Note: Descriptions are shown in the official language in which they were submitted.
CA 02203907 1997-04-28
wo 96/l5634 r~-lr~ ~ o o ~17
Overload ~ ve~ ~ion in a tel e ; ~-~tions network node
Field of the Invent$on
The invention relates generally to traf f lc
control in trlr ;~-~tion neL . More
spe~ f ~ 11 y, the invention relates to a method and an
elL~_, L for ~-~vt~ lng an overload in a
t-l e ; ~--tion network.
The invention ls intended ~pP~ l ly for so-
called Intelllgent Networks ( IN ) that are being
developed at present, but the same prlnciple can be
applied in any network wherein two or more nodes are
l.~Lc~ ected in such a way that at least one of the
nodes can be loaded by one or several other nodes.
B~c~y,~ d of the Invention
An Intr~ g~nt Network usually refers to a
network comprising more intr~ll;g~n.-r- (i.e. a better
ability to utilize information stored in the network )
than the present public ( switched ) networks . Another
characterlstic of the Intelligent Network is that the
network architecture somehow draws a distinction
between, on the one hand, the operations concerning the
switching itself and, on the other hand, the stored data
and its processing. Such a division makes it possible
that, in principle, the organization providing network
services can be different from the organization r-n_s;n~
the physical network in which the services are provided.
Conceptually, an Intelligent Network can be divided into
three parts . The f irst part comprises the nodes
switching traffic (performing connections), the second
part contains the services provided by the network, and
the third part consists of the internodal ;~tion
protocol, i.e. the "language" the r--h;nr-~ use to
, ; ~ ~te with one another . Since all services must
-
CA 02203907 1997-04-28
WO Y6/15634 r--lr- ~'7
be presented as a 9~ 'P of - g~ ~ conforming with
the protocol, the protocol defines the "intPl 1 igPnre~
of the network.
In order to facilitate the u-,deL~a..ding of the
present invention, reference is first made to a simple
basic situation illustrated in Figure l, wherein two
r--h; noe ( or network nodes ) l and 2 are shown, the
r--hinPg being inL~ ;u....ected by means of a si~n~lling
link 3. Machine l comprises a database D~, and machine
2 is a client asking questions from machine l by
transmitting 1 -sa~PR to machine l over the link 3. When
machine l receives a question, it initiates a
transaction resulting in an answer after a certain
~,~oces~lng period. When the answer is ready, machine 1
transmits it to machine 2 over the link 3. Each answer
costs machine 2 a certain sum.
A ~1-eo~ ical omnipotent machine l would answer
each question immediately so that the correlation
between the questions rate (questions per time unit) and
the answering rate ( answers per time unit ) would look
like the description in Figure 2a. However, there is in
practice a limit to how f ast machine l can provide
answers. Taking this into account, the le:S~u115t: curve
of the omnipotent machine l becomes like the one shown
in Figure 2b. When the questions rate exceeds a certain
threshold Amax corrPc:prnfl i ng to the highest possible
answering rate, the latter remains constant, i.e. some
of the questions will not be answered. However, this
situation does not correspond to a real situation,
either. In practice, the situation is such that as the
questions rate exceeds a certain threshold value for a
long period of time, machine l becomes overloaded so
that the increasing questions rate further reduces the
answering rate. This situation is illustrated in Figure
2c. The decreasing answering rate is due to the fact
CA 02203907 1997-04-28
WO96/15634 I~/rL_ ~17
that the machine starts wasting its resources, for
example in such a way that it reserves more and more
free memory for storing the questions, so there will be
A1n~ly less and less memory available for
computing the answers. The threshold value of the
questions rate at which an overload situation occurs is
not constant, but it depends on how much of the capaclty
of machine l is dedicated to answering. For example, the
threshold value is lower than usual when the database
DB of machine l is being updated.
The purpose of any overload L,lc:v~:nLlon method
is to make the curve ( Figure 2c ) describing a real
situation ~ as closely as ~Qss1 bl ~ the curve
( Figure 2b ) describing an ideal situation. On the other
hand, it is r~son~hl-~ to provide the overload
pr~:vt:nLion of machine l partly in machine 2, so that
machine 2 would not have to load the tr:~nc~mi Sc; I-n
connection between the r~~hin~c by transmitting messages
that would be discarded by machine l.
