord circuit”

• Connected to dial-tone generator until the first dialed digit.• Then the circuit is switched through a chain of two-motion selector stepper

switches, with a “motion” for each digit. Each burst of impulses (dialed digit) produces a rotary or vertical motion constituting the next stage of the wiper arm selection process

• Dial pulses from rotary dial (typically 10 impulses per second, each one approximately 60 millisec current OFF and 40 ms current ON) are passed around the cord circuit by special electro-mechanical relays

• A relay employs magnetically operated switch contacts, so that current ON/OFF status in the contacts mimics the current ON/OFF status in the wire coil causing the magnetic action.

– Special “slow release” relays hold the line finder so the 60 ms OFF intervals do not cause a disconnection

• After turning and releasing the telephone set rotary dial from an angle labeled with a specific number, the returning rotation of the dial to its normal position produces 1 to 10 current impulses

– Simultaneously, an “Off-normal” switch contact in the telephone set temporarily short-circuits earphone so clicking is not heard

• Following a stage of selection motion, a slow release relay is automatically connected into that line to prevent further disturbance of that particular selection due to the succeeding bursts of dialing impulses

©1997-2005 R. Levine Page 15

Incidental Information

• Rotary dial label “0” represents 10 impulses everywhere in the world (except Sweden, where the dial is labeled 0, 1, 2…9)

– However, touch-tone dials in Sweden use the same digit labels for DTMF tones as the world standard.

– Impulsive signaling must be converted at international boundaries to Swedish telephone system. But symbolic signaling (binary digit codes used in CCS7, etc.) is the same everywhere.

• Alphabetic dial labels (2=“ABC”, 3=“DEF”, etc.) were introduced in New York City in ~1923 when subscribers complained about “long” 5 digit directory numbers.

– Alphabetic dial labels were introduced in US, Canada, UK, France, Scandinavia and USSR (three cities only) but not all the same:

• Examples: Q on French dial, Russian (Cyrillic ) letters in Moscow, Leningrad, Odessa,

– Considered an obstacle to direct international dialing, alphabetic exchange names were purged from telephone directories in 1960s by international agreement.

• The “anti-digit dialing league” and other grass roots groups in the US opposed all-digit directories in the 1960s.

– Letter labels still appear on the dial in most of these named countries. Business users highly value so-called “Anagram” numbers such as 1-800-FLOWERS, or 1-800-NORSTAR, 1-800-AMERICAn, etc.

©1997-2005 R. Levine Page 16

Significant Properties of Stepper Switches• To add more traffic capacity, install more line finders and more

selector switches– This increases parallel path (traffic) capacity through the switch, since

multiple last stage selectors lead to the same destination lines. • Only one last stage selector can connect at a given time. The sleeve wire is

also connected to each corresponding position on the selectors and is used to divert the call to a busy signal generator if the sleeve voltage is ON for that destination line and a call is attempted while destination line is busy.

• A non-blocking Strowger step switch assembly would require 100 last stage selector switches connected to 100 destination telephone lines, and similar replication of parallel paths all the way to the originating lines (line finders, earlier stepper stages, etc.).

• This automatically increases the call processing capacity (BHCA) of the switch as well

– Each selector is both a traffic path and a part of the digit processing hardware– When there is a traffic path available to the destination, there is also the

hardware to respond to the succeeding dialed digits.

• A stepper switch assembly “automatically” has enough call processing capability if it has adequate traffic path capacity

©1997-2005 R. Levine Page 17

Stepper Properties• Stepper switches are extremely reliable overall

– Because of parallel path capability through a large stepper switch, the failure rate of these switches (when properly maintained) is very good

• Failures affecting only one user amount to only about 1 hour cumulative in 20 years• Failure of the entire switch is only 1 or 2 minutes in 20 years, and when this occurs

it is mostly due to power supply or other aspects of the system

• Steppers can be adapted to many improvements• Touch-tone dialing (by means of a tone-to-pulse converter)• Computer control has been adapted to steppers to make advanced features available (such as

call waiting, 3-way conference, etc.)• Unfortunately, basic reliability, power consumption and size not improved!

• Inter-switch signaling between stepper switches requires electrical transmission of dialing impulses

– conversion between modern digital signaling (common channel 7) and impulse switching is feasible, but slow acting

• European version of SS7 signaling allows transmission of one dialed digit at a time, but North American (ANSI) version does not send dialed number onward until the “last” digit is dialed.

– several earlier “electronic” but non-digital switching systems still used electromechanical switching (small relays) and analog transmission (example: No. 1 ESS), but digital computer central control or stored program control

©1997-2005 R. Levine Page 18

Undesirable Stepper Properties• High maintenance

– ‘“Gross Motion” or “Large Motion” wiping contacts• Require lubrication, cleaning, adjustment, etc.

– Susceptible to corrosion from sparking, air pollution (such as SO2 in the air, etc.)

• Slow mechanical operation– Even when tone-to-pulse converters support Touch-tone

dialing

• Slow signaling – Can’t take full advantage of CCS7 and other electronic

signaling systems

• Big and bulky– Digital switches use ~1/50th the floor space of steppers; ~1/10th the

floor space of crossbar switches.

