Note: Descriptions are shown in the official language in which they were submitted.
:1159~3Z
1 ~ACKGROUND OF THE INVENTION
~ . _
The invention relates to time division multiplexed
switching systems.
~ ime division multiplexed switching systems, includ-
ing those which operate under the control of digital computers,
are very well-known and widely used. For a discussion of
time division multiplexed systems, see U.S. Patent Nos.
.
3,115,552; 3,172,956; and 3,~01,235.
~RIEF DESCRIPTION OF THE DRAWINGS
1 0 ~
E'IGURE la illustrates a prior art direct link
interconnection between switching modules.
FIGURE lb illustrates a prior art ring connection
between switching modules.
FIGURE 2 is a block diagram illustrating one prior ~ -
art hierarchical interconnecting system.
FIGURE 3 is a block diagram illustrating another
prior art hierarchical interconnecting sys-tem.
` FIGURE ~ is a block diagram illustrating a prior
art interconnecting system employing a distributed architecture.
FIGURE 5 is a block diagram illustrating the inter-
connections employed in the present invention for inter-
connecting modules and their respective intertie units.
E'IGURE 6 is a block diagram of a switching module
(speciEicallyla computer-controlled PBX) and its interconnection
with an intertie uni-t.
FIGURE 7 is a general block diagram of an intertie
unit employing an intertie TDM bus.
FIGURE ~ is a block diagram of an intertie unit
employing multiplexers,
lEi~ .
llS~l32
1 FIGURE 9 is a block diagram illustrating the
carousel interconnections of EIGURE 5 where four switching
modules and their respective intertie units are employed.
In some cases/ time division multiplexing of
telephone signals is incorporated into switching devices
such as private branch exchanges ~PBXs). These devices
interconnect with all classes of trunk lines, station lines
and other lines to switch between lines and to concentrate
signals on these lines to trunk lines connected to the public
network. This permits efficient utilizatlon of these public
network access trunk lines. Numerous commercial systems are
available, includirlg those which employ redundant computers
to provide high reliability in such telephone switchesO One
such time division multiplexed
;~ ` .
~'
;
-la-
:~IL1541.~Z
swi-tching device shall be described in conjunction with FIGURE
6. Typically, these devices include, as will be described,
an interface unit (comprising answering supervision, analog-
to-digital and digital-to-analog conversion), concentration
unit, and switching means. Hereinafter for purposes of this
application, these types of devices and related devices shall
be referred to as "modules" or "switching modules".
To expand -the capacity of these modules, particularly PBXs,
switching and concentrating units are interconnected most often
in a multi-stage hierarchical architecture. Two such prior
art interconnecting schemes are discussed in conjunction with
FIGURES 2 and 3. This hierarchical archi-tecture is not always i ~ -
efficient. The single unit used in these systems to provide
; higher order switching mus-t be large enough to handle
some maxlmum number of modules. This higher order switching
is not cost-effective when used with less than this maximum
number of modules. The corollay to this is that these systems
are not readily expandable beyond the maximum number of modules.
Graded multiple architecture has been used to inter-
connect lines using modular step-by-step switches in central
offices. In such architecture, it is well-recognized that
direct links provide efficient coupling for deterministic
first offered traffic. Telephony traffic which overflows these
direct links represents more random, fluctuating traEfic. This
--2--
.. . . . . .
1~5~3~
overflow traffic is more efficien-tly handled on shared links
(denoted here as a "bus"). Often a combination of a bus
for handling overflows along with direct lin~s are employed
in distributed architecture. One such prior art system
which has been in use many years is shown in FIGURE 4.
Another architecture used in data networks employs a
plurality of TD~I switches in a ring configuration. In this
configuration, calls may be routed around the ring through
other modules to a particular module. However, if time slots
are used in each of the intermediate modules through which the
call passes, this results in relatively inefficient use of
the TD~I switch capacity in each module. Furthermore, if the
ring is broken (by the failure of one module), then the remaining
operable modules in the ring cannot be used to provide a two-
` lS way connection. This architecture is shown to some extent in
FIGURE lb.
,
As will be seen, the present invention discloses an inter-
connecting architecture of the distributed type which combines
characteristics of both the ring and graded multiple architectures.
The described architecture is well-suited for interconnecting
a plurality of commercially available digital TDM switches
or other modules for use in telephony applications. Separate
digital intertie hardware is employed which permits calls to
be effectively routed through each of the modules withou-t
ilS4~3Z
requiring the use of a time slot from within the main TDM
bus of each module. Thus, the invention provides a dynamic
form of the graded multiple architecture.