Suppose that in order to protect itself, the
overloaded machine l transmits to machine 2 a
restriction or filtering request with which it requests
machine 2 to reduce the number of questions to be
transmitted. Such a request typically contains two
restriction pal ttlL5 the upper limit of the questions
rate U ( i . e . the upper limit of the number of questions
performed per time unit ) and the duration of the
f iltration T ( i . e . of the restriction ) . When machine 2
receives such a request, it begins to f ilter ( restrict )
the questions traf f ic so that the questions rate will
be at most U, so that part of the questions will fail
( they will not even reach machine l ) . Machine 2
continues this restriction operation for the period of
time T indicated in the restriction request. If machine
2 receives a new request during this period, the upper
CA 02203907 1997-04-28
WO 96/15634 1 _ I/r_ -~ 7
llmit of the questions rate and the interval will be
updated to correspond to the new values. Instead of the
upper limit of the questions rate, the paL ~ ~er U may
also indicate the p~iue~lLaye: of all service request
g~- machine 2 should transmit to machine 1. For the
sake of clarity, only the former meaning (upper limit
of the questions rate) will be used hereinafter for the
paL ~er U.
When machine 2 uses the above-described
overload yl~:vt~ ion ~h:ln~ pm~ it has two problems.
The f irst problem is how to select the
afo.~ ~ioned paL L~L:. U and T. A long filtration
time T and a low value of the pai L~i U ~lim;niqh the
overload, but they also entail a clearly lower revenue
for machine 1. On the other hand, a short filtration
time and a higher value for the paL ~:i U do not
n"~ . lly reduce the number of questions sufficiently
for the overloading situation to be cleared up, and an
overloading situation also means a lower revenue.
A simple way of eliminating this problem is to
divide the response characteristic into consecutive
overload regions Ln (n=0,1,2.. ) according to Figure 3,
each of the regions having its own values f or the
pa, L~LS U and T. If, at all times, machine 1 is able
to determine its own load level, then the restriction
pdL ~t:,s can be stored in the machine in a format (Ln:
T, U), so that the machine can retrieve the required
values of the paL L~L~ T and U on the basis of the
load level Ln. However, this does not quite eliminate
the aroL~ Lioned problem, but shifts the trouble of
selecting the parameters to the u~eiaLoi. There are also
methods by means of which the paL_ L~LS can be selected
automatically, based on the utilization ratio of the
machine .
CA 02203907 1997-04-28
wo96/15634 1~I/r~ 7
The other problem relates to when to send and
when not to send restr$ction ~eyue~L~i. Machine 1 should
transmlt the first restrict$on request when it is close
to ~ -~ ' ng overloaded . It should then send a
restriction request elther when the restriction period
T expires (if the overload condition is still on) or
when the restriction paL teL,, change. Machine 1 should
not transmit new restriction requests if machine 2
restricts the questions correctly ( with the right
threshold value for the questions rate and the right
filtration time T). However, since there is no feedback,
machine 1 cannot know if and how machine 2 restricts the
questions. If machine 2 is the only source of questions,
then machine 1 can solve the problem by monitoring the
questions rate and by transmitting a new restriction
request when the rate of the; n~ ; ng questions exceeds
the allowed threshold value U. If there are several
r-~h;nl~q transmitting questions, then efficient book-
keeping is required to monitor the traf f ic and this
makes the aLL~Il9~ t 1 ;cated.
This second problem is thus of the
syllc~ ization type, since machine 1 must keep up to
date ( i . e . synchronize ) the restriction entity that is
in a remote machine according to the loading situation
of machine 1 at each moment.
The overload pLevt~lLion in the Intelligent
Network operates in a manner that is very similar to the
above-described example. The Int-~l l; g~nt Network
architecture is based on service switching points ( SSP )
and service control points ( SCP ) that make ~ c; 5; ~nc
concerning for example the routing and the charging of
the calls. The Service Control Points, which are
typically clearly fewer in number than the SSPs, contain
knowledge of what the different services do and how to
35 access the data that the services need. In an
CA 02203907 1997-04-28
wo 96/15634 1 _llrl ~t7
Int~l 1 ;g~nt Network, a Service Control Point is like the
machine 1 of the above-described example, containing a
database, and the SSP ls like the machine 2 that asks
questions. The above-described r~yll1ll o,~ization is also
a problem in the Int~l 1 ;g~nt Network since the
~ 1cation protocol between the nodes is not reliable
in this respect.
