DE69935775T2 - PACKAGE TRANSMISSION WITH COMMON REFERENCE TIME - Google Patents

Description

STATE OF THE ART:

These The invention relates generally to a method and to a Device for transmitting data in a communications network. Especially This invention relates to the timely forwarding and Delivering data via the network and its destination nodes. Timely forwarding is possible through time information generated by a Global Positioning System (GPS) available globally are. As a result, the parameters of end-to-end performance have the same about loss, delay and jitter deterministic or probabilistic Guarantees.

Further allows this invention provides a method and apparatus for integrating the transmission of two traffic types over a data packet communication network. More precisely, this invention secures Timely forwarding and delivery of data packets from Constant Bitrate (CBR) and Variable Bit Rate (VBR) sources Network and to their destination nodes.

Further This invention relates to a method and apparatus for sending data in a communication network via communication links with variable delays. More accurate This invention relates to timely forwarding and delivery of data via networks with transmission links with variable delays to their destination nodes while It is ensured that the end-to-end performance parameters such as about loss, delay and jitter either deterministic or probabilistic Have guarantees.

In addition, creates this invention a method and apparatus for monitoring, cell flow control and invoicing the transfer a data packet in a communication network. Specifically, this secures Invention the monitoring, Cell flow control and billing in networks with timely Forwarding and delivery of the data packets to their destination nodes. Consequently, the end-to-end performance parameters are such as loss, delay and jitter predictable so that it is possible to measure it. consequently these measurements are taken in monitoring, Cell flow control and accounting used.

In addition, creates This invention relates to a method and apparatus for transmitting Data in a heterogeneous communication network, the two types of Switching node has: (i) asynchronous and (ii) synchronous with common time reference. More precisely, this invention creates the timely Forwarding and delivery of data over the network and to its destination nodes. Consequently, the end-to-end performance parameters such as loss, delay and jitter either deterministic or probabilistic Guarantees.

Furthermore This invention facilitates a routing decision as well using timing information as well as position information of packages within timeframes. In this case, there is none Need to decode the address in the packet header, where possible is the whole data packet (including the header) like this to encrypt, how about it a public one Transfer backbone network which is an important safety feature. Therefore it is possible, about this new one Communication network a wide variety of data packets such as IP (Internet Protocol) and ATM (asynchronous transfer mode) to transport.

The strong increase of fast communication links, faster processors and affordable, multimedia-ready personal computer leads to Need for wide area networks, the real-time data such as telephony and video can. However, the end-to-end transport requirements of real-time multimedia applications a big Challenge, not by the current network technologies satisfactorily resolved can be. Such applications as video teleconferencing as well the audio and video multipoint connection creates data in one Wide range of bit rates and require a predictable, stable Performance and strict limits to the loss rates, to the average delay and the delay changes ( "Jitter"). These properties and performance requirements are incompatible with the services that the present ones Can provide line and packet switching networks.

Packet-switched networks such as the IP-based (Internet Protocol-based) Internet and IP-based intranets [see e.g. BA Tannebaum, Computer Networks (3rd edition), Prentice Hall, 1996] and the ATM (asynchronous transfer mode) [see e.g. For example, Handel et al., ATM Networks: Concepts, Protocols, and Applications (2nd Edition) Addison-Wesley, 1994] treat discontinuous data more efficiently than circuit switching because of their statistical multiplexing of the packet streams. However, the current packet switches and routers work asynchronously and provide only one best effort service in which end-to-end delay and end-to-end jitter are neither guaranteed nor limited. In addition, statistical changes in traffic often cause congestion, resulting in excessive delays and loss of pacts, thereby increasing fidelity Real time streams is significantly reduced at their receiving points. In fact, the performance characteristics of a given connection at the time of connection establishment are not even predictable according to the best effort service.