'~
`: :
'
-4-
.
~S~32
SUM~RY OE T~IE INVENTION:
.
A distributed interconnecting system for use in ~ time
division telephone switching system which employs a plurality
of similar switching modules in described. Each of the modules
is connected to a plurality of communication lines such as
station lines, trunk lines and other lines. The modules,
in a well-known manner, concentrate the signals on these lines
into time division, digital signals in order that circuit paths
may be readily comple-ted between lines. Each of the modules
also includes intertie facilities containing an input and
output port for permitting the modules to be coupled to other
modules. The improved interconnecting system of the present
invention provides a plurality of dual ring interties connecting
the modules. Each dual ring system is comprised of rings
which rotate data in opposite directions. This dual ring
structure is referred to as a "carousel". Each carousel is
connected to a module by an intertie unit. These digital
intertie units have buffers connected to the intertie paths
of the carousel. In addition, each digital intertie unit has
buffers connected to the main TDrq bus of its respective module.
A first coupling rreans within each intertie unit provides
selective transmission between the intertie buffers without
requiring use of the main TDM bus of the module. The intertie
units also include second coupling means for selectively
coupling the intertie buffers to the buffers associated with
the TDM bus of its respective module. The invention operates
by partitioning the carousel intertie structure as necessary to
connect communication lines in different modules by using the
shortest partition available given a multiplicity of carousels
interconnecting modules.
--5--
.
' ' ' ;
LS~Z
1 DETAILED DESCRIPTION OF THE INVENTIO~:
A highly e~ficent and reliable interconnecting
system for interconnecting switching modules with a distributed
architecture is~described. In the following description,
well-known components and techniques are not set forth in
detail in order not to obscure the present invention in
unnecessary detail.
The present invention, in its presen~ly preferred
embodiment, is employed to interconnect a plurality of PBXs.
It will be apparent that the described invention may be
employed with switching modules other than PBXs. In the
presently preferred embodiment, a PBX commerically available
~rom Rolm Corporation of Santa Clara, California i& employed
and is brie~Iy described in conjunction with FIGURE 6. This
PBX is generally described in "The New Rolm VLCBX", Business
Communications Revlew, March-April, 1979, beginning on
page 38. This system comprises a time division multiplexed,
computer-controlled private branch exchange, known in the
trade as a computerized branch exchange (CBX~. Analog
slgnals appearing on trunk interface buffers and telephone
set interface buffers are converted into digital signals
with we:Ll-known pulse code modulated (PCM) digital
techniques. Other well-known means are used for error
-
~;
--6-- .
~.
1~5~ 2
reduction in these digital signals. See U.S. Patent Nos.
3,877,022 and 3,999,129. Using time division multiplexing,
the digital voice signals or other digital data are
then switched, permitting connections between the station
lines and trunk lines under the control of a central processing
unit. These PBXs thus include a plurality of ports permitting
coupling with station lines, trunk lines and other lines. Also,
these modules for the present invention include digital intertie
termination units to permit the PBX to be coupled to other PBXs.
Before describing the present invention, a brief discussion
of prlor art architecture is helpful to apprecia-te the advan-tages
of the present invention. Two intertie concepts, a direct
link and a bus or ring, are commonly employed. As shown in
~ FIGURE la, the direct link provides a direct connection between
15~ two modules. This connection is very cost-efficlent when heavily
used and is particularly useful where traffic between modules is
known and stable. However, with this type of connection, dynamic
.,
redistribution to overcome system imbalances among several switching
modules is not possible. rlith the bus or ring, such as the unidirec-
tional links connected in a loop shown in FIGURE lb, loadingof the lines is not particularly sensitive to the distribution
of the traffic between modules since connections can be made
between any two nodes. Unfortunately for this arrangement,
many links are re~uired for a single conversation and, moreover,
the failure of a single module can interrup-t communications
between all the modules.
B ~
:~S4~2
(As will be described in greater detail, the carousel-
like interconnections of the present invention, as shown
in FIGURE 5, comprises two rings rotating in opposi-te
directions. This permits dynamic partitioning of carousels.
