S38.

3115 Signaling Protocols – Lecture Notes lecture 4

S38.3115 Signaling Protocols – Lecture Notes

Lecture 4 – Integrated Services Digital

Network (ISDN)
Motivation for ISDN...................................................................................... 1
Why is ISDN important even today........................................................... 3
Method of specification used in ISDN........................................................... 3
Digital transmission ....................................................................................... 3
Line Coding ............................................................................................... 3
Frame synchronization............................................................................... 4
ISDN Interface Types .................................................................................... 5
ISDN Reference Points .................................................................................. 6
Frame structure and synchronization on S-interface ..................................... 7
Power and power saving on BRI ............................................................... 8
Layer 2 or the LAPD -protocol...................................................................... 8
Telecommunications and Bearer Services ..................................................... 9
Q.931 ISDN signaling.................................................................................. 10
Identification in Q.931 ............................................................................. 10
Signaling modes....................................................................................... 11
Call setup and release............................................................................... 11
Connecting PABXs to ISDN ................................................................... 13
Signaling Efficiency..................................................................................... 14
A DSS1 signaling trace............................................................................ 15
Conclusion on ISDN signaling efficiency ............................................... 17

Motivation for ISDN

The grand idea behind ISDN was that the network should be digital end to end
and this would allow introducing new services to users and thus would help
the operators to create new value and differentiate on the market.

Also, it was seen that digital switching is going to be superior to

electromechanical switching because it will allow further automation of all
operator business processes. To take full advantage of digital switching it was
seen that native digital (= message based) signaling systems are needed. It was
obvious that this will allow eliminating such special purpose hardware as
DTMF receivers and senders, R2 multi-frequency receivers and senders1 etc.
The idea was also that message based signaling in combination with program

1
(NB: senders are mentioned for completeness, they are easy to implement for a limited set of
signals. One only needs to store the signals in a memory in digital form such that the signals
can be sent from the memory onto a PCM timeslot).

©Raimo Kantola (2008) 1

S38.3115 Signaling Protocols – Lecture Notes lecture 4

controlled implementation of switching systems would allow implementing

any number of services that would allow differentiation on the market.

First, digital signaling was introduced in trunk signaling (signaling systems 6

and later SS7). This is natural for two reasons: (1) digital switching first
appeared in trunk switches and was introduced with a delay of several years in
local exchanges. And (2) the changes in network nodes that are required to
implement something as drastic as a completely new signaling system in a live
network are limited to operator’s own sites – no work gangs need to be sent to
visit each and every subscriber. Note that a truck run to visit a single
subscriber costs much more than an operator collects from that subscriber in a
month.

On this course we will first talk about digital signaling for subscribers, i.e. the
DSS1 signaling system, although it appeared later than SS7.

It is natural to split the operator’s customers to two categories: (1) residential

subscribers and (2) business customers. The first usually have one or two
phone lines while the second often have a PBX or even a network of PBXs
that needs to be connected to a local exchange (or to several LEs). Small
businesses (SOHO – small office home office) may also have a key system
that in the analogue era was connected to the local exchange with a few
analogue subscriber lines.

For local operators, the most long term investment is the copper laid in ground
and connecting users to its local exchanges. By introducing digital signaling
on local loops, the operators thought that they will be able to add value to their
investment into their copper plant and that they will leverage the value of their
copper.

In practice it turned out that very few attractive subscriber services could be
invented such that were impossible to implement with analogue subscriber
lines and analogue subscriber signaling. Residential customers started to adopt
ISDN in their homes more or less widely only, when Internet connections at
home had become popular and modems for transferring data over analogue
phone lines started to feel clumsy and slow.

On the other hand, DSS1 was widely adopted for the purpose of connecting
PABXs to local exchanges and as a basis for specifying private network
signaling systems. With DSS1, we had the first rich signaling system that was
well specified and an international standard for corporate telephony services.
For example, DDI stopped to be a problem. Reachability became easier.
Clearly, easy reachability of company personnel is very valuable for
businesses that actively communicate with their customers.

©Raimo Kantola (2008) 2

S38.3115 Signaling Protocols – Lecture Notes lecture 4

Why is ISDN important even today

Using DSS1 for PABX connectivity and for connecting special devices such as
voice mail systems to exchanges is still common place. DSS1 has been used as
the starting point for specifying GSM signaling. In terms of the circuit
switched services 3G is quite similar to GSM.

So, even if DSS1 was not a huge success with residential customers, it became
sort of a mother of many other signaling systems and many of its principles
have been inherited to other more modern systems.