The efforts as well as IP as well as for ATM Defining advanced services is done on two levels: (1) Definition of the service and (2) specification of provisioning procedures various services for different packet streams. The former defines interfaces, data formats, and performance goals. The latter specifies procedures for processing packets Hosts and switches / routers. The service types defined for ATM include constant bit rate (CBR), variable bit rate (VBR) and available bit rate (ABR). For IP the defined services contain guaranteed performance (bitrate, Delay) controlled flow and best effort [J. Wroclawski, "Specification of the Controlled-Load Network Element Service ", IETF RFC 2211, September 1997] [Shenker, u. a., "Specification of Guaranteed Quality of Service ", IETF RFC 2212, September 1997]. For IP or ATM are signaling protocols, eg. RSVP and UNI3.1, specifies the tax information that the fixing the desired Enable services [R. Braden, "Resource ReSerVation Protocol (RSVP) version 1 Functional Specification, IETF Request for Comment RFC2205 ", September 1997] (Handel et al., "ATM Networks: Concepts, Protocols, and Applications ", (2nd edition) Addison-Wesley, 1994]. These protocols handle the transport of data to a destination, what's known as point-to-point or multi-destination broadcasting is. Furthermore is currently under the auspices of the IETF the SIP, a protocol higher Level to enable the setting of meetings, the underlying services use, under definition [Handley, u. a., "SIP Session Initiation Protocol", <draft-draft-ietf-mmusicsip-04.ps>, November 1997].

The Method for providing various services according to packet switching fall under the general title of Quality of Service (QoS). The state of the art in terms of QoS can be divided into two parts (1) local-time traffic modeling without deadlines, z. B. [M.G.H. Katevenis, "Almost Switching And Fair Control Of Congested Flow In Broadband Networks ", IEEE Journal on Selected Areas in Communications, SAC-5 (8): 1315-1326, October 1987; Demers, u. a., "Analysis and Simulation of A Fair Queuing Algorithm ", ACM Computer Communication Review (SIGCOMM'89), pages 3-12, 1989; S. J. Golestani, "Congestion-Free Communication in High-Speed Packet Networks, IEEE Transcripts on Communications, COM-39 (12): 1802-1812, December 1991; Parekh, u. a., "A Generalized Processor Sharing Approach To Flow Control - The Multiple Node Case ", IEEE / ACM T. on Networking, 2 (2): 137-150, 1994], and (2) Traffic shaping with Deadline planning, z. For example, [Ferrari et al., "A Scheme For Real-time Channel Establishment In Wide-Area Networks ", IEEE Journal on Selected Areas in Communication, SAC-8 (4): 368-379, April 1990; Kandlur et al., "Real Time Communication in Multi-Hop Networks ", IEEE Trans. On Parallel and Distributed Systems, Vol. 5, no. 10, pp. 1044-1056, 1994]. These two approaches are based on the manipulation of local queues through each router with little Interacting with other routers. The weighted fair queuing (WFQ), the these accesses embodies is based on cyclic operation of the output port queues, the service level of a specific class of packets through the length of time that determines its queue in each cycle is operated [Demers u. a., "Analysis and Simulation of A Fair Queuing Algorithm ", ACM Computer Communication Review (SIGCOMM '89), pages 3-12 1989]. If these additions for the Transport of real-time streams are used, they have inherent Restrictions. If traffic shaping is configured without scheduling, that it works without loss at high usage, are the delay and the jitter is inversely proportional to the connection bandwidth, d. h., that low-rate connections have a big delay and a big Can experience jitter in the network. In traffic shaping with deadline planning, the delay and the jitter at the expense of possible Stowage and possible Loss controlled.

The Realizing that the processing of packets through exchanges and router is a performance bottleneck, led to the development of procedures to improve performance by simplifying processing the packages. Multi-Protocol Label Switching (MPLS) sets the destination address in the packet header into a short tag which defines the routing of the packet in the network [Callon, u. a., "A Proposed Architecture For MPLS "<draft-ietf-mpls-arch-00.txt> INTERNET DRAFT, August 1997].

The Real Time Protocol (RTP) [H. Schultzrinne u. a., RTP: A Transport Protocol for Real-Time Applications, IETF Request for Comment RFC1889, January 1996] a procedure for the encapsulation of time-sensitive data packets and for attaching time-related information like timestamp and packet sequence number to the data. RTP is currently the accepted procedures for the transport of real-time streams over IP connection networks and for the Packet audio / video telephony based on ITU-TH.323.