That is, the carousel intertie emulates the behavior of both
the links and ring structure described above. ~ore importantly,
as will be described, the requirement of the ring of FIGURE lb
that calls occupy time slots in each of the modules through
which they pass, is eliminated. Intertie units permit calls
to be effectively routed through the modules without passing
through the TDM bus of the module.) -~
As mentioned in the prior art section, in order to increase
capacity, a plurality of PBXs are often coupled in a hierarchical
structure such as shown in FIGURE 2. In FIGURE 2, a plurality of
lines such as station lines and trunk lines l0 are coupled to a
multiplexer 11. The multiplexer selects linés in a well-known -
manner and couples these lines to a -time division multiplex (TD~1)
switch 14. Three multiplexers 11, 12 and 13 are coupled to this
single TDM switch 14. Similarly, the TDM switches 15 and 16
each include three multiplexers which in turn are coupled to a
plurality of communication lines. Interconnections between
mul-tiplexers coupled to the same TDM switch are made through that
TDM switch. For example, an interconnection between lines
coupled to the multiplexers 11 and 13 is made through -the TDM
switch 14. Obviously, some means must be provided for inter-
~, ~ ~~ '
.' ' :
:
~1~4~32
connecting lines associated with different TDM switches.
A hi~her order switch is required, for example, to inter-
connect a line connected to the multiplexer 11 with a line
connected to the multiplexer 18. A higher order time division
multiplex switch 17, which is coupled to each of the switches
14, 15 and 16, is used in this hierarchical architecture. It
is apparent that the switch 17 must have the capacity to handle
the maximum number of lower level switches which may be coupled
to it. Thus, if the switch 17 has the capacity to handle six
lower level TDM switches, half its capaci-ty is unused when only
three lower level swi-tches, such as shown in FIGURE 2, are used.
Similarly, if the switch 17 is slzed to handle three lower level
switches, the system may not readily be expanded to handle more
lower level TDM switches unless an even higher level TDM switch
~15 lS used. Thus, this system is not easily~expanded and, moreover,
unless the maximum capacity of the highest order switch is used,
costly hardware remains idle.
:
In FIGURE 3, another commercially employed hierarchical
architecture is shown. Again, a plurality of multiplexers, such
as multiplexer 21, is used to select a plurality of lines 20.
Three such multiplexers are coupled to a higher level multiplexer
22. The multiplexers 22, 23 and 24 are interconnected through
TDM switch 25. ~hus, for -this particular architecture, switching
does not occur within the multiplexers 22, 23 and 24 but rather
only in the TDM switch 25. To conneck a line from multiplexer
.
Je ,~, ,
.. ~
:lil5~132
21 to one associated with multiplexer 26 requires coupling through
multiplexer 22, switch 25 and multiplexer 24. This architecture
has the same shortcomings as the architecture of FIGURE 2.
That is, unless the TD~1 switch is used to its maximum design
capacity, the system is inefficient.
For many years, particularly in central offices, a distributed
architecture, referred to as "graded multiples", as opposed to
the hierarchical axchitecture oE FIGURES 2 and 3, has been
:
employed, such as shown in FIGURE 4. Three switching units 30,
~10 31 and 32 are shown~in FIGURE 4; each of these switching units
is coupled to a plurality of communication lines such as lines
'::
33. The switching units are interconnected by direct links;
for example, link 34 interconnects switching units 30 and 31
and, similarly, links 35 and 36~interconnect unit 32 with units
30 and 31, respectively. These direct links provide very
efficient hardware utilization,~particularly when the traffic
between each of the unlts is determinabIe. However, if only direct
links were used, fluctuations between units, as is the case with
typ~ical telephone communications, will~accur requiring the addition
~20 of a larg~ number of direct lin~s to handle such fluctuations.
To avoid the large number of direct links which would
be required when an overflow occurs on any of the direct links, ;~ .,
the bus 37 is employed. Obviously the bus 37 does not provide
as efficient coupling as the direct links, since it requires
ports in all of the switching units. For example, if the bus
` - ;', '
:
~
.. .. . . . ..
~5~3~
37 is handling an overflow between the units 30 and 32, the
: corresponding ports on the bus coupled to the switching unit
31 remain unused. However, the bus uses fewer ports than
would be required using direct links to achieve the same capacity
between any two modules. The overflow of telephony traffic
from direct links is statistically subject to bursts of
traffic. The effect in a graded ~ultiple design is to
interleave such overflow on the bus, providing higher utilization
of the bus.
The distributed architecture of FIGURE 4 may be more readily
expanded than the hierarchical structure of FIGURES 2 and 3.