The above-described example uc,-~c-3L"ed a network
that was as simple as poss;hl~ with respect to its
topology. For example an Int~l 1 ;g-~nt Network is a
network with a (typically) star topology. A star network
bAC;-~lly comprises two kinds of nodes: central nodes
and peripheral nodes. Peripheral nodes generate traffic
that f lows towards the central node . When the
Int-~l l; g~nt Network comprises more than one SCP, the
architecture corresponds to several superimposed star
networks sharing peripheral nodes. Figures 4a to 4c
illustrate the above-described alternatives with
reference CN (in the IN: SCP) denoting a central node
and reference PN (in the IN: SSP) denoting a peripheral
node. Figure 4a shows a star network having one central
node CN and three peripheral nodes PN. Figure 4b shows
a star network in its simplest form corresponding to the
example of Figure 1 (one central node and one peripheral
node ), and Figure 4c shows two star networks sharing
peripheral nodes PN.
In addition to the Intelligent Network, many
other networks have a star topology. Examples of such
networks include a network formed of a satellite and
earth stations, wherein the satellite switches traffic
ye:"eL~ed by the earth stations, or a network consisting
of a base station controller and base stations of a
cellular network.
In some known ( intelligent ) networks, the
above-described ~y~ ul-ization is ; r- l ~ ted with a
CA 02203907 1997-04-28
WO96/15634 I~l/rl ~6~17
I,.~adc~a~ing method, wherein the central node transmits
a common restriction request to all the peripheral nodes
c~..~e~:~,ed thereto wl-en~v~r its load level changes (or
the restriction parameters change for some other reason,
for example when the Uy~ changes them ) and the
peripheral nodes respond to each restriction request
they receive with an acknowle~ . The central node
keeps a record of the acknow~ t ~sag~c, and if
some nodes have not transmitted an acknowledy~ t
message within a certain control period, the central
node retransmits the restriction request to these nodes.
The broadcast is repeated again to all the nodes as the
restriction period (T) expires if the overload condition
ls still on. ~owever, it is difficult to 1 l~ t, such
a method in a typical network comprising several nodes,
and in addition the method is not reliable, since a
peripheral node may be damaged for example immediately
after it has transmitted the acknowl .,, ~ message,
in which case the central node will not be informed of
the situation. Another drawback of such a method is that
the central node also transmits a restriction request
in vain to nodes which cause an insignif icant loading
on the central node ( this could only be avoided by
monitoring separately the traf f ic f rom each peripheral
node, which, however, is a, , l icated and therefore
undesirable solution ) .
The latter problem has been solved in some
known Intelligent Networks in such a way that as the
restriction parameters change, a restriction request is
always transmitted in response to a service request
message (which may be for example the guestion of the
example provided in the beginning ) sent by a peripheral
node. Therefore, the peripheral nodes with light traffic
will corrPsp~n~lin~ly receive fewer restriction requests.
The drawback of this method is, however, that it causes
CA 02203907 1997-04-28
WO 96/15634 r~l/rL s/~ ~,17
a great deal of traffic over the si~n~ n~ link between
the central node and a peripheral node. It also causes
a large number of updates in the peripheral node.
Summary of the Invention
The purpose of the present invention is to
provide a new kind of a~ang~ t by means of which the
above-described drawbacks can be eliminated and
.~y..._l..u--ization can be; ~ ed in a simple and
suf f iciently reliable manner ( i . e . that the loading
machine operates as well as possible in accordance with
the current load level of the loaded machine ) . That
purpose is achieved with the method according to the
invention that is characterized in that the service node
adds to the restriction request it transmits information
identifying the group of parameters, on the basis of
which the node should perform restriction, the node
copies said information into its memory and transmits
the data identifying the groups of parameters it uses
back to the service node, and the service node compares
said data with the data it has transmltt~ nd decides
on the basis of the comparison ~ th~r . r~striction
request is transmitted to the nod~ l~'''.IC~; ' r i~r~srrltted the
data. The arrangement accordin~ t~, t~ riv~ntion is in
turn characterized in that the n~t~ori~ service node
comprises comparing means for comparing the data stored
in the node and identifying the group of parameters to
corr~p~nfl;n~ data received from another network node,
the comparing means being operationally connected to
control the tr~nrm;scion of the restriction request from
the service node.
The idea of the invention is to maintain, in
a node providing services, information identifying the
restriction parameters currently utilized and to
transmit this data together with the restriction request
CA 02203907 1997-04-28
WO96/15634 I~l/r 5 ~-7
f irst to another node to be stored there, and to
transmit then the data identifying the restriction
pa. . Ler s used by this other node back to the node
providing services, so that the latter can cr~n~lude on
the basis of the received data whether the other node
restricts the traffic and if it restricts it in the
correct manner.
In an individual overloading situation, the
f irst restriction request that contains the data
identifying the new restriction pdL Lérs can be
transmitted f rom the node providing services either
a.lt Lically ( without an ; n~ ' n~ service request ) or
in response to an i n~ ng service request. Furth~ e,
the restriction request may either be an individual
message or it may be contained in a message that would
in any case be provided in response to the service
request message.