Compared to the packet switching, the main Asynchronous methods are mainly used in circuit switching to find synchronous methods. However, some synchronous packet switching methods have been proposed. IsoEthernet or IEEE 802.9a (IEEE 802.9a Editor: IEEE 802.9A "Isochronous Services With CSMA / CD MAC Service", IEEE Draft, March 1995] combines CSMA / CD (IEEE 802.3), providing asynchronous packet switching is, with N-ISDN and H.320, which is a circuit switch, over an existing Ethernet infrastructure (10Base-T) .This is a hybrid solution with two different switching methods: N-ISDN circuit switching and Ethernet packet switching The IsoEthernet TDM uses a fixed allocation of bandwidth for the two methods regardless of their degrees of utilization.This access to resource sharing leads to unwanted side effects such as underuse of the circuit switch part while the asynchronous packet switch is overloaded while not being able to use the idle resources in the line switching part.

One Access to an optical network that uses synchronization became in the synchronous optical hypergraph [Y. Ofek, "The Topology, Algorithms And Analysis Of Synchronous Optical Hypergraph Architecture ", PhD Dissertation, Electrical Engineering Department, University of Illinois at Urbana, Report No. UIUCDCS-R-87-1343, May 1987], itself as well refers to how the packet telephony using the synchronization to integrate is [Y. Ofek, "integration Of Voice Communication On A Synchronous Optical Hypergraph ", INFOCOM '88, 1988] synchronous optical hypergraph will relay via hyper-edges, d. H. passive optical stars, executed. In [Li u. a., "Pseudo-Isochronous Cell Switching In ATM Networks ", IEEE INFOCOM '94, Pages 428-437, 1994; Li, u. a., "Time-Driven Priority: Flow Control For Real-Time Heterogeneous Internetworking ", IEEE INFOCOM '96, 1996) became the idea of synchronous optical hypergraphs on networks with any topology and with point-to-point transmission links applied. The two treatises [Li u. a., "Pseudo-Isochronous Cell Switching In ATM Networks", IEEE INFOCOM '94, pp. 428-437, 1994; Li, u. a., "Time-driven Priority: Flow Control For Real-Time Heterogeneous Internetworking ", IEEE INFOCOM '96, 1996] an abstract (high-level) description of what is called "RISC-type routing" in the one similar to a package Way as in the execution a statement in a machine of a computer with reduced Command set (RISC machine) with little detail, if any, in each time frame a jump is forwarded.

The US Pat. No. 5,418,779, Yemini, u. a., and "Isochronets: a High-Speed Network Switching Architecture "- published in the daily report of IEEE INFOCOM 1993, pages 740-747 a switched network architecture with common time reference. The Time reference is used to determine the time in which a Plurality of nodes simultaneously via a preferred routing tree can send to a destination. The plurality of nodes send to another one at a time Destination node.

The Routing - the Selection of an output port for an information segment (i.e., for data packets), that arrives at an input port of a switch - is one Basic function of communication networks. In circuit networks the unit of the switch is one byte and becomes the switch executed on the basis of the location of the byte in a timeframe.

Of the Establishing a connection in a circuit-switched network requires the network reserves a slot for the connection in each frame. The position of the byte in the frame is from link to link different so that each switch has a translation table at each input port from incoming frame positions to respective output ports and Maintains frame positions in it. The sequence of frame positions on the transmission paths of the route forms a conduit for the exclusive one Use of a specific connection is assigned to what a significant inflexibility leads: The connection is in terms of traffic volume through the capacity of the line limited, wherein if the connection does not use the line, no others may use them. This feature is useful for CBR traffic as for PCM telephony, but leads to a low use of the network, if the traffic is bundled is [C. Huitema, Routing in the Internet, Prentice Hall, 1995, and A. Tannebaum, Computer Networks (3rd ed.), Prentice Hall 1996].