The particular distributed architecture of FIGURE 4, however,
has certain technological disadvantages, particularly when
used to interconnect commercially available PBXs.
Referring now to FIGURE 6, this FIGURE illustrates a :
computer-controlled telephone switching system and is the
switching module employed in the presently preferred embodiment
of the invention. As may be seen, a central processing unit
57 under the control of a program stored in the memory 58 ¢ontrols
the switching interconnections between a plurality of trunk
llnes 47 and a plurality of lines 48 connected to telephone
sets via a digital network, specifically TD~q network control
means 56 and a module TD~I bus 50. Each trunk line 47 is coupled
through a trunk interface,unit 54 and a pair OL multiconductor
r~
~ -.;,3-
3~
~ ., ~
paths 43 and 44 to a coder 49 and a decoder 51. Similarly,
each of the lines 48 is eoupled through the telephone set
interface 55 through multipath eondue-tors to the eoder 52
and deeoder 53.
~hile the present invention may be employed with a plurality
of different modules, the module of FIGUR~ 6 eomprises a time
dlvision multiplexed eomputer-controlled private branch exehange,
known in the trade as a computerized branch exchange (CBX).
In this module, analog signals appearing at trunk interface 54
and telephone set interfaee 55 are eonverted to digital signals
by eoders 49 and 52, respeetively. These signals are routed
ln digital form to an appropriate decoder-51 or 53 where the
digital signals are converted baek to analog form and coupled-
through the trunk interface or telephone set interfaee to an ~ ; -
appropriate one of the lines 47 or 48, all under the control of
the TD~l network eontrol means 56. The control means 56 is,
in turn, controlled by the CPU 57 in aceordance with the program
stored in the memory 58.
The module TDM bus 50 provides multiconduetor paths for
both digital information signals (e.g., voiee signals from
lines 48); multldigit address signals identifying
speeifie trunk lines; and timing and control signals for direeting
sequential operation of the interfaee units 54 and 55, eoders
49 and 52, and the deeoders 51 and 53.
~l15~ Z
he operation of the module, speci~ic~lly the ~B~: of
FIGURE 6, is well-known, as is its construction. For this
reason and the fact that a detailed discussion of -the module
is not necessary for an understanding of the present invention,
a detailed description of this module is not set forth
herein.
For purposes of the present invention, however, the module
TDrl bus 50 is interconnected to an intertie unit 60. This
intertie unit is used to interconnect the module with other
: 10 modules. The intertie unit 60 includes the input por-ts I1 and
I2 for coupling to output ports of other lntertie units and
the output ports l and 2 for coupling to the input ports
of other intertie units. In general, the in-tertie unit 60
~ includes first switching means for routing signals between the
; 15 lnput port Il and the output port 2~ and between the input port
I2~and the output port l. These lnterconnections are made
without coupling through the module TDM bus 50. The intertie
unit 60 includes second switching means which permit signals
to be routea from the intertie ports to the module TDrl bus.
. .
One embodiment of the intertie unit 60 of FIGUR~ 6 is
shown in FIGURE 7 and includes a time division multiplexed
switch. Thus the unit 60 includes a separa-te intertie unit
TD~1 bus 70 which is separate and apart from the module TDrl
bus 50. The intertie ports Il and I2 cor,lmunicate with the
..:
~4~3Z
bus 70 through the decoders 63 and 65, respectively. The output
ports l and 2 co~unicate with the bus 70 through the coders
6~ and 66, respectively. ~The module TD~1 bus 50 communicates
wit:h tlle intertie unit TD~I bus 70 through a TD~1 in-terface
switch 62. The unit of FIGURE 7 may be constructed employing
well~]~noun TDM components such as the components employed in
the PBX of FIGURE 6.
Through the bus 70, signals at the port Il may be directly
routed to the port 2 without requiring use of time slots on
the bus 50. Also, calls received at the port I2 may be routed
to the port l through bus 70, again without requiring time
slots on the bus 50. The ports Il, I2 and l~ 2 may
communicate with the module TD~1 bus 50 -through the switch 62
in a well-known manner. - -s.
Referring now to the presently preferred embodiment of
the-intertie unit as shown in FIGURE~8, the unit is again coupled
to a module TD~I bus 50. The intertie ports Il and I2 are
coupled to decoders 63 and 65, respectively, and the outpu-tports l and 2
are again coupled to the coders 64 and 66,respectively. The signals from
the decoder 63 are routed to a buffer 71. The buffer 71
communicates with the bus 50 and also has a path to the multi-
plexer 74. Signals from the bus 50 may be coupled to the
module TDrl buffer 72. This buffer provides another input
to the multiplexer 74. The multiplexer 74, which may be an
~ .
