 |
Claims  |
|
|
We claim:
1. A circuit- and packet-switching unit of a switching node for a switching
system having one or more communication switching nodes each including a
plurality of circuit- and packet-switching units comprising:
means connectable to either one of a communication endpoint or a unit of
another switching node, for receiving simultaneously a plurality of
multiplexed circuit- and packet-switched communications in a same format
from a connected either one of the endpoint or the unit of the other
switching node;
means connectable to the units of the switching node that includes the
unit, for broadcasting all communications received by the receiving means
to each unit within the switching node;
means connectable to the units of the switching node that includes the unit
and responsive to receipt of communications broadcast by the units of the
switching node that includes the unit, for determining which of the
received broadcast communications are destined for receipt or transmission
by the unit and discarding all other received broadcast communications;
means responsive to information included in received broadcast
communications that are destined for receipt by the unit, for multiplexing
together received broadcast circuit- and packet-switched communications
that are received from a plurality of the units of the switching node that
includes the unit and are destined for transmission by the unit; and
means connectable to either one of a communication endpoint or a unit of
another switching node, for transmitting simultaneously a plurality of
circuit- and packet-switched communications multiplexed together by the
multiplexing means in the same format to a connected either one of the
endpoint or the unit of the other switching node.
2. The unit of claim 1 wherein
the determining means includes a plurality of means, a different one
connectable to each unit of the node that includes the unit, each for
receiving and storing communications broadcast by the connected unit; and
wherein
the multiplexing means is for multiplexing together circuit- and
packet-switched communications that are stored by a plurality of the
receiving and storing means and intended for transmission by the unit.
3. The unit of claim 1 wherein
the determining means further includes:
means for determining whether an individual received packet-switched
communication is destined for receipt or transmission by the unit from an
address included in the individual packet-switched communication; and
means for determining whether an individual received circuit-switched
communication is destined for transmission by the unit from information
included in a packet-switched communication destined for receipt by the
unit, and received by the unit prior to the individual circuit-switched
communication.
4. The unit of claim 1 wherein
the format is
a multiplexed format wherein a bandwidth is divided into a plurality of
identifiable portions and wherein each communication occupies a different
identifiable portion of the bandwidth; and wherein
the determining means includes:
first means for determining whether an individual received packet-switched
communication is destined for receipt or transmission by the unit from an
address included in the individual packet-switched communication; and
second means for determining whether an individual circuit-switched
communication is destined for transmission by the unit from the portion of
a link's bandwidth that is occupied by the communication.
5. The unit of claim 4 wherein
the second means further includes
third means responsive to a packet-switched communication addressed to the
unit and received by the unit prior to an individual circuit-switched
communication, for storing information included in the received
packet-switched communication that identifies the portion of a link's
bandwidth that will be occupied by the individual circuit-switched
communication.
6. The unit of claim 5 wherein
the third means further store information that identifies the portion of
the bandwidth of the unit's connected link that is to be occupied by the
individual circuit-switched communication; and wherein
the transmitting means includes bandwidth-portion interchange means for
transmitting the individual circuit-switched communication in the
connected link's bandwidth portion specified by the information stored by
the third means.
7. The unit of claim 2 wherein
the format is time-division multiplexed format having different selected
time slots carrying different circuit-switched communications and having
other time slots carrying packet-switched communications; wherein
each receiving and storing means of the determining means comprises means
for deciding whether an individual received broadcast communication is
packet-switched or circuit-switched; wherein
the determining means comprises means for determining from addressing
information included in packet-switched communications which of the stored
packet-switched communications are intended for receipt or transmission by
the unit and discarding all other stored packet-switched communications;
and wherein
the multiplexing means comprises
time-slot-interchanging means responsive to information included in stored
packet-switched communications that are intended for receipt by the unit,
for placing selected circuit-switched communications stored by a plurality
of the receiving and storing means into selected time slots of the
time-division multiplexed format transmitted by the transmitting means and
discarding other stored circuit-switched communications, and for placing
packet-switched communications stored by the plurality of the receiving
and storing means that are intended for transmission by the unit into
other time slots of the time-division multiplexed format transmitted by
the transmitting means.
8. The unit of claim 7 wherein
the broadcasting means include means for broadcasting all communications
received by the receiving means to each unit within the switching node in
the same format in which the communications are received.