DISCLOSURE OF THE INVENTION:

In accordance with the present invention, a system for scheduling and managing data transmission of data packets over packet-switched networks in accordance with claim 1 is disclosed. The arrangement combines the advantages of both circuit and packet switching. It considers both the assignment for the exclusive use of predefined connections and those connections whose lossless transport with little delay and jitter guarantees them. If predefined connections do not use their associated resources, they can be shared by other unreserved data packets without affecting the performance of the predefined connections. On the Internet, the traffic of unreserved data packets is called "best effort" traffic.

It is a method for sending and forwarding packets over Packet switching network described in which the delay between two Placements increases with time, decreases or changes arbitrarily. About every transmission link in the network are in predefined periodic time intervals Forwarded packets. The network's mediations hold one common time reference, either from an external source (such as GPS - global Positioning system) or internally generated and distributed becomes. The time intervals are with simple periodicity and with complex periodicity (like the seconds and minutes of a clock).

At least in its preferred embodiment the invention provides the advantages of timely forwarding in a predefined time interval over two types of link delay changes: (i) increase the delay and (ii) reducing the delay. If the delay At a time, a packet may be eligible for its predefined forwarding time interval delayed be. In this case, the package will be up to the next time interval delayed virtual pipe. When the transmission link delay decreases, the packets are up to their first time interval of their virtual Pipe buffered.

Packages, the on the net over the same route and at the same time intervals forwarded form a virtual pipe and share a pipe ID. The pipe ID can either be explicit, such as an identifier or a label that is generated in the network, or implicitly like be about a group of IP addresses. A virtual pipe creates deterministic service quality guarantees. The time interval in which a switch is a specific packet is passed through the pipe ID of the package, by the time, to which it reaches the mediation, and by the momentary value of the common Time reference determined.

At least in its preferred embodiments the present invention offers the advantage that a compound associated bandwidth and the delay and jitter in the network independently are. To identify virtual pipes, the present invention can be used to identify virtual pipes Invention the MPLS be used. To use the time-based transport too enable, the packet timestamp transmitted in the RTP header may be used becomes.

According to the above mentioned Prior art method of providing packet switching services work exchanges and routers asynchronously. The present invention creates through synchronous methods that use a time reference, the for the Exchanges and terminals, comprising a wide area network, are common, real-time services. The common time reference can realized using the UTC (Coordinated Universal Time) be, z. B. over the GPS (Global Positioning System - see for example http://www.utexas.edu/depts/grg/gcraft/notes/gps/gps.html) globally available is. UTC is everywhere according to international convention equal to the world. UTC is the scientific name for what is usually GMT (Greenwich Mean Time) is called, the time on the 0-line (Original line) of length in Greenwich, England. 1967 set an international agreement the length one second as the duration of 9,192,631,770 oscillations of the cesium atom firmly. The assumption of the atomic second led to the coordination of the clocks around the world and 1972 fixing the UTC. For coordination with the International Bureau for weights and dimensions (BIPM) in Paris in maintaining UTC is the time and Frequency Division of the National Institute of Standards and Technologies (NIST) (see http://www.boulder.nist.gov/timefreq) responsible.

For individual PCs is the UTC timing over GPS maps readily available. For example, the PCI-SG of TrueTime, Inc., (Santa Rosa, CA) for Computers that have PCI expansion slots, an exact time with zero latency. Another way through the the UTC over A network can be provided using the network Time Protocol (NTP) [D. Mills, "Network Time Protocol "(Version 3) IETF RFC 1305]. However, the clock accuracy of the NTP is sufficient for the coordination between exchanges on which this invention is based is not sufficient.

Although the present invention relies on time to control the flow of packets in the network in a manner similar to that of circuit switching, there are great differences between the two approaches. In circuit switching, for each data unit (eg, one byte) at the time it has been sent from its source, it is possible to deterministically predict the future times at which it will be sent by some switch along its route [ Ballart et al., "SONET: Now It's The Standard Optical Network", IEEE Communications Magazine, vol. 29, no. 3, March 1989, pages 8-15]. The time resolution of this advanced knowledge is much shorter than the data unit transmission time. On the other hand, in accordance with the present invention, it is possible for each data unit (e.g., a cell) at the time it has been sent from its source to join to know the future timeframes at which this data unit will be forwarded along its route. However, the time frame that constitutes the accuracy of this advanced timing knowledge is much larger than one unit time of transmission. For example, the transmission time of an ATM cell (53 bytes) over a gigabit per second link is 424 nanoseconds, ie, 294 times shorter than a typical 125 microsecond time frame used in one embodiment of the present invention. There are several sequences that further distinguish the present invention from circuit switching.