Due to the dL L ~ t, according to the
invention, ~y.,~ Lv.,ization can be implemented very
reliably in a simple manner.
Brief Description of the Drawings
The invention and its preferred embodiments
will be described in greater detail below with reference
to the examples according to the A~"- ,- ying drawings,
in which
Figure 1 illustrates the questions traffic
between two r-~h; n~c,
Figure 2a illustrates a response of a
hypothetical machine to service requests,
Figure 2b illustrates a response of an
omnipotent machine to service requests,
Figure 2c illustrates a response of a real
machine to service requests,
CA 02203907 1997-04-28
WO 96/15634 . ~, I/rL ;/~ 17
Figure 3 ill~ l,Lc~ s the division into
different load levels, pe:~L~- ~t' in a node,
Figure 4a shows a star network comprislng four
nodes,
Figure 4b shows a star network in its simplest
f orm,
Figure 4c shows a star network comprising two
superimposed star ne~wuL}.X sharing peripheral nodes,
Figure 5 shows an int~l l jgPnt network
comprising two central nodes and three peripheral nodes,
Figure 6 illustrates the ;c~tion between
the nodes in an intelligent network,
Figure 7a illustrates an intelligent network
and the formation of its central node from functional
blocks situated at different hierarchical levels,
Figure 7b shows the division of one block shown
in Figure 7a with respect to the call restriction
f unction,
Figure 8 illustrates the operation of a known
traffic restriction method,
Figures 9 and lO illustrate the operation of
an intel l; g~nt network in a loading situation,
Figure 11 illustrates the operation according
to the invention in a service control point ( central
node of the network) of the intelligent network,
Figure 12a is a flow chart illustrating the
tr~ncm; sc; on of a restriction request from a service
control point,
Figure 12b shows schematically the means of the
network central node performing the operation according
to the invention,
Figure 13 illustrates the addition of the stamp
according to the invention to a restriction request
message to be transmitted f rom a service control point
in the intelligent network,
CA 02203907 1997-04-28
W096/15634 ._I/r~ ~17
11
Figure 14 i~ aLes the addition of the stamp
array according to the invention to the initial message
transmitted from an peripheral node in the intel 1 ;g~nt
network, and
Figure 15 is a flow chart illustrating the
trAnPm;ss;~n of a restriction request from a service
control point, according to a preferred ' _'; t of
the invention.
Detailed Description of the Invention
In the following, the invention will be
described in greater detail by using as an example a
( star ) Intelligent Network wherein calls are
transmitted. As described above, the architecture of the
Intelligent Network is based on service switching points
( SSP ) and service control points ( SCP ) . These nodes are
inLt:~,ol.ne.;Led by means of a network SN according to the
p j ~nAl l; nt7 system number 7 ( SS7; described in greater
detail in the CCITT Blue Book Specifications of
~ nA77in~ System No. 7, ~lhournp 1988), in the manner
shown in Figure 5 . In mutual In; oation the SSP and
the SCP utilize the Intelligent Network application
protocol ( INAP ) described in the ETSI ( European
Tel~ ~; cations Standard Institute ) standard ETSI IN
CSl INAP Part 1: Protocol Specification, Draft prETS 300
374-1, r:o~ ' 1993. In the SS7 protocol stack
illustrated in Figure 6, the INAP is the upmost layer
situated on top of the Transaction rArAh; l; ties
Application Part (TCAP), the Si-7nAl 1 ;n-7 Connection
Control Part (SCCP) and the Message Transfer Part (MTP).
The SSP is generally a commercial tPl prh~n~ exchange
with a modified call control software, and the SCP
comprises the service control logic and has access to
the services database . Call traf f ic passes through the
SSPs. The Service Control Points make some of the
CA 02203907 1997-04-28
W096/15634 r~l/rl ~17
12
~pr;c;cmq rrnrPrn;ng the routing and the charging of the
calls . During a call in the IntPl l; gPnt Network, there
may be one or more INAP dialogues between the SSP and
the SCP . Each of these rl; A 1 ogl~c begins with a
~.ede~ ;nPd message (initial detection point message)
hereinafter referred to as an initial message.