©1997-2005 R. Levine Page 19

Some Other Historical Electro-Mechanical Switches

• Panel– A huge mechanical “monster” switch using continuously running electric motors

and electrically operated clutches to move wipers vertically and horizontally on a rectangular wall panel of contacts. A high maintenance problem.

• Crossbar– An assembly of rocking contacts attached to vertical and horizontal rotating

actuator axles. Because of relatively small motion and compact size, this was the heir apparent to the stepper switch in both North America and Europe until electronic switching appeared.

• X-Y– A horizontally platform with rows and columns of contacts with wipers actuated

by magnetic coils. Gross motion problems, but more compact than Strowger design.

• Rotary– Similar to X-Y switch, but platforms had contacts arranged in semi-circles of

increasing radius. More compact than Stepper, but same gross motion problems.

• Multi-relay– Rocking contact motion, but still rather complex and difficult to maintain.

The last 3 were mainly used by “independent” telcos in North America. All here except Crossbar and Multi-relay were “gross motion” switches.

©1997-2005 R. Levine Page 20

Common Control• Many of these electro-mechanical designs had separate relay

assemblies to count (“decode”) the dial impulses, completely separate from the switching portion of the system. These so-called “common control” portions were analogous to the computer control in a digital switch.

• Once the desired destination directory number was decoded, it was “translated” by special purpose wired logic devices

– One method for this was to use magnetic core memory of a special wired type (not addressable RAM like modern computer memory)

– The equipment numbers resulting from the translation were used to select a path through the switching part of the system.

• The result of the “translation” was a code designating the proper bay, shelf, and switch outlet wire for the internal destination calls, or the proper outgoing trunk group for outgoing (other switch) calls. The first non-busy channel in a trunk group was selected by an appropriate special outgoing trunk switch.

• These systems first demonstrated the need for provisioning separately both sufficient call processing capacity (BHCA) and also sufficient switching capacity (Erlangs)

©1997-2005 R. Levine Page 21

Electronic Switches• ESS No. 1: Electronic but not Digital!

– Computer control/stored program control (SPC)– Analog Relay switching, using sealed contact reed switches

• Most of the design problems for high reliability were addressed in this design.– Duplicated processors, etc.

• ESS No. 4: Fully Digital but Trunks Only– When designed (1960s-70s) the cost of A/D conversion

(CODECs) on each subscriber line was seen as prohibitive– Depended on T-1 channel banks at distant ends of trunk

groups for A/D (analog/digital) conversion– 4 ESS is a transit or tandem switch, not a central office or end office

• 4 ESS has only T-1 links at its ports (no telephone sets except for a few test telephones)

• Above are all Lucent (then AT&T) products. Competitors had similar designs shortly after or almost contemporaneously. In that era Western Electric (manufacturing division of AT&T) only sold products to the Bell System operating companies, and was required to license all its patents as one result of an earlier anti-trust settlement.

©1997-2005 R. Levine Page 22

Subscriber Interface to Electronics Was Difficult

• Integrated circuits made the digital central office (end office switch) economically feasible

– The most elegant hardware design requires a dedicated analog-digital converter at each subscriber loop. A long time economic problem for end office switches.

• Subscriber loop circuitry usually on a removable printed wiring card

– Replaceable in case of failure or damage from lightning, etc.

• Handles some so-called BORSCHT* functions– Battery Feed – Over voltage (from lightning and accidental power line contact)– Ringing– Supervision– Codec (A/D inter-conversion, also low pass audio filtering)– Hybrid (directional coupler, 2-wire to 4-wire inter-conversion)– Testing (a capability of the switch, not of the telephone set)

* This term credited to John Iwerson of AT&T Bell Labs ca. 1960s

©1997-2005 R. Levine Page 23

Modern Digital Switch Subscriber Loop Block Diagram

Hybrid andmatchingnetwork

Hybrid andmatchingnetwork

_

+

Telephone set (dial,ringer, cradle switchcircuits for loop length level compensation not shown)

Wireloop,up to~8 km

Central office switch equipment. Actual switching is not shown, but isoff to the right of this page. Audio frequency voice signals coupled via transformer. Ringing power, loop current detection (supervision) not shown.

Amplifierand A/Dconverter

Amplifierand D/Aconverter

Transmitsignal

Receivesignal

CO part

telephone set andsubscriber loop

Common battery feedand voice coupling

Tip

Ring

©1997-2005 R. Levine Page 24

Subscriber Line Interface Card/Chip (SLIC)

• Due to large volume of use, integrated circuits are available which perform most of these BORSCHT functions

• ICs designed for line card in switch and chips for use in a low-cost telephone set are both available

• Spoken acronym SLIC /slIk/ sounds like another acronym, Subscriber Line Concentrator (SLC). Ask for fully spelled out version if context is not clear!

©1997-2005 R. Levine Page 25

Some BORSCHT Explanations• Battery Feed via split winding on audio coupling

transformer• Over-voltage (lighting, power line crossing)

protection primarily based on arc-over at spark gaps installed where outdoor wiring enters subscriber premises and CO building.

– Enclosed gas spark gaps provide uniform electrical “breakdown” at ~300 volts between wires or wire-to-ground

• Hermetic enclosure prevents variations due to air pressure and humidity, a problem in older lightning arrestor devices

– Non-linear series resistance devices limit high current surges due to lightning or accidental “cross” with power voltage wiring

• Light bulbs or “heat coils”• Positive Temperature Coefficient (PTC) resistors using conductive

polymer plastics

– Not on subscriber loop circuit card, except for PTC resistor.