9. A circuit and packet switching system comprising:
a plurality of interconnected identical switching units forming one or more
communication switching nodes, each switching node including
a plurality of different ones of the identical switching units, and
a communication medium interconnecting all of the units of the switching
node; each unit comprising
means connectable to either one of a communication endpoint or a unit of
another switching node, for receiving simultaneously a plurality of
multiplexed circuit- and packet-switched communications in a same format
from a connected either one of the endpoint or the unit of the other
switching node;
means connectable to the units of the switching node that includes the
unit, for broadcasting all communications received by the receiving means
to each unit within the switching node;
means connectable to the units of the switching node that includes the unit
and responsive to receipt of communications broadcast by the units of the
switching node that includes the unit, for determining which of the
received broadcast communications are intended for receipt or transmission
by the unit and discarding all other received broadcast communications;
means responsive to information included in received broadcast
communications that are intended for receipt by the unit, for multiplexing
together received broadcast circuit- and packet-switched communications
that are received from a plurality of the units of the switching node that
includes the unit and are intended for transmission by the unit; and
means connectable to either one of a communication endpoint or a unit of
another switching node, for transmitting simultaneously a plurality of
circuit- and packet-switched communications multiplexed together by the
multiplexing means in the same format to a connected either one of the
endpoint or the unit of the other switching node.
10. The system of claim 9 wherein
the determining means includes a plurality of means, a different one
connectable to each unit of the node that includes the unit, each for
receiving and storing communications broadcast by the connected unit; and
wherein
the multiplexing means is for multiplexing together circuit- and
packet-switched communications that are stored by a plurality of the
receiving and storing means and intended for transmission by the unit.
11. The system of claim 9 wherein
the plurality of interconnected identical switching units form a plurality
of communication switching nodes each comprising a plurality of different
ones of the units, and wherein
the system further includes
at least one communication link interconnecting the plurality of switching
nodes and each having an identical communication format, at least one unit
of each switching node each being connected by one of the communication
links to a unit of another switching node, to form a hierarchical circuit
and packet switch.
12. A circuit- and packet-switching network comprising:
a plurality of interconnected identical switching units forming one or more
communication switching nodes, each switching node including
a plurality of different ones of the identical switching units, and
a communication medium interconnecting all of the units of the switching
node;
a plurality of communication endpoints each connected to the one or more
switching nodes for at least one of transmitting communications to others
of the endpoints or receiving communications from others of the endpoints
through at least one of the one or more switching nodes; and
a plurality of communication links each connecting a different one of the
units of a switching node to either (a) one of the communication endpoints
or (b) a unit of another switching node, all of the links having an
identical communication format;
each unit comprising
means connectable to either one of a communication endpoint or a unit of
another switching node, for receiving simultaneously a plurality of
multiplexed circuit- and packet-switched communications in a same format
from a connected either one of the endpoint or the unit of the other
switching node;
means connectable to the units of the switching node that includes the
unit, for broadcasting all communications received by the receiving means
to each unit within the switching node;
means connectable to the units of the switching node that includes the unit
and responsive to receipt of communications broadcast by the units of the
switching node that includes the unit, for determining which of the
received broadcast communications are intended for receipt or
retransmission by the unit and discarding all other received broadcast
communications;
means responsive to information included in received broadcast
communications that are intended for receipt by the unit, for multiplexing
together received broadcast circuit- and packet-switched communications
that are received from a plurality of the units of the switching node that
includes the unit and are intended for retransmission by the unit; and
means connectable to either one of a communication endpoint or a unit of
another switching node, for transmitting simultaneously a plurality of
circuit- and packet-switched communications multiplexed together by the
multiplexing means in the same format to a connected either one of the
endpoint or the unit of the other switching node.
13. The network of claim 12 wherein
the determining means includes a plurality of means, a different one
connectable to each unit of the node that includes the unit, each for
receiving and storing communications broadcast by the connected unit; and
wherein
the multiplexing means is for multiplexing together circuit- and
packet-switched communications that are stored by a plurality of the
receiving and storing means and intended for retransmission by the unit.