-~6- .~l
. " .
~15~ 3~
ordinary digital multiplexer, selects the contents from either
the buffers 71 or 72 and couples these selected signals to
the buffer 79. The buffer 79 is,in turn,coupled through the
coder 66 to the output port 2 The signals from the decoder
65 are coupled to the buffer 78 and from there may be coupled
either to the bus 50 or to the multiplexer 77. The multiplexer
77 also receives an input from the buffer 80 which
is coupled to receive signals from the bus 50. The
multiplexer 77 may select the contents of elther the buffers
78 or 80 and couples the selected signals through the buffer 73
and coder 64 to the output port l.
.:
In opexation, assume that a call is received at the port
Il. This call may be routed either to.the bus 50 (through buffer
71~ or to the output port 2 (through buffer 71, multiplexer
74 and buffer 79).: Si~ilarly, calls received at the input
port I2 may be routed either to the bus 50 or the output port
l. Signals from the bus 50 r,lay be coupled to.the output port
l through the buffer 80, multiplexer 77 and buffer 73. Similarly,
calls from the bus 50 may be routed to the output port 2
through the buffer 72, multiplexer 7~ and buffer 79. It should
be noted that the calls routed between the ports Il and 2
and between the ports I2 and l do not pass along the bus 50
and thus do not require time slots on this TDM bus. The
intertie unit of FIGURE 8 may be constructed from well-known
digital buffers,~multiplexers and coders and de.coders.
~ .
.~p, '
:.
~15~
1 ~he bu:E~ers and other lines of the above-described
intertie unit may have the capacity to handle a single
digital word, in the presently preEerred embodiment, each
buffer provides storage for eight digital words. The reason
for this is that, as currently implemented, time division
multiplexi.ng ~eight time slots) is employed on the inter-
connecting lines between the intertie units.
Referring again to FIGURE 5, three modules such
as the module of FIGURE 6 with their i.ntertie units such ~ -
as unit 60 of FIGURE 6 are shown as module and intertie
units 40, 41 and 42~ In the carousel interconnecting ..
arrangement used between the modules, the output port 2 oE
each of the mbdules is coupled to the input port Il of the
next moduJ.e with the output port ~2 of the last module
42 belng coupled to the input port I1 of the first module
40. .Similarly, the output ports l are coupled to the
input ports I2 of the next module with the output port l
of the first module being coupled to the input port I2
the last module. As lS appa~ent, two buses rotating in
opposite directions result.
.
. .
..
`:
.
-16-
:
B
~ , , ; .. ,
115a~ 3z
In general, in the presently preferred embodiment all the
CPUs 57 of FIGUR~ 6 are coupled to a common hub providing common
channel interface signaling. This is in contrast to a system
employing a master CPU which controls the CPUs in each of the
modules. Through use of the hub, race conditions are avoided.
The lines interconnecting each of the rnodules consists of a bus
for handling 16 bits. Twelve (12) of these bits are used for
the converted analog signals and four (4) bits are used to
indicate channel availability. The status of the path through
a switching module is determined by having an idle coder on
the originating end sent out of the originating modules ID
over the idle channel. Switching modules can then break into
idle links and sample the furthest idle module. Comparing this
to the internali~zed system architecture, either module can
teIl whether or not a connection extends far enough in either
direction. A module's own ID on the intertie connections
indicates it is completely ldle. After verification, a module
can seize a link through commands to the control hub. This
approach has the advantage of keeping status lnformation
d~namically updated. Furthermore, if in the seizure control
sequence, the terminating module were to request to break-in,
then path verification with a terminating module's ID can be
made. The seizing module would notify intermedia-te modules
that the channel was now inactive, and upon release -that it is -
inactive.
~ .
115~3%
A hierarchical fixed routiny table for searching for the
intertie connections between modules is employed. The lowest
entry in the table which represents most of the traffic flow
will be effectively direct links, for example Call l shown by
dotted line 45 between the modules 41 and 42 of FIGUR~ 5.
In addition, the status of each terminating channel in a given
module will be stored in the module containing the channel.