©1997-2005 R. Levine Page 26

More BORSCHT Explanations• Ringing voltage from ringing generator via

electro-mechanical relay contacts on tip & ring• Supervision (dc loop current sensing) via various

methods:– Sensing relay coil in series with subscriber loop. Loop current

actuates separate relay contacts.• Inductance of relay coil affects frequency response somewhat, but

can be bypassed for audio frequencies via a capacitor– Non-linear magnetic material (“saturable magnetic core”) with loop

current coil and sensing coil• Loop current changes magnetization point behavior• Sensing coil’s small signal inductance decreases when loop

current is on.• Smaller, less costly, less effect on voice signal.

©1997-2005 R. Levine Page 27

Still More BORSCHT

• CODEC (COder/DECoder) in switch (except for ISDN sets): – Low pass filter the audio on analog side of A/D conversion

• Active RC filter, switched capacitor filter, or CCD (charge control diode array) transversal filter are different analog technologies

• Purpose is to attenuate audio power above about 3.5 kHz– Then “measure” voltage, 8000 samples/second for coding– Encode each voltage sample as a compressed digitally coded 8 bit sign-

magnitude binary code• Mu-law approximately logarithmic compression rule used in North

America and Japan• A-law log-linear compression rule used in other national PSTN

systems– Digital/Analog conversion in opposite direction as well

• Hybrid or directional coupler is analog device using multiple windings on transformers, together with a “matching network” composed of resistors and capacitors

– Separates incoming and outgoing electrical waveforms on 2-wire subscriber loop into separate unidirectional signals with good but not perfect accuracy

– All digital transmission operations in a digital switch comprise two opposite-flowing unidirectional signal paths

©1997-2005 R. Levine Page 28

Automatic Test in Digital Switches• Electro-mechanical relay installed in the line card can switch

subscriber loop temporarily to an auxiliary test device– Main test operation is to measure (tiny) microamperes of

“leakage” current between wires and from each wire to ground• High leakage current indicates imminent failure due to moisture in cable,

damaged insulation, etc.

– Testing is usually done circa 2 AM when traffic is minimal• Even so, if subscriber lifts handset or a call comes in, test is

suspended until loop is again idle– Suspicious test results are automatically reported to repair crafts

persons• Tremendous reduction in staff is feasible when their repair

work can be scheduled, rather than waiting for an emergency, unexpected customer complaint, or loop failure!

• Most repairs are thus done before customer notices noisy line problems!

• Consequently, most unexpected failures today are due to human error or accident, rather than slow cable deterioration

©1997-2005 R. Levine Page 29

Digital Switch Advantages• Automatic test reduces staff costs significantly

– Predominant cost saving in many cases!!• Feature-rich, increases income of public

telephone company by selling optional “vertical” features (e.g. Call waiting, conference, etc.)

• Inter-works with digital trunks (T-1 etc.) without use of channel banks

• Smaller size allows more CO capacity growth in same building

• Less electric power consumption, reduces operating costs somewhat

©1997-2005 R. Levine Page 30

Internal SwitchingMatrix orNetwork

Block Diagram• Control Processor and Switching Matrix are duplicated for reliability

– Line and trunk interface cards are not duplicated because failure rate on outside plan wiring is greater than loop or trunk electronics.

Internal SwitchingMatrix orNetwork

Line/LoopInterface

TrunkInterface

ControlProcessor(CPU, RAMetc.)

Line/LoopInterface

Line/LoopInterface

TrunkInterface

TrunkInterface

Morenot shown

Morenot shown

Multiple station loopsMultiple T-1 trunk groups

optional remote line concen-trator

©1997-2005 R. Levine Page 31

Translation Tables• Each end office switch has at least 3 translation tables in its

control processor– 1. Internal line appearance number (ILAN) translated to directory

number (DN)• Identifies billing number for originated calls, and for calling line ID• ILAN is a proprietary number indicating the rack, shelf and circuit

card number of a line– 2. Inverse table of above: DN to ILAN

• Used to route incoming call to proper destination line– 3. Translates from NPA/NXX (or just NXX) into the proper outgoing

trunk group to reach that destination.

• Two inverse tables are used for fast look-up– Like using both a Spanish-English and a separate English-Spanish

dictionary for human language ‘translation’– Tiny switches (example: 16 lines) use just one table and perform

exhaustive search for the “inverse” translation function

• Additions, removals, and changes in DNs are made by entries in these tables, not by rewiring the external subscriber loops.

©1997-2005 R. Levine Page 32

Other Switch Configurations• A switch can be configured with only trunk interfaces (no line

interfaces). Applications include everything except traditional end central office use:

– Tandem or transit switch use in local or long-distance network– A cellular or Personal Communication System (PCS) radio system switch

• Connections to base radio stations are via trunks (e.g., T-1)

• Historically, a switch can be configured with only line interfaces (no trunks) for use as an “intercom” or PAX inside a building. Seldom installed today since a standard PBX with both inside and outside connections is less costly than “two track” systems.