When the network traffic is heavy, the SCP may
become overloaded. In order to prevent this, the
IntPl l; ~Pnt Network has a dece~ lized load control
system that uses a so-called call gapping method to
restrict ~ 9Pg arriving towards the SCP ( the term
"call gapping" is used in several international
standards, for example in the CCITT Blue Book,
Re ' -tion E . 412, 3 .1.1. 2 and RP~ ~Ation Q . 542,
5.4.4.3). The call gapping method is a known traffic
control method that is based on the frequency of call
oc.;ul~ence (rate of arrival), in which method the number
of calls is limited in such a way that at most a certain
maximum number of calls per time unit are allowed to
pass. In addition to the aforementioned standards, such
a method is also described for example in US Patent
4,224,479. The SCP monitors the loading situation and
the SSPs restrict the traffic, if necessary, by
rejecting some of the calls before the related dialogue
is started.
Assume that the network comprises, in the
manner shown in Figure 7a, two nodes SSP1 and SSP2, and
one SCP. The SCP can be considered to contain a
hierarchy of functional blocks A to E. Each block can
be considered to comprise, according to Figure 7b, a
gapping gate 70 operating according to the call gapping
method, and a subsystem SS located behind the gapping
gate. All tel~ ; ration with the subsystem passes
through the gapping gate, and the gapping gate gathers
35 statistics about the traffic, the condition of the
CA 02203907 1997-04-28
W0 96/15634 ~ /r~ 17
13
aub~yal,~..., and the condition of the other parts of the
SCP. From this data the gapping gate calculates the load
level of the subsystem in question.
The normal load level of the subsystem is L0
(cf. Figure 3). When the load level changes from L0 to
Ll, the gapping gate will try to limit the traf f ic by
sending a call gapping request to both SSPs. Such a
request typically comprises the following groups of
paL l.eLa ( 1 ) gap criteria, ( 2 ) gap indicators, and
(3) gap treatment. The gap criteria identify the portion
of the traffic that is the object of the call gapping
operation, for example, only calls starting with 800 can
be limited. Gap indicators define the maximum number of
initial messages ( calls ) U allowed in a time unit ( in
fact the gap indicators define the shortest allowable
interval I = l/U between two successive initial
_ -, which, in principle, amounts to the same
thing ) and the duration of the restriction T, wlleL~:u~u
the rate of initial ~S8~C between the arrival of the
call gapping request and the end of the duration can be
at most the aforementioned maxlmum. Shc operation of
this call gapping method is lllustr~s~ ' ln Figure 8.
When the traffic rate (shown o:: t~ ..ontal axis)
offered by the network is less tt'l~l.. t~ orementioned
maximum U, there is no call gappln-,. Wncn the offered
traffic rate exceeds this value, the SSP rejects some
of the calls so that the rate of the transmitted traffic
( shown on the vertical axis ) will be U. An ideal case
is described by a broken line, and a real situation by
a continuous line. In practice, the characteristic is
a continuous approximation of the piecewise linear
characteristic of the ideal case. This is due to the
fact that the offered traffic is not divided evenly on
the time axis.
CA 02203907 l997-04-28
wo 96/15634 . ~l/rl c ~17
14
The gap treatment parameters determine what to
do with rejected calls. For example, the speech channel
of a rejected call can be connected to a voice
Annollnl t or to a busy tone. In addition, the call
gapping request contains a control field which indicates
whether the call gapping request comes from an automatic
overload pl~:v.~"~ion ---h~nl ~m or from an SCP u~eLcl~oL .
The above-described groups of parameters are .1; qcl osed
in the afoL~ ~ioned standard ETSI IN CS1 INAP Part 1:
lû Protocol Specification, Draft prETS 3ûO 374-1, ~-J~. '
1993, Item 7 . 3 . 6, which is referred to for a more
detailed description.
When a call gapping request arrives at an SSP,
the SSP creates, based on the information it has
received, an image of the sending gapping gate (i.e. the
subsystem controlled by the gapping gate ) . This is
illustrated in Figure 9, wherein the overloaded block
( C ) is denoted by hatching and the call gapping request
transmitted by the SCP by the reference CG. By means of
the gap criteria and this image, the SSP identifies the
traffic that is directed to the overloaded subsystem and
restricts the rate of this traffic. When the period of
time indicated in the call gapping request expires, the
SSP destroys the image of the subsystem from its memory.
The gapping gate in the SCP is "static", i.e.
it exists all the time. The image of the gapping gate
(or the corr~oCpnnfl;n~ subsystem) in the SSP is
~Ly; the SSP creates the image when it receives
a call gapping request and destroys it when the duration
T specif ied in the call gapping request has expired .
When the SSP receives a call gapping request containing
the same gap criteria as an already existing image, the
other paL ters of that image will be updated to
correspond to the new ones.