The Use of reserved resources through the entire package traffic can be admitted every time the reserved resources are not in use.

Preferably are the synchronization requirements of the transmission speed of physical transmission path independently, while the synchronization in the circuit switching ever more difficult will, while the transmission link speed increases.

For the routing do not need to use timing information, so the Routing such as. B. on the Internet using IP addresses or an identification mark / label can be made.

The Internet "Best Effort" packet forwarding strategy can be integrated into the system.

It becomes a method of monitoring and for the cell flow control of packet traffic in a packet switched network discloses, in which the exchanges a common time reference maintained.

It can be enabled be that the special points in the network in predefined Time intervals determine the degree of packet traffic and the flow control of packages and him in cases, in which the volume of traffic exceeds given degrees, to given Bring back degrees. To facilitate the information gathered through the special points the accounting for Internet services based on network usage and identification erroneous conditions and malicious Forwarding of packets over beyond the predetermined value may cause excessive delay.

It will be a procedure for the exchange of timing messages and data packets between synchronous investigations with a common time reference and between Terminals / Gateways and other synchronous exchanges over asynchronous network, d. H. above a network with asynchronous switches. The procedure requires the broadcast of messages that the common time reference too Transport end stations / gateways, who have no direct access to the common time reference, and data packets that respond in response to the timing information and predetermined scheduled time intervals are sent.

It will be procedures for the maintenance of timely forwarding in a predetermined Time interval over three network configurations created: (i) end station too synchronous Mediation over asynchronous (LAN) switches, (ii) end station to the gateway via asynchronous Switches (LAN switches) and then to synchronous switching (iii) synchronous switching for synchronous switching over asynchronous Negotiations.

These and other aspects and features of the present invention will be described in the following drawings and the attached description discussed.

BRIEF DESCRIPTION OF THE DRAWINGS:

1 Figure 3 is a schematic representation of a virtual pipe and its timing relationship with a common time reference (CTR), the delay being determined by the number of time frames between the forwarding time at node A and the forwarding time at node D;

2 Figure 12 is a schematic representation of a plurality of virtual pipes sharing particular ones of the exchanges;

3 Fig. 10 is a schematic block diagram representation of a switch disclosed in accordance with the present invention which utilizes a common time reference from the GPS (Global Positioning System) for the timely forwarding of packets;

4 illustrates the relationship between the local common time reference (CTR) at the switches and how the local time multiplicity is projected onto the real-time axis, the time being divided into time frames of a predefined duration;

5 Figure 12 is a schematic representation of how the common time reference is organized into contiguous time cycles of every k time frames and into a contiguous supercycle of every l time cycles;

6 Fig. 12 is a schematic representation of the relationship of the common network time reference and Coordinated Universal Time (UTC) such that each time cycle has 100 time slots of 125 microseconds each and that 80 time cycles are grouped into a one second super cycle;

7 is a schematic representation of a data packet pipeline as in FIG 1 and correlates with data packet movement over the switches 10 versus the time for forwarding via a virtual pipe with the common time reference (CTR);

8th Figure 12 illustrates the mapping of the time frames into and out of a node on a virtual pipe, wherein the mapping repeats in each time cycle, illustrating the input time versus the forwarding output time;

9 Fig. 10 is an illustration of a serial transmitter and a serial receiver;

10 FIG. 12 is a table of the 4B / 5B encoding scheme of the data in accordance with one embodiment of the present invention, such as used by the TAXI AM7968 chipset; FIG.