• A line module can be located remote from the switch location when a distant cluster of subscribers needs service.

• Connects to main switch via T-1 links thru a trunk interface• Subscriber Line Concentrator (SLC-96) is an example of this.• PBX also performs this function, but has different signaling and is typically

subscriber owned. PBX and SLC systems are not covered here In detail.

©1997-2005 R. Levine Page 33

State Machine• Standards, user documentation, or software design documents can be

written in natural human language, but this often leads to misunderstandings and differing implementations– Readers disagree on meaning of natural language, regarding sequence

of steps, etc.– Actual operational test of compliance to standards requires testing via

inter-working against pre-existing implementations• Often, first-to-market implementations actually supersede the written

standard when discrepancies occur

• A better form of description is needed for:– Software algorithm design– Description and documentation of existing systems

• for testing or design of compatible equipment• for user training

• Finite state machine (FSM) formalism (also called Discrete State Machine -- DSM) serves this purpose

• SDL-Specification and Description Language, ITU-T standards Z.100 and SDL 2000, formalize a graphic flow-chart-like symbolism for this purpose.

©1997-2005 R. Levine Page 34

Finite State Machine (FSM)• FSM formalism can describe a computer or a telephone

switching system quite well• A FSM has a number of distinct states

– States are distinguished from each other by the unique binary value of:

• At least one bit somewhere in the CPU, RAM or mass storage (disk, etc.) is distinct (1 vs. 0 value) from the corresponding bit value in another state

• In an electro-mechanical telecom switch, at least one relay/switch contact is distinct (ON vs. OFF)

• A FSM is “driven” from state to state by events– An event is often an external cause such as a customer dialing

a digit, lifting or replacing a handset, etc.• The expiration of a timer/counter is also an event

©1997-2005 R. Levine Page 35

Some Simplifications•Strictly speaking, each combination of busy vs. idle telephone lines in a switch is a different state of that overall state machine

– Because of the similarity of operation of all telephone lines, we can simplify the description by describing the telephone switch in terms of just the 2 (or 3, etc. lines for a conference call) involved in the conversation

– The distinctions due to different optional vertical features (call waiting, etc.) can be handled by means of a general FSM description which handles every possible feature, with clearly defined options to allow or deny each specific feature dependent on a data table entry defining the class of service (COS) for that line*.

•We consider the states of one telephone and the aspects of the switch which relate to it, and also the events at the callèd telephone as well.•The general historical approach to FSM design is to describe what historical electro-mechanical switches do, and then program the digital switch to present the same behavior to the customer•When new features are designed, feature conflicts sometimes arise. These include discovery of ambiguous operations, etc.

– Feature conflicts are usually resolved by re-design of the feature at the human interface level.

*Some local service providers (e.g. SouthWestern Bell -- SBC) now allow all subscribers to use most previously “optional” services, charging on a per use basis until a maximum monthly fee is accumulated.

©1997-2005 R. Levine Page 36

FSM Description Formats• Logically equivalent (isomorphic) descriptions can be made

in several forms1. Graphic point or picture for each state, with directed lines or

arrows representing event-caused transitions between states

• Useful for human visualization, particularly with “cute-sy” pictures.

2. Table with column for each state, row for each event (or the reverse), and entries describing the target state and related information.• Often very large if all events are treated explicitly, and often

has many null (not possible) entries. Good for certain table-driven computer software systems.

• Usually not instructive for human visualization3. Flow-chart like description such as SDL

• Convenient starting point for software development• One-to-one correspondence to formal software language is

under study (e.g. ITT CHILL language occasionally used)

©1997-2005 R. Levine Page 37

Simplifying Conventions• Certain events invariably lead to the same result,

regardless of current state (whenever logically consistent)– Example: “hang up” of the handset leads to

disconnection (“this line idle”) state– To avoid pictorial clutter, this is omitted but implied in

graphic point-line diagram. Shown only where essential for understanding.

• Multiple states can be symbolically combined into one “covering” state to clarify the explanation aspects– All the internal details must still be explicitly defined for

a working description (perhaps separate diagram)– Example: dialed digit collection (“digilator”)

©1997-2005 R. Levine Page 38

Pictorial FSM ExampleIllustrates only 3 states

Idle state

Event: called byanother caller.

Ringing state

Conversation (answered) state

Event: lift handset

Event: replace handset (note: some designs have intermediate state with 10 sec. timeout)

Event: other callerabandons call attempt,ring-no-answer.

“Babble, babble”

©1997-2005 R. Levine Page 39

Originate-Answer Distinctions• Most PSTN “wireline” switches actually handle disconnect

differently for an originator vs. an answerer. Such switches disconnect:– immediately when originator disconnects– after (typically) 10 second timer expires, when answering person

disconnects.– Distinction is software controlled based on a bit set in RAM

• Mimics a historical property of some electro-mechanical switches– Allows callèd person to hang up and then quickly run from one

extension set to another on same line without disconnection

• Many PBX and Cellular/PCS switches do not distinguish originator vs. destination (except for billing!!)– We do not distinguish in our diagram for simplicity– If distinguished, two separate conversation states would be required

in the diagram

©1997-2005 R. Levine Page 40

Example ExtendedMore states but still not “complete”

Idle

callèd

Ringing

Conversation

replace handset

other callerabandons

Lift handset Dial tone

Dial firstdigit

Distant lineanswers

“Digilator”

~~

Validdigits

Event: timer expiresor invalid digits.