CA 02203907 1997-04-28
WO96/1S634 ~I/r ~'7
Another approach is to view the images (copies)
in the SSP as objects with two states: active and
passive. When an image receives a call gapping request,
it turns active and starts to restrict traffic. When the
image is in the active state it can receive several call
gapping L~lueOI,s from the SCP. When the duration
cre,-i f; Pd in the last call gapping request expires, the
image turns passive again.
When two of the sub~yO ~_...s in the SCP are
simultAnPoucly overloaded, there is corrpsprn~;n~ly an
image ( copy ) of each gate in the SSP . As more and more
subOy~l become overloaded, the logical ::~LLL~ ,UL~ of
the images in the SSP starts to resemble the hierarchy
of the gapping gates in the SCP. This process is
illustrated in Figure 10.
The aforementioned ETSI standard ( Item
7.3.19.1.1) also defines a special "c811 gap
t:.,cuul-L~Led" indicator, which the SSP adds to the
initial message if the call has passed through the
gapping gate. This indicator thus informs the SCP that
the concerned SSP peL~uLl~o call gapping. However, the
SCP cannot be certain that the SSP performs the call
gapping with the correct paL lels, wherefore the SCP
cannot trust this indicator in making ~Pr; ~ nC about
whether to send a call gapping request or not. An
example of this is a network which comprises one SCP and
several SSPs, and in which network one of the SCP
subsystems is on the load level Ll having a
corrPcpr~ntl;ng upper limit U, to be indicated to the SSP,
of e.g. 10 initial messages (10 calls) per second. If
the load level now changes from Ll to L2, having a
corresponding upper limit of e . g . 5 initial messages ( 5
calls ) per second, the SCP transmits a call gapping
request CG containing a new upper limit to each SSP. In
this situation, if the data of some SSPs will not be
CA 02203907 1997-04-28
WO96/15634 .~l/r~ ~'7
16
updated, for example due to faults, then these SSPs
continue restricting the traf f ic with the old ( higher )
value of U until the duration indicated in the call
gapping request expires. Due to this, the ~;u..~el--ed
subsystem may move f urther to the next load level L3 .
The SCP cannot distinguish between updated and non-
updated SSPs, since the same indicator is received from
all SSPs.
There have been efforts to solve the problem
in the manner described in the beginning, so that the
same call gapping request is repeated after each initial
message arriving from the SSP. However, this alL~ng~ t
produces (a) more traffic over the si~n~l1;n~ link
between the SCP and the SSP, and (b) repeated updating
of the information ( subsystem images ), concerning the
SCP, in the SSP.
According to the present invention, the
operation ~l u.:~ds in such a way that the central node
SCP maintains in its memory, by means of e.g. a counter,
an integer which is called in this connection a "global
stamp". Initially the value of the stamp is zero. When
the load level of any subsystem of the central node
changes ( i . e . the restriction parameters change ), the
corresponding gapping gate increments the value of the
global stamp by one, reads the value of the new stamp
and stores it in its internal data. The stored stamp is
called in this connection a "local stamp". The value of
the local stamp identifies, at any moment of time, the
restriction paL - tei 5 used by that gapping gate ( and
the cu~ -led subsystem ) . The principle is illustrated
in Figure ll, wherein the counter is denoted by
reference numeral llO and the local stamp is shown only
for the subsystem E, for the sake of clarity.
When a gapping gate transmits a call gapping
request, it adds this local stamp to the call gapping
-
CA 02203907 l997-04-28
wo 96/l5634 ~-/rL r ~..7
17
request message CG transmitted to a peripheral node.
When the peripheral node SSP receives this call gapping
request message, it stores the value of the stamp in the
message to the data of the corresponding image ( copy ) .
After this, if a call from a peripheral node passes
through the call gap control, the initial message to be
transmitted contains this stamp identifying the
restriction pai, L~ x used by the peripheral node .
In the central node, the gapping gate decides,
in the manner shown in Figure 12a, on the tr~nqmiqq;nn
of a call gapping request. A received initial message
is first ~ ;n~ to see if it contains a stamp (phase
112). If the received initial message contains no stamp,
it means that the peripheral node SSP does not have the
image of the central node gapping gate (i.e. the
peripheral node does not restrict the calls ) . A call
gapping request is then transmitted (phase 114). If the
received initial message contains a stamp, its value RS
is either equal to or different from the value of the
local stamp LS. If the examination performed in phase
113 shows that the values are equal, then the image of
the gapping gate is correctly updated and no call
gapping request will be transmitted. If the values are
different, a call gapping request will be transmitted
to the ~_u,~ ,-ed peripheral node ( phase 114 ) .
The tr~nc~m;Rq;on of the call gapping request
can also be described in the following manner:
transmit a call gapping request if there is no
stamp in the initial message or if the stamp of the
service request message r local stamp.