11 Figure 4 is a table of the 4B / 5B coding scheme for control signals such as the time frame delimiter (TFD) used by the AM7968, for example, in accordance with an embodiment of the present invention;

12 Fig. 10 is a schematic block diagram of an input port with a routing control unit;

13 FIG. 12 is a schematic diagram of the routing control unit that determines to which output port an incoming data packet should be switched, and that appends the arrival time information (ToA information) to the data packet header;

14 Fig. 10 is a flowchart of the routing control unit operation;

15A and 15B illustrate two generic data packet headers with the virtual pipe ID (PID) and the priority bit (P), where 15A illustrates a package without timestamp field and where 15B illustrates a packet with timestamp field and also shows how the common time reference value, the arrival time (ToA), is applied by the routing control unit;

16 Figure 3 is a schematic block diagram of an output port with a scheduling controller and a serial transmitter;

17 Fig. 10 is a schematic block diagram of the double buffer scheduler;

18 is a flow chart of the dual buffer scheduling controller 46 ;

19 Figure 13 is a functional block diagram of the general scheduling controller with its transmit buffer and select buffer controller;

20 Fig. 10 is a flowchart describing the packet scheduling controller operation for calculating the forwarding time of a packet based on the following input parameters: PID 35C, ToA 35T, and CTR 002;

21 is a flowchart that illustrates the operation of the select buffer controller 45D illustrated;

22 illustrates the Real-Time Protocol (RTP) packet header with 32-bit timestamp fields; and

23 Fig. 10 is a flowchart describing the packet scheduling controller operation for calculating the dispatch time of a packet based on the following input parameters: PID, ToA, CTR, and the RTP timestamp;

24 Fig. 12 is a schematic description of a switch having a common time reference distribution in time frames with predetermined positions such that the input port can uniquely determine the positions;

25 is a description of the timing sharing of the common time reference in cycles of k time frames each, with each time frame further divided into four predefined parts: a, b, c and d;

26 FIG. 10 is a schematic diagram of the time-based routing control unit. FIG. This unit determines to which output port a data packet should be switched and attaches the input time and position information to the data packet header;

27 FIG. 10 is an example of a routing and scheduling table at one of the incoming ingress ports using the arrival time or arrival time frame (ToA) and the position count to determine: (i) the output port, (ii) the outbound time frame, and (iii) the outbound packet position in the outgoing time frame;

28 Figure 12 is a schematic representation of a data packet being sent over the fabric to the output port;

29 FIG. 10 is an example of a routing and scheduling table at one of the incoming ingress ports that uses the timestamp and position information to determine (i) the egress port, (ii) the outbound time frame, and (iii) the outbound packet position in the outbound time frame ;

30 Fig. 10 is a flowchart of the routing control unit operation;

31 is a flow chart of the operation of the data packet planning control unit 45A ;

32 Fig. 10 is a flowchart of another embodiment of the selection buffer control unit 45D ;

33 Figure 10 is a schematic diagram of another alternative embodiment routing control unit that determines which output port an incoming data packet should be switched to, and that attaches the arrival time information (ToA information) to the data packet header;

34A and 34B 12 are schematic diagrams of two generic data packet headers having a virtual pipe ID (PID) and priority bits (P1 / P2): (A) a packet without a timestamp field and (B) a packet with a timestamp field. In addition, this drawing shows how the common time reference value, the arrival time (ToA), is attached by the routing control unit;

35 is a table that classifies data packets;

36 FIG. 10 is a flowchart of an alternate embodiment of the routing controller operation; FIG.

37 Figure 13 is a schematic diagram of the scheduling and congestion control unit, with each buffer divided into two parts, one for constant bit rate (CBR) and the other for variable bit rate (VBR);

38 FIG. 10 is a flowchart describing an alternate embodiment of the packet scheduling and re-scheduling controller operation for calculating the forwarding time of a packet based on the following input parameters: pipe ID 35C , Arrival time 35T and the common time reference 002 ;

39 FIG. 13 is a flowchart illustrating an alternative embodiment of the selection buffer and stowage control unit. FIG 45D describes;

40 is a schematic representation of a virtual pipe with a delay whose time varies between node B and node C;

41 Figure 12 is a timing diagram of a virtual pipe with an increasing delay between nodes B and C;

42 Figure 12 is a timing diagram of a virtual pipe with a decreasing delay between nodes B and C;

43 is a schematic representation of a virtual pipe p with an alternative virtual pipe p ';