Collect digits

Announcement: “Please hang up andtry again…”

replace handset

~~ ~~

“Inert” Howler

Distant line rings or busy.

or

©1997-2005 R. Levine Page 41

Other Representations• Not shown explicitly in these notes are two other FSM

representations of a telephone switch:• Tabular representation could have (for example) a column

for each state, and a row for each event– Entry for Idle column and “Lift Handset” row is “Go to

Dial Tone state”– Entry for Dial Tone column and “Lift Handset” row is

“not applicable”– Entry for Dial Tone column and “Dial first digit” row is:

Start timeout for inter-digit max time, go to next Digilator internal state

• SDL description comprises a distinct flow-chart-like diagram for the computer processing steps which must follow each event– Proto-representation of the software in the interrupt

handling routine invoked by that event.

©1997-2005 R. Levine Page 42

Time-Space Diagrams• In some cases the sequence of events and messages between

different parts of a telecom system is displayed via a time-space diagram.

– Time (not to correct scale) usually increases in downward direction– Various horizontal positions (not to scale) represent different physical

devices (subscriber set, end office switch, transit switch, etc.) in the system

– Only one (usually representing a “successful” case) event sequence is displayed.

End OfficeSwitch

Distant End Office Switch

TransitSwitch

DestinationStation Set

OriginatingStation Set

Incr

easin

g Ti

me

Off Hook

Dial Tone

Dial Digits

Recognize non-local number andextend to destination switch

Ring

AnswerConnect Voice Channel

©1997-2005 R. Levine Page 43

Variations on a Theme• Use of diagonal lines (as in previous page example)

emphasizes transmission time delay aspect– Often called a “zig-zag” diagram

• Horizontal lines may also be used– Then often called a “ladder”diagram. Still exhibits sequence of

events but de-emphasizes transmission delay

– Another name: “Ping Pong”Diagram

• Similar diagrams in other subject areas illustrate sequential events at different locations– e.g., Feynman diagrams in quantum physics

• Time-Space diagrams cannot clearly illustrate all exceptional cases in one figure– e.g. destination busy, or ring-no-answer, timer in software

expires, etc. etc.

©1997-2005 R. Levine Page 44

Distant Line Ring or Busy

• When the destination telephone is in another switch and the trunk signaling is one of the more primitive types (not common channel No.7), the origination switch cannot distinguish distant ringing from busy– The human caller must listen to the call

progress tones to distinguish busy/ringing• When the callèd telephone line is in the same

switch or CCS7 signaling is used, this switch can distinguish, and two states should be drawn.– Two distinct states are not illustrated in these

notes.

©1997-2005 R. Levine Page 45

Digilator Digit Collection• A contraction of the words “digit” and “percolator,” the Digilator

“state” is really a collection of many internal states and events.• Digit collection strategy can be described as a tree-structured

data collection decision process• Any pause before or between digits which exceeds maximum time

causes a state transition to a recorded announcement– Internal timer (typically 6 to 20 seconds) is started after each digit but the last,

and reset again when the next digit is sensed– Expiration of the timer causes a “time out” interrupt

• Valid digit strings are described by both– Numbering plan: assignment of specific number groups to local, long-

distance, and service lines– Dialing plan: assignment of specific digits (usually prefixes) for

purposes not described in dialing plan:• Example: dial 9 prefix to get outside line in a PBX• Example: dial initial 1 (North America) or 0 (many European

nations) for non-local or special service calls (1-411, etc..)

©1997-2005 R. Levine Page 46

Digilator Information• In North American Numbering Plan (NANP)

– Initial digits 2,3,… 9 (not 1,0 ; traditionally represented by N) are valid for local directory number

• Subsequent digits can by any of 1,2,3,…9,0 (called X). First 3 digits form exchange code NXX, remaining digits XXXX. Special treatment for 3-digit service codes NXX=911, 311, and in some areas 611, 411 etc.

• Digit Collection will be stopped after 7 digits (local call) are collected– Recently, mandatory 10 digit local dialing in numerous cities like Dallas.

• Then further testing or connection required to determine if this number is in service, corresponds to a valid NXX, etc. Data for this decision may not be available in the originating switch.

– The complicated cases are initial zero (0) and initial one (1)– Initial 1 may imply:

• service call: 1-411 for directory assistance, 1-611 for repair (initial 1 not used universally)

• Non-local (inter-exchange carrier) but in e.g. Los Angeles, just non-local• first digit of 11XX, a rotary dial substitute for special feature prefix/code

such as *69 (1169) for “call back most recent caller”• first digit of 1010XXX prefix for selected inter-exchange carrier (e.g., 1010222

for MCI, 1010288 for ATT, etc.)

©1997-2005 R. Levine Page 47

NANP Initial Zero• Initial zero (0) has several subsidiary choices, depending

upon succeeding input:• Followed by no other digits (typically 6 second timeout)

connects to local operator/attendant (0|)*– Followed by second 0 and timeout (00|), connects to

preferred inter-exchange carrier’s operator/attendant– Followed by 11 indicates international call (011-) prefix

• 010- is international operator assisted prefix– Followed by any valid foreign directory number (country

code, area code, local number) indicates an operator assisted call (so-called “zero-plus” call)

• Example, person-to-person or English-language-only international call, etc.