Method of specification used in ISDN

ISDN is specified in terms of logical functions and interfaces. When a standard
is implemented, one or several functions can be mapped to a single device
(box). Interfaces are for the purpose of connecting devices that may be
manufactured by different vendors. The idea is that each vendor implements
its own box as it pleases for example using a combination of hardware and
software. Because the implementation is left for the vendors, ISDN does not
specify Application Programming Interfaces (APIs).

ISDN was specified in a series of CCITT books (red book 1984, blue book
1988, Euro ISDN 1992). The blue book was the first specs that was more or
less accurate enough for implementation. However, only Euro ISDN gave
enough detail for implementing supplementary services so that it started to
make sense to produce user devices in reasonable volumes.

Digital transmission
Message based signaling can be understood by breaking the required
functionality into layers and features on each layer. Messages are made of bits,
these are organized into frames and finally inside the frames, we can find the
messages that carry the telephony signaling information.

Line Coding
In order to transfer bits to a distance on a wire, an agreement what a logical
one and a logical zero look like is needed. This agreement is called the line
code. Physical limitations of the wired connection may dictate some properties
of good line codes. Let us take an example.

The simplest code one can think of, for example on a physical copper
connection like the local loop, would be to send a short pulse to indicate a
logical one and zero voltage would indicate logical zero. In case of local loop
this has the disadvantage that the code itself creates a constant current
component (in a Fourier breakdown of the coded signal) that is dependent on
what is being sent. This is not nice because the local loop is used for power
supply and there may be things like transformers on the line card (something

©Raimo Kantola (2008) 3

S38.3115 Signaling Protocols – Lecture Notes lecture 4

like this is needed for transforming a two wire connection to a four wire
connection that is done on the line card).

To avoid creating a constant current component, a slightly more complex code

can be used: the code is called Alternate Mark Inversion or AMI-code. The
principle is that each logical one is sent with a different polarity than the
previous one. It is also possible to use Reverse-AMI: logical zeros are mapped
to pulses with alternate polarity and logical ones are sent with zero voltage.

The limitation with the AMI code is that a long string of zeros is difficult to
count. The receiver needs to have a very accurate clock to be able to count the
number of zeros that was sent. In reverse-AMI the difficulty comes with a long
string of ones.

The difficulty of a possible long string of zeros is resolved in the HDB3 –code
(more generally HDBn), see Figure 4.1. HDB comes from the words “high
density bipolar 3”. This line code is used for example in the PCM system.

In HDB3, the forth consecutive zero is replaced by a pulse of the same polarity
as the previous one violating the idea that there should be no constant current
component and the basic encoding of logical zeros and ones. This violation
needs to be balanced out. This is done by sending a balance pulse instead of
the next zero.

1 0 1 0 0 0 0 1 1 0 0 0 0 0 0 0 0 1 0
+V
AMI 0 or inverse
-V
AMI - alternate mark inversion

+V
B 0 0 V
HDB3 0 0 0 0 V B V or inverse
-V

HDB3 - high density bipolar 3

V - violation
B - balance

Figure 4.1 AMI and HDB3 line code.

Exercise: Make a coding table of 32 lines for HDB3 showing how all
combinations of 5 bits are presented by positive and negative pulses and zeros.

Frame synchronization
In digital serial transmission, once we are able to detect logical zeros and ones,
we need to spot bytes and strings of bytes. This requires that we detect the
beginning and the end of a string, we call it a frame. Different transmission

©Raimo Kantola (2008) 4

S38.3115 Signaling Protocols – Lecture Notes lecture 4

methods use different framing methods. The ISDN signaling channel uses a
well known method, inherited from HDLC (high level data link control).

HDLC framing is most suitable for transferring messages that are from tens to
a few thousand bytes in length. The idea is to mark the beginning and the end
of a frame with eight bits: “01111110”. This combination is called a delimiter.
The immediate question is: what if the user wants to send 01111110? How can
we make the distinction with the delimiter? The problem is solved by bit
stuffing.

In HDLC, the sender, when it detects a series of 5 consecutive logical ones,

always adds a zero onto the line. The receiver, always after 5 consecutive
ones, removes the immediately following zero. If the receiver detects 6
consecutive ones, it determines that this is a delimiter.

Bit stuffing and HDLC framing are implemented on silicon and higher
protocol layers need not worry about them. Processing HDLC framed
messages is natural for computers: the contents are moved to memory and the
message can be parsed using the CPU.