At least one pair of comparing means 122 have
thus been added to the central node, as shown in Figure
12b, the comparing means comparing the stamp received
from the receiver means 121 with the local stamp LS, and
controlling the node transmitting means 123 to transmit
CA 02203907 1997-04-28
WO 96/15634 ~ J~ 7
18
a call gapping request CG according to the result of the
comparison .
The afo~ . ~ioned ETSI standard ( Item
7.3.6.11) defines four different gap criteria, one of
which is the called address. As it was already described
above, the SCP can be seen as a hierarchial DLlul_~uLe
of SUbDyDt , wherein the traffic of each aubDyDL~... is
controlled by a gapping gate. Such a subsystem is the
part of the SCP having its own unique gap criteria. This
also means that the less precisely the gap criteria are
det~rm;n~d, the greater the portion of the traffic they
relate to, and the higher the position of the
corresponA;n~ DubDyDL~m in the hierarchy. For example
a subsystem having the criteria "called address starts
lS with number 3" is higher up in the hierarchy than a
subsystem having the criteria "called address starts
with numbers 314". When the SSP contains at least two
active images ( copies ) which are located one after
another in the above-described hierarchy, for example
subsystems A, 8 and C in Figure 11, it must be decided
how to restrict calls to these subsystems. The SSP can
contain for example two images in such a way that in the
first image the criterion is "called address starts with
numbers 800" and the maximum rate U is 10 initial
55~ C/S, and in the second image the criterion is
"called address starts with numbers 80012" and the
maximum rate U is 5 initial messages/s. When a
subscriber dials for example the number "800123", it is
possible in principle to filter that call according to
the ~ L L~l~s of either image. However, using the
pal ~eLs of the first image is incorrect, since then
the gate which created the image located at a lower
level in the hierarchy may receive as many as 10 initial
messages/s, even though only 5 initial messages/s are
allowed. Neither is it right to use the pc~ lD of
CA 02203907 1997-04-28
wo 96/lSG34 r - l/rJ -17
19
the second image only, whereas using the pal tt:L~ of
both images in the call gapping provides the correct
result in all cases. When the call data matches the gap
criteria of several active images, the call must pass
through the call gap control of all these images. If the
call cannot pass through all the lmages (gates), it will
be re j ected .
The principle described in the previous
E~a~ayLaPh has the following effect on the present
invention. The initial message must contain an array of
stamps: that array contains the stamps ( stamp values )
of all the active images the call has passed. The array
may thereby be empty ( no stamps ) or there may be one or
several stamps. The gapping gate of the subsystem of the
central node SCP P ;nPs whether the array contains the
value of its own local stamp. If the value is absent,
the gate transmits a call gapping request to the SSP.
According to the invention, additional data,
a stamp or a stamp value identifying the restriction
parameters used in the SCP is added to the data f ield
concerning call gapping in the message to be transmitted
from the SCP. The data field, referred to in the
standards as "CallGapArgument", to be transmitted in the
call gapping request message is shown in Figure 13. For
the sake of clarity, the figure shows the fields with
the same names they are ~licrlosPcl with in the standard.
Gap criteria, gap indicators, control type, gap
treatment and extension are in succession within the
data field. Control type and gap treatment are optional.
The ext~ncionc form a free space that the manufacturers
can utilize in several ways. According to the invention,
the stamp is either added between some of the
aL~,L ~ioned f ields as a ~paL ate f ield, or it may be
placed inside the extPnc;~nc. The advantage of the
CA 02203907 1997-04-28
WO 96/15634 ~ /rL _ ~17
latter alternative i8 that such a call gapping request
message ,~ q fully to the present standard.
It is also roeq1 hl e that the stamps of two or
more simult~n~o~c1y overloaded sub~y~ . are placed in
the same call gapping request message. The message then
contains two or more CallGapArgument data fields and
corr~sp9n(l1 ngly at least two stamps .
The a~u.r Lloned array of stamps is added to
the initial message transmitted from the SSP. Figure 14
shows the data field of the initial message (i~Le--,2d
to in the x Lc...da ~ ~ as the InitialDPArg ) . The data f ield
of the initial request message contains the service key,
called party number, calling party's number, calling
party category, the aLo-~ tioned call gap -en.iL,u--Lc:-~d
indicator and the extenei nne . According to the
invention, the stamp array (containing one or several
stamps if the SSP performs call gapping, or no stamps
if the SSP does not perform call gapping ) is added
either between some f ields as a separate f ield or the
stamp array is placed within the ext~neinne. The
advanLag~ of the latter is again that this kind of
initial message complies fully with the present
standard .