* The non-standard symbol “|” is used here to represent a timeout with no succeeding digits.

©1997-2005 R. Levine Page 48

NANP Areas• US, Canada, and certain coastal* and Caribbean islands are under the NANP

– Each area has a 3 digit area code of form NXX, sometimes represented by the 3 letters NPA. In some documents, the 10 digits are represented ABC DEF GHIJ.

• Historically the B digit was restricted to 1 or 0 due to historical use of NNX for local office codes. No longer done (now NXX)

• Size of an area is dependent on total directory numbers in use in that area. High population density areas have required many area code splits or overlays in recent years and will again in the future

• “Caller pays” special surcharges is the source of questionable billing in some Caribbean nations. Beware of area code 809 and others...

• Certain pseudo-area codes are used to cover the entire NANP:– 800, 888 and 877 for callèd-line-pays long distance numbers.

• Actual target number is determined by a translation table data base using the succeeding 7 digits

• Call forwarding to existing line can be altered based on originator’s directory number, time/date or other factors

– Your call to Sears Roebuck or Domino’s Pizza is routed to the geographically nearest store (central office) to your point of origin

– When calling some large firms, your call to the same number may go the the east coast customer service department in the morning, and the west coast department in the afternoon and evening.

*e.g. St.Pierre and Miquelon, French possessions off Canada’s Atlantic coast. Also Bermuda, etc.

©1997-2005 R. Levine Page 49

ITU Numbering Plan• The world is divided into 9 zones, each with a

specific initial digit used in national prefix:– 1 North America (US, Canada and some islands)– 2 Africa– 3 Europe (part)– 4 Europe (other part)– 5 Central and South America– 6 Australia and pacific (part)– 7 Former USSR– 8 China, Japan and Pacific (other part)– 9 India and Middle East

• smaller nations have 3-digit code, larger nations have 2- or 1-digit codes.

©1997-2005 R. Levine Page 50

How is Connection Routed?• A connection is directed from an originating line or trunk to

a destination line or trunk outlet by setting the correct numbers in the connection memory associated with the switching matrix.– The choice of destination or outlet is based on an internal

translation table which uses a part of the directory number as input (e.g. NPA or NXX) and physical equipment identification number (trunk number or ILAN) as output.

– The contents of this translation table may be changed from time to time (or another table substituted) to accommodate:

• Malfunction or traffic saturation of certain links (use of alternate path or route)

• Known or expected better choices for economic or traffic reasons

– Time of day changes in traffic in various time zones– Utilize different price strategies on leased or outgoing

trunks

©1997-2005 R. Levine Page 51

Historical North American* Signaling Methods• DC (direct currrent) or baseband dial-pulse signaling 1900-1940s• SF (single frequency) impulsive tone signaling for supervision and dialing.

Needed for FDM multiplexing.• MF (multi-frequency) dialing digit representations.

– The above 3 methods are each historically called “in band” even when analog frequency band is not involved (for example, in T-1).

• When digital multiplexing (eg, T-1) arrived, MF dialing signals and “robbed bit” supervision signaling was used.

• Many telephone networks formerly or presently use(d) in-band tone signals to represent dialed digits (DTMF “touch tone,” or another set of audible tones called multi-frequency MF)

– DTMF uses two simultaneous tones taken from a set of 8– MF uses two simultaneous tones taken from a different set of 5

• Common channel signaling (SS7 ISUP) is dominant today– Earlier versions were SS6 and SS7 TUP.

*R2 and other signaling methods used outside of North America are not covered in this lecture.

©1997-2005 R. Levine Page 52

DC Pulsing Signals• Before dial telephones, for inter-office (inter-switch) signaling (usually in same

city) the human “operator” connected through a trunk to another human operator at destination switch. Requests for a destination number were verbal. When subscribers ended the call, buzzers at each cord board signaled the operators to manually “tear down” (unplug) each link.

• Earliest step switch dial signaling methods accessed specific inter-office trunks in response to dialing the first (typically) 3 digits. The remaining dialed digits were out-pulsed in the same form as “locally” dialed signals (that is, brief 40 ms interruptions in the loop current) on the the same wires as the eventual voice channel, and operated the last stages of step switches in the destination switch. A “long” interruption in the loop current ultimately caused automatic disconnect of all the step switches involved in the call, just as for a “local” (same switch) call.

• This system was difficult to evolve further:– Required many permanently installed, directly-connected but not always frequently

used inter-office trunks.– Long distance voice transmission with electronic (vacuum tube) amplifiers began

about 1914. (Before that, very costly thick low-resistance trunk lines were installed between some cities.) Amplifiers also required conversion from 2-wire switching (used in each end switch) to 4-wire switching in the trunk link.

©1997-2005 R. Levine Page 53

Single Frequency (SF) Signaling• In the 1920s and 1930s, analog frequency division multiplexing

(FDM) was introduced into the long distance telephone network. Typically 12 conversations were multiplexed on the same 4 trunk wires, by means of amplitude modulation of each conversation (in each direction) onto a distinct carrier frequency.– Similar to having 12 radio broadcasting stations, with 12 pre-tuned

radio receivers. Except the radio frequency signals were all carried by wires and not “over the air.”