44 describes a node, node E, in which two virtual pipes, p and p ', converge;

45 is a schematic representation of a resynchronization mechanism of two virtual pipes, pipe ID = p and pipe ID = p ', at node E;

46 Figure 4 is a schematic representation of the delay analysis and scheduling control unit with its transmit buffer and select buffer controller;

47 Fig. 10 is a flowchart describing an alternate embodiment of the selection buffer controller;

48 Fig. 11 is a flowchart describing the operation of the delay analysis and scheduling control unit for calculating the forwarding time of a data packet;

49 specifies a program executed by the delay analysis and scheduling controller for mobile nodes with increasing and decreasing delay in their incoming links;

50 specifies a program executed by the delay analysis and scheduling control unit for communication links in which their delay can change instantaneously, as is the case for SONET transmission links in self-healing SONET rings; and

51 specifies a program executed by the delay analysis and scheduler unit to combine second alternative paths p and p 'into one way.

52 is a schematic representation of the delay monitoring control unit;

53 Fig. 10 is a flowchart of the program executed by the delay monitoring control unit;

54 is a schematic representation of the cell current control and load control unit;

55 Fig. 10 is a flow chart of the program executed by the cell power control and load control unit;

56 Figure 3 is a schematic representation of the connection between an end station and a synchronous virtual pipe switch separated by asynchronous LAN switches;

57 Figure 4 is a schematic representation of the connections between end stations and synchronous virtual pipe switches separated by some asynchronous switches and a gateway that resynchronize the data packets before being forwarded to the synchronous switch;

58 Figure 4 is a schematic representation of the connection between two segments of virtual pipe switches separated by asynchronous switches and routes;

59 Figure 3 is a diagram of the timing relationship between the common time reference signal from a synchronous switch and the timely forwarding of packets from an end station;

60 is an illustration of possible time of arrival changes (ToA) to synchronous switching of data packets routed via asynchronous switches;

61 Fig. 10 is a schematic representation of a data packet resynchronization mechanism at the input port; and

62 specifies a program executed by the delay analysis and scheduling control unit in the case of determining the schedule that is exactly at the delay limit using the timestamp at the data packet header.

DETAILED DESCRIPTION OF THE ILLUSTRATIONS Exporting-EXTENSION FORMS:

Although this invention is an embodiment in many different forms, are in the drawing specific embodiments shown therein, which are described in detail here, wherein the Of course, the present disclosure is exemplary explanation the principles of the invention and the invention not to the specific illustrated embodiments restrict should.

The The present invention relates to a system and method for sending and forwarding packets over a packet-switched network. The exchanges of the network receive a common time reference upright, either from an external source (such as GPS Global Positioning system) is obtained or generated internally and is distributed. The time intervals are in simple periodicity and in complex periodicity (like the seconds and minutes of a clock). A package, that arrives at an input port of a switch is on the Based on specific routing information (eg IPv4 destination address on the internet, VCI / VPI labels in ATM) in the header of the packet to an output port. Any switch along a route from one source to another Target forwards packets at periodic time intervals that are under Use of the common time reference have been predefined. The time interval duration may be longer to be the time duration for the shipment of a parcel is required, with the exact position of a packet in the time interval not specified in this case is.

Packages, within the network the same route and in the same periodic time intervals be forwarded, form a virtual pipe and use the same Pipe ID in common. A pipe ID can either be like an identifier or a label generated within the network, explicitly or how about being implicitly like a group of IP addresses. A virtual pipe can be used to packet data from multiple Sources and to transport multiple destinations. A virtual Pipe ensures deterministic service quality guarantees. The time interval, in which a switch forwards a specific packet is by the pipe ID of the packet, by the time it is mediated reached, and by the instantaneous value of the common time reference certainly. In accordance with the present invention, a stagnation-free for pipe IDs Packet switching in which the capacity in their corresponding forwarding transmission links and time intervals are reserved in advance. Furthermore reach packages that over to ask a virtual pipe their goal in predefined time intervals, which guarantees that the delay jitter is less than or equal to a time interval.