How far the channel activity extends a~ound the carousel can
be dyn~mically determined or verified at the time of the desired
seizure~as described. The hierarchical approach -to routing
assures the lowest paths to make a connection. I~hen a call
lower in the hierarchy disconnects, it provides a hole which -
is filled by shifting down calls located at higher levels.
The value of dynamic rearranging calls to fill such holes is
substantial.
- ' . -
For example, assume that all the direct links between the
modules 41 and 42 as shown by dotted line 45 are busy. Additional
calls are routed between the moduIes 42 and 41 through the
module 40 as indicated by the dotted lines 46. ~hen it is
determined that links indicated by the dotted line 45 are
available, calls are rerouted through this direct link and
removed from the longer routing. This dynamic rerouting frees
the direct links between the modules 40 and 42, and 40 and 41. ;~
It is significant to note that when the call is routed from
module 42 to 41 as indicated by the dotted lines 46, that is,
through the module 40, the TDM bus within module 40 is not
required.
,
1;~, . ~- :
1~
"
,
1::159L~L32
Thus the described intertie struc~ure provides standardi~ed
hardware for each module capable of crea-ting links, buses or
rings. In addition, when connected in -the carousel architecture
the carousel can be dynamically par-titioned to effectively
create direct linked paths, tandem linked paths, through inter-
tieing modules without usin~ time slots within -the modules.
Another important feature of the i,ntertie arrangement of
-the present invention is that is provides more reliable inter-
connections, particularly when compared to a standard ring
architecture. Assume that the module ~0 alony with its
intertie units are inoperative. Even in this case, direct links
exist between moduIes 41 and ~2. Note that in a typical uni-
direotional ring arrangement such as shown in FIGURE lb, if
any one of ,the modules becomes inoperative, all communications ;~
15' from module-to-module are lost. Even with the hierarchical -
architecture of FIGURES 2 and 3, malfunctions in the higher
order TDM switches can disrupt all inter-module communications.
The interconnection system of the present invention has
particular advantage when used with three switching modules
as shown in FIGURE 9. With three modules, direct links exist
between each of the modules. The same principle, however,
may be applied to any number,of switching modules. ,In FIGURE
9, switching modules and their intertie units are shown as
switching units. That is, for example, the switching module (PBX)
of FIGURE 6 along with the intertie unit 60 are shown as the
: '~ ' /9
.: . .
l~S~ Z
switching unit 92 in FIGURE 9. Four switching units 92, 93, 94
and 95 are lllustrated in this figu:re. The units are coupled
in the same carousel manner as the units of FIGURE 5. For
example, one input port of the unit 92 is coupled -to an input
port oE the unit 93 through line 99, and one output port of
the unit 93 is coupled to one input port of the unit 92 via line
96. The other output port of unit 92 is coupled to an ou-tput
port of the unit 95 via line 97. Units 93 and 94, and 94 and 95,
are coupled in a similar manner. `;
~ ' ~
While direct links exist between units 92 and 93, units
94 and 95, and units 95 and 92, direct lin]cs do not exist
between units 92 and 94, units 93 and 95, and units 92 and 94.
Some direct links between units are lost when more than three
units are used. I~owever, in these cases, two paths exist
through the intertie unlts associated with each of the units,
again without employing time slots within the modules. For
.
example, consider the possible paths between units 92 and 94.
A connection may be made through unit 93; another connection may
be made through unit 95.
While only the case of three and four switching modules
has been shown above, it will be obvious to one skilled in the
art that the principle of the present invention may be extended
to any greater number of modules. Also it wi].l be apparent
that the intertie units may be reali7ed in many different
configurations.
:,
.
1~5~2
Comparing for a moment the interconnecting systems of
FIGURES 5 and 9 with the prior art system of FIGURE 4, it should
be noted that, in effect, the lines between the modules and
intertie units 40, 41 and 42 serve the function of both direct
links and the function of bus 37 of FIG~RE 4 in tha-t they are
used for overflow. However, unlike the arranaement of FIGURE 4,
the unusable ports associated with the bus 37 do not exist.
Thus, an interconnecting system for interconnecting modules
has been described. The system employs a general distributed
architecture which permits effective routing of signals
through modules without utilizing time slots within the modules.
Intertie units are employed which permit signals to be routed
in a carousel or ring-like fashion. The invented system provides
a highly reliable and cost-effective interconnecting architecture
15~ for time division multiplexed cor.lmunicatlon systems.
'
~ ;,t3 _ ~_