– DC or baseband pulsing was not feasible with FDM.

• In each conversation signal in each direction, a single frequency (2600 Hz) tone was transmitted in the voice channel to indicate the absence of subscriber loop current.– Steady SF tone indicated that the channel was idle, or that the

subscriber at the relevant end had hung up.– Out-pulsing of the SF tone at the beginning of a call setup conveyed

dialing digits as from a rotary dial.

©1997-2005 R. Levine Page 54

Multi-Frequency (MF) Digit Signals• SF for dialing was “slow,” and was also susceptible to theft of

service fraud from SF tones generated at the originating subscriber end by “hackers.”*

• MF dialing encoded each digit, and also a “start” (KP) and “stop” signal, each by means of a pair of tones taken from a menu of five distinct audio frequencies.– MF dialed digit signals were initially used with SF supervision.– This required a more sophisticated multi-tone receiver to decode

the tones and produce electric current pulses compatible with the switching equipment.

– New types of switch equipment (example: crossbar) were introduced in the 1930s that did not need sequential digit pulsing and thus connected the call faster when MF was used.

– MF signaling can out-pulse 10 digits in one second. Baseband and SF out-pulsing could take up to 10 seconds for this.

*Large scale “hacker” fraud did not become a widespread problem until the 1960s

©1997-2005 R. Levine Page 55

MF with Digital Multiplexing• When T-1 transmission was introduced (1961),

MF was still used to convey dialed digits.– Supervision was signaled via “robbed bit” signaling

(described in EETS8302 lecture notes and in Bellamy recommended reference)

• The same advantages and disadvantages still applied to MF encoding of the destination number

• Plus a new requirement:– Not flexible for evolving new services. Next slide shows

and example where other information is needed.

©1997-2005 R. Levine Page 56

Automatic Routing of Subscriber Dialed Calls• Automatic routing of long distance calls via several intermediate switches

(first introduced in late 1950s) called for more sophisticated signaling• Knowing only the destination telephone number was not sufficient

– A problem called “ring around” or “looping” can occur when an automatic call routing system tries several suburban intermediate switches surrounding a destination city, and eventually revisits the same intermediate switches

– The call signaling must identify each call so that the previous intermediate switches on its attempted route are not visited twice…

• Better to signal to originator to “try again later” and abandon the call

• Unlimited “ring around” will eventually gobble up all circumferential trunk channels, making the situation worse than it already is.

from originating switch

Destination switch

Blue represents trunksUsed in call setup attempt.

Red represents trunk groupsWith all trunks already busy.

Attempted connectionIs heading again forSwitch A…..

A

B

C

D

©1997-2005 R. Levine Page 57

Common Channel Signaling• The first common channel signaling system (1970s,

CCITT No.6 or common channel interoffice signaling – CCIS) used fixed length fixed structure messages via a 2400 bit/second modem in a pre-designated channel.

• Signals in the “common”channel controlled call setup and disconnect and some routing features in any voice channel.– Prevented hacker fraud– Still not versatile for future evolution

• Replaced starting in 1980s with common channel system number 7.– Higher bit rate, faster call setup– Versatile for future evolution

©1997-2005 R. Levine Page 58

Signaling System 7• Almost a world wide standard

– Some variations in versions and features in different national telephone networks

• High bit rate. Two implementations:– Uses one voice channel, carrying 56 kb/s or 64 kb/s– Alternative implementation uses all of a T-1 bit rate (1536 kb/s)

except the synchronizing bit.

• Flexible– Message formats provide for future addition of new parameters

• Reliable– Typical SS7 network has extensive duplication of data packet

switches and geographically diverse alternate transmission channels, and includes continual built-in channel testing and route around failed links/switches.

©1997-2005 R. Levine Page 59

Main SS7 Message Types• Standard SS7 uses 4 Message types for ordinary call setup, but

other types exist and more can be added– Initial Address Message (IAM) orig->dest

• Contains dialed and originator’s numbers, and other information– Address complete msg. (ACM) dest->orig

• Indicates if destination Line is ringing or busy– Answer (ANS) dest->orig

• Indicates that destination line has been answered. • SS7 is designed to allow postponing assignment of a voice

channel until this event. This feature uses voice channel capacity much more effectively but will be implemented only in the future when the world PSTN is 100% SS7.

– Release (RLS) and its acknowledgements- either direction

• Causes disconnect process.

• In addition to the “standard” parameters of each message type, new parameters can be added when required.

©1997-2005 R. Levine Page 60

SS7 Services Examples: • Calling Line Identification (Caller ID)

– Originator telephone number sent as a parameter in the IAM to the destination switch. Sent to destination line between first two rings via a modem tone, if not “blocked.” This was the economic “killer app” for SS7

• Call back to last unanswered caller• Call completion to busy subscriber

– Monitors busy line when denied caller requests, and establishes connection with caller after it hangs up.

• Redirect “800” calls and other pre-designated calls to alternate numbers– When you dial 1 800 DOM-INOS, directs call to the

nearest Domino’s pizza retail store. Uses data base indexed by first 6 digits of the originator’s number.