What if we need to send more than a few thousand bytes? We will discuss
other frame structures later on this lecture.

ISDN Interface Types

ISDN Basic Rate Interface (BRI) is meant for residential customers and small
offices. It uses a single copper pair of quality that was typical of countries
where PSTN was widely spread. (Note that in offices today wiring is often
CAT5 or better while the copper wiring in the ground is mostly of much
poorer quality.)

Basic rate interface offers 2 data (or voice) channels of 64 kbit/s each, plus a
signaling channel on 16 kbit/s. So, a customer, using the existing copper
wiring, could have 2 independent voice channels instead of one provided by
PSTN. This turned out to be of limited attraction to the users. The added value
came to fruition when faster connections to the Internet were needed. The data
(or voice) channels are called B-channels (short for bearer) and the signaling
channel is called the D-channel.

ISDN alternatively offers the Primary Rate Interface (PRI) that in practice
makes use of PCM transmission. PRI is meant for connecting PABXs and
servers like Voice Mail systems to exchanges. One PRI offers 30 data (or
voice) channels and one signaling channel each on 64 kbit/s.

Both in BRI and PRI the signaling channel is used for signaling related to calls
that use any of the B-channels on that interface. The D-channel can also be

©Raimo Kantola (2008) 5

S38.3115 Signaling Protocols – Lecture Notes lecture 4

used for transferring packet data. This possibility of transferring packet data
has been made some use of in ISDN. The possibility later found a great
success in GSM where signaling channels are used to carry short messages
between users. One example of the use of packet data on D-channel is the
always on-line connection to the Internet that was implemented using BRI.
The use of this service in ISDN was short lived because ADSL came very
quickly after its introduction.

From the ISDN exchanges point of view B-channels are continuous 64 kbit/s
bit streams that are connected to its switching fabric, switched to PCM
timeslots for wide area connectivity and allocated for calls based on signaling
that takes place on the D-channel.

ISDN Reference Points

Reference points are standards for connecting devices that come from different
vendors or may be owned by different stakeholders (operators or customers).

BRI separates the home wiring (that is typically owned by the house owner)
from the copper pair operated by the operator. On this border, a Network
Termination box (NT) is installed. Originally, the operators wanted to make
this the demarcation point for responsibilities. Regulation forced the operators
to offer BRI using different interfaces. Towards the user devices an NT offers
the S-interface (or reference point). On the copper wire towards the exchange
we have the U-interface.

The relationship of a customer extension in the local exchange to user devices

on a BRI is one to many (max 8). This means that the max of 8 user devices
such as ISDN phones can be connected to a single BRI.

S-interface wiring uses 8 wires, two in each direction for data and four for
power, connectors are RJ45. The S-interface was in international standard
from the beginning. Its gross data speed is 192 kbit/s containing 2 × 64 kbit/s +
16 kbit/s = 144 kbit/s of user accessible data and 48 kbit/s of overhead. Line
coding on the S-interface is reverse AMI. The U-interface that became a
standard was based on 2B1Q line coding on 160 kbit/s. Naturally, the same 2
data channels and the signaling channel with the total of 144 kbit/s need to be
carried. On U-interface, the amount of overhead is 16 kbit/s.

On PRI, ISDN also provides the T-interface between the Network Termination
1 and Network Termination 2.

Between the Line Termination (LT) and the Exchange Terminal in the local
exchange, ISDN specifies the optional V-interface. This was elaborated to
V5.1 and V5.2 interfaces between subscriber multiplexers and LEs during the
1990’s. These were however implemented by major wire-line switching
system vendors in a slightly different manner and the goal that the operators

©Raimo Kantola (2008) 6

S38.3115 Signaling Protocols – Lecture Notes lecture 4

had in mind was only partially reached. The idea was that since ca. 70% of LE
cost is in the access part, let’s put a standard interface between the core switch
and the access part and let the vendors compete in the two areas separately.

ISDN also talks about the R-interface. In fact this is a pseudonym for any
legacy interface in the customer premises that can be mapped using a Terminal
Adaptor (TA) box to the S-interface.

When ISDN was deployed the boxes that the operator installed at homes were
often actually combinations of the functions that are described in ISDN
specifications. The home box could contain a TA for example for 2 analogue
phone lines and an NT1 with 3 RJ45 jacks for S-interfaces.