14. The network of claim 12 wherein
the units have no information on routing of communications through the
system other than the information to which the multiplexing means
responds, and wherein
the information to which the multiplexing means responds is provided to the
units by the endpoints. |
|
 |
|
 |
Claims |
|
|
|
Description |
|
|
TECHNICAL FIELD
This invention generally relates to the architecture of a multi-node
integrated circuit- and packet-switching system, and particularly relates
to the system control architecture and to the architecture of individual
switching nodes and their switching units that make up the system.
CROSS-REFERENCE TO RELATED APPLICATION
A. D. Franklin and R. W. Gebhardt, "Control Architecture of a Multi-Node
Circuit- and Packet-Switching System", Ser. No. 410,777, filed on Sept.
21, 1989 and assigned to the same assignee, shares herewith a
substantially-identical disclosure.
BACKGROUND OF THE INVENTION
It is known that different kinds of communication traffic are
advantageously switched by means of different switching techniques. For
example, the characteristics of voice communications are typically most
compatible with circuit-switching techniques, while the characteristics of
data communications are typically most compatible with packet-switching
techniques.
When faced with switching of both voice and data traffic, the art has
sought to provide the best-suited switching technique for each kind of
traffic within a single switching system. An example thereof is U.S. Pat.
No. 4,535,448, which discloses a single-node switching system having both
a time-division multiplex (TDM) bus and a packet bus. Pulse-code modulated
(PCM) voice signals are circuit-switched over the TDM bus between various
ports that are connected to the TDM bus, while packetized data are
packet-switched over the packet bus between the ports that are connected
to the packet bus.
An advantage of the dual-bus system shown in this patent is that the packet
bus permits efficient bandwidth utilization and high data rates. This is
in contrast to a PCM-only switching system, which allocates bandwidth in
fixed increments to data connections just like it does for voice
communications, and maintains those allocations for the duration of the
connections irrespective of actual use of the bandwidth. Hence, the
dual-bus system uses bandwidth more efficiently and obtains higher
throughput rates for a given bandwidth than the PCM-only system.
While the dual bus system shown in U.S. Pat. No. 4,535,448 provides
efficient circuit-switching and packet-switching service for a single-node
switching system, it offers no solution to the problem of transmitting
both circuit-switched voice and packet-switched data between the various
nodes of a multi-node switching system wherein each node thereof
comprises, e.g., a switching system of the type shown in U.S. Pat. No.
4,535,448.
Arrangements for routing circuit-switched and packet-switched traffic
between the nodes of a multi-node switching system are known. For example,
U.S. Pat. No. 4,556,972 discloses an arrangement that provides for such
interconnection by carrying both circuit-switched and packet-switched
traffic between nodes in TDM circuit-switched form. However, the
inter-node packet transmission capabilities of this system are not at the
high data rates characteristic of packet buses. Instead, the inter-node
packet transmission rate of this system is relatively low and is limited
to that of a single system time slot. The reason for this is that this
system serves inter-node packet communications by breaking up an
inter-node packet into bytes and then transmitting the bytes one-by-one
during successive occurrences, in successive transmission frames, of a
time slot to which the packet transmission is assigned. The packet bytes
are combined at the receiving end and reconstituted into a complete
packet.
It can be seen that the system of this patent serves inter-node packet
connections in the same manner as it serves inter-node voice connections.
Namely, it serves both types of connections by assigning a single time
slot to each such connection and by then transmitting the voice signals or
data associated with the connection between the transmitting and receiving
ends via the assigned time slot during successive occurrences of the time
slot. This is the same manner in which the abovementioned PCM-only
switching system handles intra-node voice and data transmissions.
Consequently, the system disclosed in U.S. Pat. No. 4,556,972 fails to
realize for inter-node transmissions the advantages that are offered for
intra-node transmissions by the abovementioned system of U.S. Pat. No.
4,535,448.