©1997-2005 R. Levine Page 61

Future Telephone Signaling• Voice over Internet Protocol (VoIP) using speech digitally

coded at low bit rate into data packets has caused great interest in use of the Internet and the traditional PSTN to set up calls. Two protocols are in use for these applications:

• H.323, originally designed for multi-media and conference calling via Internet, is the first historical method. Similar in many ways and designed to be backward compatible with SS7. Uses binary number parameters in messages.

• Session Initiation Protocol (SIP) is simpler for non-telephone experts to understand, and uses alphabetic (ASCII) characters and printable numeric digit parameters.

©1997-2005 R. Levine Page 62

Advanced “Intelligent” Network (AIN)• Numerous advanced features utilize SS7 messages. Examples:

– Call completion to busy subscribers (CCBS). Originator dials *66 (auto redial‡) upon reaching a busy line, then hangs up.

• Destination switch automatically sends a special SS7 message when desired but busy destination person hangs up.

• Origination switch rings back the originating/requesting line. If origination person answers, origination switch sends a message to destination switch causing destination line to ring.

• If destination person answers, conversation proceeds.

• SCP data base can “translate” 800/888/877 dialed numbers into destination numbers appropriate to the caller, the time of day, etc.

– Callers to Sears Roebuck, Domino’s Pizza, actually reach the geographically closest retail outlet, or the east coast customer order center in the morning and the west coast center later in the day, so order takers work shorter hours at each such location!

– Translation based on origination calling line ID and extensive data base relating each NPA/NXX to the directory number (DN) of the nearest retail outlet, etc.

‡ This marketing name is misleading, because the network does not actually re-dial the desired destination repeatedly. It is “dialed” only once, after the originator answers the ring back.

©1997-2005 R. Levine Page 63

Local Number Portability• The objective of LNP is to promote local service competition.• The FCC (and some other national telecom regulators such as

OfTel in UK) have legally mandated “portability” of telephone directory numbers when a subscriber changes from an existing (“donor”) LEC switch to a new (“recipient”) competitive local exchange carrier (CLEC) telephone service provider switch.

– In North America, several centralized duplicated SCP/STP data bases provide a translation between the subscriber’s DN and a special “dummy” number (called a location routing number -LRN) located on the new recipient CLEC switch.

– After this translation, the call is routed to the recipient destination switch using the LRN’s NPA/NXX. The actual dialed number is carried in a separate special information parameter in the IAM message.

• When the call is routed to the recipient destination switch, the destination switch connects the call to the internal subscriber line designated in a special translation table which relates such “ported” DNs to the physical lines (ILANs) in the switch used for that purpose.

©1997-2005 R. Levine Page 64

Proposed “System Beta”

• Your instructor proposed a software-based network technology* in 1998 to allow all of one subscriber’s business or residence telephone lines to each be dialed using the same 7 (or 10) digit decimal telephone number.

– The subscriber may group all business lines (voice, fax, cell/PCS, pager, data, etc.) under one number and all residential lines (teenager’s line, home fax, etc.) under another number

– The different lines in a group are distinguished from each other by means of functional purpose (FP) codes which are mostly pre-set by the user

• FP codes for each line are stored in a suitable data base (perhaps the same data base is used for LNP!)

• The internal network SS7 signal messages carry the FP codes in separate parameters of existing messages

– Multi-use lines have a preset normal FP (typically voice) which can be temporarily superceded on individual calls by a dialed prefix (e.g., a fax machine automatically dials a *329 prefix, indicating that the desired destination is a fax line during this call only).

*US patent 6,076,121 issued June 13, 2000, and foreign patents pending.

©1997-2005 R. Levine Page 65

Beta Status• Aside from making it easier to remember all the different telephone numbers of your

correspondents, System Beta reduces the problem of telephone number exhaustion by reducing the quantity of telephone numbers for most subscribers to just one or two, regardless of the quantity of lines in service!

– In some cases most of the lines used by a subscriber will have internal non-decimal BCD digits in their internal network telephone numbers. This is not visible to the end user, but it stops the present growth in use of multiple decimal telephone numbers by subscribers.

• Also permits blocking undesired callers by functional purpose (e.g., block all unsolicited sales calls) and other new optional features

– FCC now considering System Beta due to above benefit and uses for the disabled (e.g. automatic routing of special 911 or teletypewriter for deaf calls)

• Estimated complete one-time development and installation cost is approximately $7 Billion to the telephone industry (mostly administrative and testing costs).

– At present the telephone industry spends about $1billion each year due to area code changes, with no end in sight to this recurring cost. Possibly additional $7 Billion in 2005 to 2020* to change to 4-digit area codes.

– Cost of all area code changes and 10-digit dialing to the public is estimated to be much more ($50 Billion to $150 Billion).

• System Beta was tabled by the T1S1.3 standards committee which defines North American SS7 signaling message standards.

– Waiting for direction from a major carrier or FCC to take it off the table.

• Could permit “recombining” previously split area codes, restoring 7-digit local calling, in perhaps 2 to 8 years after installation.

*Industry predictions of date when 11 digit NANP will be needed varies with the source and the date of estimate.

Documents

©1997-2005 R. Levine Page 1 Digital Switching EETS8320 SMU/NTU Lecture 9a Switching Software (print slides only, no notes pages)