Frame structure and synchronization on S-interface

Transmission on the S-interface is meant to be continuous. Line coding is
AMI. To make sense of the bits, frames of 48 bits, 4000 times per second are
sent in both directions (NT to user device and vice versa). The beginning of
the frame is detected based on AMI code violation. Bit number 1 (F-bit) and
number 14 (FA) are zeros with a wrong polarity. There is a two bit time shift
between the downstream frames (from NT to terminal) and upstream frames
(TE to NT).

The code violation needed for frame synchronization is balanced using the L-
bit.

Figure 4.2 shows the S-interface frame structure. The 48 bits in a frame have
two samples for each B-channel (4000 times per second instead of 8000 times
per second in PCM). Four of the bits in a frame are allocated to the D-channel.
Next to D-channel bits, the NT echoes back downstream the D-channel bits it
has received from a terminal. This uses another 4 bits on the downstream
frame. The A-bit is used for Activation of the interface in the downstream
direction. All this uses 1 + 1 + 4 × 8 + 2 × 4 + 1 = 43 bits. The rest is for
achieving DC balance and future extensions.
NT -->TE
48 bit frame

FL 8B1 ED AF N A 8B2 ED M 8B 1 ED S 8B2 ED L

00 01 0 D 0 0

‘wrong’ polarity Echoed D-channel

TE -->NT
FL 8B1 L D LF L
A 8B 2 L D L 8B 1 L D L 8B2 L D L
00 0 000 0 D 0 0 0 0 0

2 bit delay
F Frame alignment bit B1 channel 1 bits A Activation bit
FA Frame synchronization bit B2 channel 2 bits M Multi-frame bit (normally 0)
L DC balance bit D D-channel bits S Future use (0)
N Inverted FA (normally 1) E Echo channel bits

Figure 4.2: Frame structure on the S-interface.

©Raimo Kantola (2008) 7

S38.3115 Signaling Protocols – Lecture Notes lecture 4

Because up to 8 terminals can be connected to an S-interface, they need to

compete for channel access. The echo channel shows to a sending TE that NT
has received what it is sending. If the content does not match to what a TE is
sending, signals from two TEs may have collided on the S-interface or if the
problem persists, there may be a problem with the NT itself. For the case of
collision a back-off algorithm is needed.

Let us summarize the discussion of framing in digital transmission so far. We

are now familiar with HDLC framing for short sequences of bytes and two
different framing systems for continuous digital signals: PCM framing and S-
interface framing.

Power and power saving on BRI

For emergency calling it is required that besides data the LE sends power on
the copper pair to the NT. It should be possible to make an emergency call
without mains power supply (e.g. 240V AC) being present on the customer
premises. Feeding subscriber lines that are not in use most of the time would
be a burden and waste of energy. Therefore, the ISDN BRI interface usually
goes to a power saving mode when it is not in use. To wake up from this
mode, an activation procedure is required. Activation can be initiated either by
a terminal or the LE.

A Terminal starts by sending INFO1 frames. When LE detects INFO1 or when

it wishes to start the activation, it sends INFO2 frames. When Terminal detects
INFO2 frames in the downstream, it starts sending active state INFO3 frames.
LE acknowledges by starting to send INFO4 frames.

After this activation procedure, framing on the S-interface is in place and

signaling on the D-channel can start based on the HDLC framing that we
discussed earlier.

Layer 2 or the LAPD -protocol

The transfer of D-channel frames corresponds to layer 2 in Open Systems

Interconnection (OSI). Layer 2 frames can carry either signaling messages of
layer 3 or packet data. Layer 2 protocol in ISDN is called LAPD for “Link
Access Protocol on D-channel” and it is specified in Q.920 –Q.921. Layer 2 is
concerned with a reliable transfer of data between a set of terminals and the
LE. Layer 2 does not understand about calls. Calls are the concern for layer 3,
the DSS1 signaling that is specified in Q.930-Q.931.

To be more exact, LAPD serves for the purpose of reliable transfer of chunks
of data between many terminals and many layer 3 entities in the LE.

©Raimo Kantola (2008) 8

S38.3115 Signaling Protocols – Lecture Notes lecture 4

The first thing one almost always needs to understand about a protocol is
how the communicating parties are identified.

In case of LAPD, the parties are identified with a Data Link Connection ID
(DLCI) field that is carried in each frame. DLCI normally takes 16 bits after
the delimiter marking the beginning of the frame. The DLCI can be extended
to larger values if needed. DLCI has two subfields: SAPI or Service Access
Point Id and TEI for Terminal Endpoint Id. SAPI identifies the layer 3 or
management entity to which the frame is addressed. TEI values can be
dynamically allocated to terminals in a BRI. LAPD has procedures for the
dynamic allocation of TEI values.