The art has sought to introduce the advantages of high-speed packet data
transfers to communication systems wherein inter-node transmissions of
both circuit-switched and packetized traffic are accomplished by means of
TDM facilities. For example, U.S. Pat. No. 4,731,785 discloses an
arrangement for inserting circuit-switched and packetized traffic into
different time slots of an inter-node TDM bus such that the
circuit-switched traffic is carried by time slots of the TDM bus in a
substantially conventional manner while packets are broken up into bytes
and sequential bytes of a packet are inserted into sequential "idle" time
slots of the TDM bus. "Idle" time slots are those that are not presently
carrying circuit-switched traffic. "Idle" time slots are distinguished by
the value of a special information bit that is inserted into each time
slot, and which thereby specifies whether the traffic carried by the time
slot represents circuit-switched or packetized traffic. The bytes of an
inter-node packet may thus be carried by a plurality of time slots of a
single transmission frame. This is in contrast to the system of U.S. Pat.
No. 4,556,972, wherein the bytes of an inter-node packet are carried by
only one time slot of a single frame. The arrangement of U.S. Pat. No.
4,731,785 thereby significantly increases the inter-node packet
transmission rate to the full bandwidth of the TDM bus that is represented
by the "idle" time slots.
While it does provide the requisite transmission rates for packetized
traffic, U.S. Pat. No. 4,731,785 only discloses an arrangement for
transporting integrated circuit-switched and packetized traffic between
only two nodes. Unlike U.S. Pat. No. 4,556,972, it does not disclose a
switching system network capable of interconnecting a plurality of
endpoints, each one of which comprises a source of separate
circuit-switched and packet-switched traffic, in a manner whereby the
switching systems of the network are capable of routing individual
communications, be they circuit-switched or packet-switched, to different
individual ones of the endpoints. Rather, it discloses a switching network
capable of interconnecting only two such endpoints and having no routing
capability. And neither one of the patents offers any suggestions on how
their respective advantages might be achieved within a single integrated
circuit-and packet-switching network.
It is therefore a problem to provide a multi-node switching system having
the capability of carrying and routing integrated inter-node
circuit-switched and packet-switched traffic, with the packetized traffic
being transmitted at the high data rates characteristic of packet
switching systems.
A further difficulty is encountered in the area of capacity of the
switching system to handle traffic growth. Traditionally, switch design
has proceeded by first determining a maximum switch size, and then
implementing a design that meets this size objective. Maximum designed
switch size has often been based on the call-handling capacity of a
feasible call control processor, and on engineered switch fabric capacity
and switching bandwidth requirements. But once the switch design has been
defined and the switch has been built, adding switching capacity beyond
the predefined limit has been impossible, or very difficult at best.
It is therefore another problem to provide a multi-node switching system
that avoids limitations on system growth, and that provides integrated
circuit- and packet-switching bandwidth and capacity for as-yet unforeseen
bandwidth-hungry applications and for unbounded line size growth--in other
words, to provide a system that has a substantially-limitless growth
architecture.
Certain switching fabric architectures do offer the possibility of
substantially-limitless growth--the banyan network topology is a good
example. However, most switching systems which are available today,
irrespective of their switching fabric topology, are of the common control
type. Common control systems, in general, comprise a switching fabric such
as an array of crosspoints forming a network, and a centralized control
which operates the fabric in order to establish a communication path. The
centralized control typically has a maximum size expansion limit which is
determined by the capacity of the control. Beyond a predetermined point,
further size expansion requires replacement of the control, which
generally requires total replacement by a different system. This also
means that the total foreseeably-required control capacity must be
provided right from the start, regardless of the line size of the system
as initially implemented or put into service. Therefore, common control
design is not economical for small-sized systems. And furthermore, even if
the switching fabric itself is modularized and distributed, the number of
control links required to connect the central control to all of the
modules of a growing system quickly becomes prohibitively cumbersome and
expensive.
In order to alleviate some of the problems associated with common control
systems, the prior art has sought to develop control architectures that
use distributed or progressive control. Distributed control systems, in
general, comprise a number of switching stages which combine both control
and switching in each stage. Since control and switching are provided in
coordinated amounts, distributed control systems are economical at small
line sizes and have virtually unlimited growth potential. Examples of
distributed control systems are found in U.S. Pat. No. 3,860,761 which
applies the distributed control concept to a space-division circuit
switch, and in U.S. Pat. No. 4,488,288 which applies the concept to a
banyan packet-switching network.