Since the size of the global stamp is in
practice limited, it cannot be incremented forever. If
the size of the variable is, for example, one octet ( 8
bits ), its maximum value is 2a-l = 255 . When a gapping
gate tries to increment a stamp that is at its maximum,
the value will not increase, but it will jump back to
~0 zero ( roll-over of the counter ) .
After this, it is possible that two or more
gapping gates have the same local stamp value. The
frequency of occurrence of such an event depends on the
relative speed with which the gates update their
restriction pai ~ L~L ~ and on the maximum value of the
CA 02203907 1997-04-28
Wo 96/15634 1 ~l/r_
21
counter. The problem resulting from the roll-over of the
counter can be completely ignored, however, DCpeC;Al ly
if it is certain that it does not occur too often. If
it is not certain, the problem can be avoided in such
a way that the roll-over provides a new value for the
stamp of each gate, even if the restriction paL ~t:LS
of the gate have not changed. The call gapping ~e~lu~Ls
transmitted after this provide new stamp values for the
images of gates in the peripheral nodes.
Instead of a counter to be in-,L, ~d, the
global stamp can naturally also be maintained with a
counter that ls de~L, ted. Instead of an integer
maintained by the counter, the global stamp may in
principle consist of any data that uniquely identifies
the restriction par, -~ ~C:L~, for example of the current
value of the central node clock.
While a call gapping request is on its way to
the central node, the peripheral node may transmit some
initial ~ ~ca~De with the old stamp value. Each of these
initial messages brings about the tr~n~m;Sc;on of a call
gapping request with a new stamp value from the central
node. Thus due to the network delays, the call gapping
request might have to be repeated several times.
According to a preferred ' ~ of the
invention, the above-described basic principle can be
modified in such a way that the central node only
~:~y:~m; n~:~C a predetermined PLUIJOL ~ion of the total number
of initial messages it has received. In the following,
this predet~rm; n~ O~)UL ~ion will be described with p
( 0 < p S 1 ) . For a source ( peripheral node ) transmitting
at a rate ~Yr~ ; ng the threshold U ( initial messages/
second ) this guarantees a call gapping request on
average within at least 1/ ( pU ) seconds . The advantage
of this ~ ' ~';- t is that in the central node the
amount of work due to comparison is reduced by l/(l-p).
CA 02203907 1997-04-28
wo96/15634 1_I/rL ~W~7
22
This kind of operation is preferably
1~ ~ed in such a way that for each ~- ~n~ initial
message, a random number R, for which O s R 5 1 holds,
is yeneL~ d in the central node SCP. If this number is
smaller than the parameter p, the array of stamps is
e nPd . The dPri c; ~n-making can thus be described in
the following manner:
examine the received stamp array, if R < p,
where O S R S 1, and O < p S 1.
The operation is also illustrated in the flow
chart of Figure 15, which corresponds to the basic
principle shown in Figure 12, except that in this case
the comparison of the random number R with the ~.'L-. tt:i
p (phase lllb) is peLL~ before the examination of
the stamp (phase 112). If R < p, the operation continues
in the above-described manner, but if R 2 p, then the
array of stamps is not I nPd at all (i.e. the
operation proceeds to phase 115 ) .
The invention is described above with reference
to the Intelligent Network. This means that the node
providing services comprises several subsystems in the
above-described manner. However, the method according
to the invention is applicable in any network with a
similar basic situation. This means that there might be
only one subsystem in the node providing services,
wl-eLI~ ol- it is sufficient to transmit the value of a
maximum of one stamp in the - ~s~yPc .
Even though the invention is described above
with reference to the examples according to the
~ nying drawings, it is clear that the invention
is not restricted thereto, but it may be modified within
the scope of the inventive idea disclosed above and in
the appended claims. The method according to the
invention can be applied for example in a certain part
of the traffic. The number of the paL tt:L::> used in the
CA 02203907 l997-04-28
W096/15634 I~,l/r
23
call gapping operation may also vary, wherefore a group
of pa~ s must be u--de~ :, L~,od as a group contalning
one or several pa- ~e.~. Since the a~
according to the invention can in principle be applied
in any t~-1 f ; cations network having the basic
situation according to Figure l, the central node is
~:re:L ~ c:d to in the ~ cl claims as a service node
( not restricted to an Int~ g~nt Network ) and the
peripheral node in turn as a node ( not restricted to a
star network ) . The service requests must also be
u"dei ~ ~ood to relate generally to any services the
peL Lo~ ance of which loads the service node .