Before any higher level (Q.931) functions can be performed, each ISDN
device must be assigned at least one unique TEI value. These numbers can be
pre-assigned (TEIs 0-63), or dynamically assigned (TEIs 64-126). Most TEI
assignment is done dynamically, using the TEI management protocol. Using
dynamic allocation eases configuration.

Next after the address field, LAPD has a control field of two bytes. This is
used for identifying the types of frames (I, S and U –frames) and carrying
frame numbers. LAPD supports both an acknowledged mode of data transfer
with I-frames and unacknowledged mode with IU-frames. In acknowledged
mode LAPD supports windowing: a certain number of frames can be sent
before an acknowledgement MUST be received prior to sending more. By
frame numbering reliable transfer of data can be achieved.

Each layer in a protocol stack usually needs some management functions. This
is called layer management. An example are the TEI management procedures
we already mentioned.

After the address and control bytes, a LAPD frame has room for the payload.
The frame ends with a two byte checksum and the ending delimiter.

Telecommunications and Bearer Services

In ISDN the term “telecommunication service” refers to functionality that
covers all OSI layers. The term bearer service refers to data transfer on layers
1 up to 3. In such bearer services as audio, speech or transparent 64 kbit/s data
only layer 1 seen by a local exchange.

9 Circuit switched bearer services include:

› Speech
› 3,1 kHz audio
› 7 kHz audio
› transparent 64 kbit/s.

©Raimo Kantola (2008) 9

S38.3115 Signaling Protocols – Lecture Notes lecture 4

9 Packet switched bearer services include:

› virtual call and permanent virtual connection,
› connectionless packet switched service on the D-channel,
› user-to-user signaling information.

Telecommunication services make use of some bearer service.

Telecommunication services are further broken down into basic services and
supplementary services. An example of a basic service is a telephone call.
Supplementary services can not exist without a basic service.

Q.931 ISDN signaling

Q.931 is the layer 3 protocol used on Digital Signaling System Number 1
(DSS1). It is carried in the payload of LAPD.

Because D-channel is separate from the voice channels, DSS1 is out-of-band

and can continue independent of the use of voice channels. Encoding of Q.931
is binary (not for example text) and the style is that information in the
signaling message is grouped into information elements. Each information
element has a type field that identifies the kind of information that can be
contained in the element. If the element is of variable length, a length field
gives the number of bytes in the element.

Each message has a message type field. This gives an idea about the main
function or state transition that the receiver should execute on reception.

Identification in Q.931

Q.931 uses telephone numbers for identifying subscribers. The layer 3 entities
related to a call are identified by Call References that are arbitrarily chosen by
the communicating parties (Terminal or LE). A signaling message can identify
the B-channel that is allocated for a call in an information element that is
carried inside the signaling message.

All this means that Q.931 signaling is also logically separated from the voice
channels. Signaling entities can be created and they can live independent of the
allocation of voice channels.

ISDN BRI supports the max of 8 terminals in a single S-interface. They can all
use the same telephone number or each one can have its own telephone
number. The first option leads to a need to identify compatibility of the
terminal with the service that is being requested. For this purpose Q.931
supports Lower layer compatibility information and High layer compatibility
information elements during the call setup procedure. Compatibility info can
for example define that a G4 fax machine is requested and that other terminals
in the S-interface should not react to an incoming call. Actually, this violates

©Raimo Kantola (2008) 10

S38.3115 Signaling Protocols – Lecture Notes lecture 4

the layering principles that are assumed in OSI. Compatibility relates to higher
layers than layer 3. Nevertheless, the information about compatibility is
carried in Q.931 that is supposed to be a layer 3 protocol. The advantage is that
if there is no compatible device connected to the concerned S-interface, no
charges will be incurred to the calling subscriber. Instead the call will fail. The
alternative would be to let the user devices exchange whatever information
they wish prior to setting up a call (for example of less than N bytes). This,
however, did not meet the operator’s charging requirements. Letting users
communicate in any way prior to setting up a call and starting charging just
does not follow the telephone operator principles.

Signaling modes
Q.931 supports two modes of signaling: the functional mode and the stimulus
mode. In functional mode information is encoded in service specific
information elements. This sounds very logical but unfortunately, it means that
software in both ends of communications needs to be updated if new
functional information elements are needed for new services. There is not such
thing in ISDN as software download to Terminals from the network. This
leads to cumbersome updating of Terminals.