However, successful application of distributed control to other types of
switches, such as time-division circuit switches and integrated
circuit-and-packet switches, has heretofore been lacking. For example, a
highly-touted attempt by a major international telecommunications
equipment manufacturing company to develop a distributed-control TDM
switch has been reported in the recent past to have failed drastically,
after reported expenditures of over a billion dollars in development
costs. It is therefore a further problem to provide a multi-node TDM
switching system having distributed control, and particularly to provide a
multi-node integrated circuit- and packet-switching system having
distributed control.
SUMMARY OF THE INVENTION
The invention is directed to solving these and other problems and
disadvantages of the prior art. According to the overall invention, there
is provided a modular, substantially infinitely-growable, multi-node
switching system, as well as the modules therefor, that operates under
distributed control to serve integrated circuit-switched and
packet-switched traffic at the data rates appropriate for each type of
traffic.
The system comprises a plurality of interconnected identical switching
units that form at least one communication node. A communication node is a
modular building block of the system; a switching unit is a modular
building block of a node and hence also of the system. A plurality of
communication endpoints is connected to the node or nodes for
communicating with each other through the node or nodes. A plurality of
communication links each connect a different one of the system's switch
units to either (a) one of the communication endpoints or (b) a unit of
another switching node. All of the links have the same communication
format. Each node includes a plurality of different ones of the identical
switching units, and a communication median that interconnects all of the
units of the node. In turn, each switching unit comprises a transmitting
and a receiving section which are structured and function as follows.
The receiving section is connectable to either one of a communication
endpoint and a unit of a switching node other than a switching node that
includes the unit (e.g., through one of the links), for receiving
simultaneously a plurality of multiplexed circuit- and packet-switched
communications in the same format from whichever one of the endpoint and a
unit of the other node it is connected to. The receiving section is
further connectable to the units of the switching node (e.g., through the
medium), for broadcasting all received communications to each unit within
the switching node.
The transmitting section is connectable to the units of the switching node
that includes the unit, and is responsive to receipt of communications
broadcast by the units of the switching node that includes the unit, for
determining which of the received broadcast communications are intended
for receipt or retransmission by the unit and discarding all other
received broadcast communications. The transmitting section responds to
information included in received broadcast communications that are
intended for receipt by the unit, and multiplexes together received
broadcast circuit- and packet-switched communications that are received
from a plurality of the units of the switching node that includes the unit
and that are intended for retransmission by the unit. The transmitting
section is further connectable to either one of a communication endpoint
and a unit of a switching node that includes the unit, for transmitting
simultaneously a plurality of circuit- and packet-switched communications
multiplexed together by the multiplexing means in the same format to
whichever one of the endpoint and a unit of the other node it is connected
to.
Preferably the transmitting section includes a plurality of stores a
different one connectable to each unit of the node that includes the unit,
each for receiving and storing communications broadcast by whichever unit
it is connected to. Furthermore, each store preferably stores received
circuit-switched communications separately from received packet-switched
communications.
The modular, or building, or building-block, switching architecture
characterized above is a growable flexible hardware architecture unbounded
by design constraints. A system that uses this architecture can be
expanded substantially without limit by adding switching units and
associated links to switching nodes, and by adding new switching nodes and
links to the system. The system provides ever-increasing communication
processing power as units and nodes are added. System size is not limited
in size and growth by processor throughput; rather, the architecture
allows for substantially unbounded system growth. System cost is directly
proportional to the number of links used in the configuration of interest.
In other words, the aforementioned additions increase system bandwidth at
a proportional increase in cost. Furthermore, the cost of increased
bandwidth can be traded off against a lowered likelihood of blocking
and/or interconnection redundancy. Redundancy has the beneficial trait
that link failures do not lead to disruptions of traffic, but merely to
increased likelihood of blocking. Furthermore, there is no central point
of control; rather, control is decentralized and each node is capable of
independent operation, thereby avoiding the hazard of relying on a
vulnerable central processor. Consequently, the switching components may
be geographically distributed; for example, modules may be remoted.
The architecture logically implements a completely non-blocking circuit
switch for switching communications such as voice and compressed video,
and a self-routing packet switch for switching communications such as
wide-band data and images. The building blocks are non-blocking bandwidth
(e.g., time-and-space)-division switches, capable of switching any time
slot incoming on any link to any time slot outgoing on any link. Hence,
the system as a whole may be engineered therefrom to be either blocking or
fully non-blocking. By making packets | | |