Stimulus mode means that signaling message carries the keying sequence on
the phone to the LE. It is possible to program new keying sequences on
programmable phones (soft keys) and implement new services just by
updating the LE.

Q.931 supports overlap sending. This means that the dialing sequence can be
as it was for PSTN phones: pick up the receiver, hear dial tone, push buttons
for dialing and wait for the ringing tone.

Call setup and release

DSS 1/Q.931 signaling is symmetric. This means that originating call

signaling and terminating call signaling use the same messages, just the
direction is (LE –terminal) is reversed. Symmetry applies to both call setup
and release.

The first message for call setup is the SETUP message. SETUP message
contains at least the bearer service (= is this a data call or audio call). In
practice it is a good idea to include as many digits in the SETUP message as
there are in the shortest directory number in the local dialing plan (4 in the
Finnish ISDN network e.g. 4511 for TKK), so the originating exchange can
attempt routing immediately having received the SETUP message.

©Raimo Kantola (2008) 11

S38.3115 Signaling Protocols – Lecture Notes lecture 4

SETUP_ACK acknowledges the reception of SETUP. It is more useful on the

terminating side – tells that the kind of device that can support the bearer and
telecommunication service requested has been found (e.g. a G4 fax machine).
INFO messages support overlap sending – the result is that the routeing can be
done through the network with the minimal number of digits (in Finland
usually 3). In practice INFO messages are useful only in the TE to LE
direction for the TE to send each new dialed digit in a separate message. This
will lead to the fastest possible call setup: often the ringing tone is heard
immediately after pushing the last button on the phone.

NB1: In case there is no number portability (NP), the last digits (in Finland
4…8) are needed only at the terminating exchange, the rest of the exchanges
do not even look at the last digits. If NP is allowed in the number block, all
digits need to be received prior to routeing decision at the originating
exchange.

NB2: There is no such thing as number complete indication from the terminal,
instead the network deduces when all digits have been received by executing
number analysis after each digit or at times instructed by former routing
information retrieved based on earlier digits received for the call.

CALL_PROCEEDING in the backward direction – tells that no more digits

will be needed even at the terminating exchange, it also tells that at least the
network is not busy (the B-subs can still be busy).

ALERTING in the backward direction tells that the phone at B-subscriber

(called party) is ringing. The D-channel message is accompanied by a ringing
tone from the terminating exchange to the caller on the audio channel. This
tells that the audio channel is clear at least in the backward direction from the
terminating exchange to the caller.

CONNECT message starts charging and all exchanges through-connect the B-

channel in both directions.

CONNECT ACK is really necessary at the terminating side: tells which of

max of 8 devices that can be connected to the S-interface the call has been
awarded (imagine two or more phones being picked up almost at the same
time each sending a CONNECT).

DISCONNECT – charging stops and the B-ch connection is torn down.

RELEASE tells that DISCONNECT was received. RELEASE COMPLETE

confirms the reception of RELEASE and tells that after this the B-channel can
be used for another call on the subscriber interface. For call tear down in
exceptional cases even RELEASE and RELEASE COMPLETE are enough.

©Raimo Kantola (2008) 12

S38.3115 Signaling Protocols – Lecture Notes lecture 4

One should also note that the call can be released any time during call setup if
the caller goes on-hook. Instead of CONNECT, there may be a
DISCONNECT in the backward direction etc.

If one takes a look at the call setup procedure, it becomes clear that Q.931 does
not follow the client-server model. Instead the model is that of communicating
finite state machines.

Connecting PABXs to ISDN

We already mentioned that DSS1 has been widely used for connecting PABXs
to ISDN networks and as a basis for private network signaling systems. This
trend started already prior to CCITT (predecessor of ITU-T) specifications of
DSS1 were ready. In the UK pre-ISDN standards for DASS and DPNSS were
developed.

DASS is a signaling system for connecting a PABX to a public network.

DPNSS is a digital private network signaling system for setting up PABX
networks. Such a network is needed in a multi-site company with thousands or
tens of thousands of employees. DPNSS supports a rich set of supplementary
services that may be useful in a corporate network. A signaling system that is
used to connect a PABX to a public network (DASS or DSS1) does not
necessarily need to and usually does not support all the corporate network
services.

It is also possible to offer PABX-like corporate network services using a

public network switching system for service implementation. This possibility
was invented because at some point in the US the Bell Operating Companies
providing local telephone services were forbidden to offer PABX services to
companies. They tackled the market limitation by implementing similar
services in their own Central Offices and they called the service offering:
Centrex. Also a combination of PABX and public switch implementation of a
private network is possible. In that case the signaling between the concerned
PABXs and the public switch is for example DPNSS or Q.SIG. The latter is an
international standards based private network signaling system. Q.SIG was
defined taking DSS1 as the starting point and extended with features that are
typical in a private network.

A PABX is normally connected to an LE using subscriber criteria. This means

that the LE records the outgoing call minutes or seconds or creates Call Data
Records for the calls and the local operator sends a bill for the calls each
billing period based on business subscriber tariffs. The set of supplementary
services supported across the connection is the set of services that the
concerned public network supports. The bill may be charged to the PABX
number rather than recorded for each extension of the PABX.

©Raimo Kantola (2008) 13

S38.3115 Signaling Protocols – Lecture Notes lecture 4

Because DSS1 or any of its variants support sending numbers downstream to a

PABX, they all naturally support Direct Dialing In (DDI). So, with a digital
signaling system it is easy to support a setting where in the private network
short numbers (e.g. 4 digits) are used for calling inside the private network
(private numbering plan) while all extensions still have the DDI capability.

Signaling Efficiency

Signaling efficiency generally refers to the performance aspects of signaling.

Different metrics are used for different purposes. If we are interested in the
user point of view, post dialing delay is a useful metric. According to E.721 of
ITU-T call setup delay (also known as post-dialing delay or post-selection
delay) is defined as the interval between entering the last dialed digit and
receiving a ringing tone.

DSS1 and particularly SS7 improved network performance significantly by

reducing post-dialing delay.

Another angle is the signaling system designers and operators point of view. In
this case we are interested in the use of network transmission capacity and
other resources that are needed for signaling. For digital signaling systems a
good metric is look at the size of typical signaling flows created using the
signaling system. Flow size can be measured in bits. Bits can be turned into
time spent by the signaling flow by a simple formula:

Transfer delay = size in bits/signaling channel speed. (1)

This measure is easy to use for a signaling system. For post dialing delay
measurements need to be made across a network or calculations must be based
on numerous assumptions about the network. Therefore, post dialing delay is
suitable for characterizing a network rather than a single signaling system.

We can make the following considerations:

All bits in all calls create signaling traffic – the amount of signaling traffic
should be significantly less than the signaling channel capacity
• Users pay for the service: businesswise signaling is pure overhead – it
follows that the following should apply:

Amount of signaling bits in a call flow << amount of bits in the service

• Amount of bits in the service is e.g. amount of voice bits in a call

• If calling is a flat rate service, amount of signaling bits is not that
important

©Raimo Kantola (2008) 14

S38.3115 Signaling Protocols – Lecture Notes lecture 4

A DSS1 signaling trace

Let us measure ISDN signaling efficiency for BRI in terms of bits and transfer delay.

The flow has been captured in the TKK Netlab by Vesa Kosonen. Layer two flags and the LAPD
checksums are not shown in the captured flow. The presentation uses hexa code instead of binary.
In hexa code four bit combinations are replaced by hexadecimal numbers from 0, 1, 2, …A, B, C,
D, E and F.

From one ISDN phone (29) to another ISDN phone (32) ENBLOCK SENDING- HEXA
representation

Conn:1 Card:1 Channel:D 52 11:50:09.565

RESYNCRONIZATION
Conn:1 Card:1 Channel:D 53 11:50:09.567
ACTIVATION REQ
Conn:1 Card:1 Channel:D 54 11:50:09.570
ACTIVATION IND
Conn:1 Card:1 Channel:D 55 11:50:09.580
SABME
Conn:1 Card:1 Channel:D 56 11:50:09.601
UNNUMBERED ACK
Conn:1 Card:1 Channel:D 57 11:50:09.644
L2: INFO 009900000801130504038090A31801836C02008070038033327D0291817E0104
L3: SETUP 0801130504038090A31801836C02008070038033327D0291817E0104
Conn:1 Card:1 Channel:D 58 11:50:09.692
RECEIVER READY
Conn:1 Card:1 Channel:D 59 11:50:09.817
L2 INFO : 029900020801930218018A
L3 CALL PROCEEDING : 0801930218018A
Conn:1 Card:1 Channel:D 60 11:50:09.826
RECEIVER READY
Conn:1 Card:1 Channel:D 61 11:50:10.370
L2 INFO : 0299020208019301
L3 ALERTING : 08019301
Conn:1 Card:1 Channel:D 62 11:50:10.378
RECEIVER READY
Conn:1 Card:1 Channel:D 63 11:50:14.019
L2 INFO 0299040208019307290569010714247C038090A3
L3 CONNECT : 08019307290569010714247C038090A3
Conn:1 Card:1 Channel:D 64 11:50:14.027
RECEIVER READY
Conn:1 Card:1 Channel:D 65 11:50:14.866
L2 INFO : 0299060208019345080281901E028188
L3 DISCONNECT : 08019345080281901E028188
Conn:1 Card:1 Channel:D 66 11:50:14.874
RECEIVER READY
Conn:1 Card:1 Channel:D 67 11:50:15.117
L2: INFO : 009902080801134D08028090

©Raimo Kantola (2008) 15

S38.3115 Signaling Protocols – Lecture Notes lecture 4

L3: RELEASE : 0801134D08028090

Conn:1 Card:1 Channel:D 68 11:50:15.166
RECEIVER READY
Conn:1 Card:1 Channel:D 69 11:50:15.282
L2 INFO : 029908040801935A
L3 RELEASE COMPLETE : 0801935A
Conn:1 Card:1 Channel:D 70 11:50:15.290
RECEIVER READY
Conn:1 Card:1 Channel:D 71 11:50:20.414
DISCONNECT

By cleaning up we get Table 4.1. LAPD (layer 2 content) is shown in bold. The reminder is Q.931.
The used telephone numbers are just two digits. In our comparison, we need to add a reasonable
amount of digits to this flow to account for some average number length.

Table 4.1: An ISDN Signaling Flow.

009900000801130504038090A31801836C02008070038033327D0291817E0
SETUP 104
CALL PROC 029900020801930218018A
Alerting 0299020208019301
Connect 0299040208019307290569010714247C038090A3
Disconnect 0299060208019345080281901E028188
Release 009902080801134D08028090
Release
Complete 029908040801935A

Counting the bits message by message and accumulating the bits over the whole flow we get Table
4.2.

Table 4.2: Summary of the ISDN flow

A B C D E F G H
Cumu- Add
N-len lative ms CK+Deli-
Message Hexa content LEN(B) Bits =8 of E of F miters
00990000080113050403809
0A31801836C020080700380
SETUP 33327D0291817E0104 64 256 352 352 22 24
CALL
PROC 029900020801930218018A 22 88 88 440 28 30
Alerting 0299020208019301 16 64 64 504 32 34
02990402080193072905690
Connect 10714247C038090A3 40 160 160 664 42 44
Discon- 02990602080193450802819
nect 01E028188 32 128 128 792 50 52
Release 009902080801134D08028090 24 96 96 888 56 58
Release
Complete 029908040801935A 16 64 64 952 60 62

©Raimo Kantola (2008) 16

S38.3115 Signaling Protocols – Lecture Notes lecture 4

The cumulative milliseconds of the ISDN signaling flow, assuming a 16 kbps signaling channel in
the 2B+D interface, en-block sending, 8 digits for both the Called Party Number and the Calling
Party Number and taking into account both the Q.931 bits and the LAPD bits in the frames that
carry the signaling information are in column H of Table 4.2. The result is also presented
graphically in Figure 4.3.

Figure 4.3: Summary of an ISDN call setup and Release

Signaling message delay on 16 kbps in ms

Enblock sending, Number length is 8 digits

70
60 62
58
50 52
44
40
MS

34
30 30
24
20
10
0 e
t
t

e
g
C
P

ec
ec

et
as
tin
O
TU

n
n

pl
PR

e
er

on
on
SE

om
el
Al

R
isc
L

C
AL

e
C

as
e
el
R

Table 4.3 presents some bit count metrics on the ISDN signaling flow.
Number length
ISDN signaling bitcount summary =8
L3 L2+L3 L2 overhead %
Call setup signaling (enblock) 536 664 24 %
Avg message in call setup 134 166
Call setup and release total 728 952 31 %
Avg message in total 104 136

Conclusion on ISDN signaling efficiency

The signaling transfer delay occurs due to the need to transfer the signaling bits on a signaling
channel of finite speed. We calculate signaling transfer using formula 1.

Transfer delay = size/signaling channel speed.

The total signaling transfer delay for call setup on a 2B+D access is less than 50 ms and total delay
including also call release is slightly above 60 ms.

©Raimo Kantola (2008) 17