Packets that are forwarded from one source to multiple destinations share the same pipe ID, with the links and the time intervals on which they are forwarded consisting of a virtual tree. This allows the congestion-free forwarding of one Input port to multiple output ports and thus from one source to a variety of destinations. Packets destined for multiple destinations reach all their destinations at predefined time intervals and with a delay jitter no greater than a time interval.

It will be a system for the management of data transmission created from data packets from a source to a destination. The transfer the data packets is during a predetermined time interval supplied, the plurality includes a predefined time frame. Furthermore, the system comprises a Plurality of exchanges. At least one virtual pipe is included two of the exchanges that have communication links are connected in a way. With each of the negotiations a common time reference signal is coupled, and a time allocation control unit has selected predefined ones in response to the common time reference signal Timeframe for the transfer in and out of each of the respective exchanges too. For every mediation There is a first predefined time frame in which a respective Data packet is transmitted to the respective exchange, and a second predefined time frame in which the respective data packet from the respective mediation is forwarded. The time allocation ensures constant fixed intervals between the time of entry into the virtual pipe and the output from it.

In a preferred embodiment is there a predefined subset of predefined timeframes, while where the data packets are transmitted to the switch, where it for every the respective exchanges a predefined subset of predefined Timeframe gives while where the data packets are transmitted from the switch.

each the switches include one or a plurality of addressable ones Input and output ports. A routing control unit forms each of the data packets, each at one of the input ports arrive at the respective investigation, in each case one or more the output ports of the respective exchange.

For each The data packets have an associated time of arrival at one respective input ports. The arrival time is a special one assigned to predefined timeframes. For each of the pictures through the routing control unit has an associated map by a scheduling control unit, which synchronizes each of the data packets the arrival time and the forwarding time. The Forwarding time is a specified predefined Time frame assigned.

In the preferred embodiment There are a plurality of virtual pipes which are at least two of the communication links in a path involving interconnected exchanges. The Communication connection is a connection between two adjacent ones negotiations; Furthermore can any of the communication links of at least two of virtual pipes are used simultaneously. Using of At least two of the virtual pipes can transmit multiple data packets become.

In Some configurations of this invention exist between timeframes for the associated arrival time and those for the forwarding time of departure for each the data packets have a fixed time difference that is constant for all switches is. For some In the exchanges, the fixed time difference is a variable time difference. A predefined interval includes a fixed number of adjacent ones Timeframes that include a time cycle. Data packets that have a given virtual pipe will be forwarded by one Output port in each time cycle in a predefined subset forwarded by timeframe. In addition, the number of data packets, the for a given virtual pipe in each predefined subset of Time frames can be forwarded, also predefined.

The Timeframe, which is a special one of the exchanges in the virtual Pipe are assigned for all time cycles are assigned to the same exchange and are also one the entries in the particular respective mediation or one of Assigned expenses from her.

In Some configurations of this invention are for everyone the time frame in each of the time cycles between entering into one each of the negotiations and the issue of her a constant firm Time. Form a fixed number of contiguous time cycles a super cycle that is periodic. The data packets that have a given virtual pipe are routed in each supercycle within a predefined subset of time frames of one Output port forwarded. About that In addition, the number of data packets that are predefined in each Subset of timeframes within a supercycle for a given one virtual pipe to be forwarded, also predefined.

In the preferred embodiment, the common time reference signal is coupled by a GPS (Global Positioning System) and conforms to the UTC (Coordinated Universal Time) standard. The UTC time signal does not need to be received directly from the GPS. This signal may be from a switch using various means as long as the delay or time uncertainty associated with that UTC time signal does not exceed half of a time frame.

In an embodiment is the supercycle duration measured using the UTC standard (Coordinated Universal Time standards), equal to one second. The supercycle can also be equal to several UTC seconds or equal a fraction of a UTC second.

A Selection buffer control unit forms one of the time frames to issue from a first exchange to a second time frame for input via Communication link to a second exchange. The selection buffer controller uses the UTC time signal to the boundaries between two consecutive To identify timeframe. The selection buffer controller sets in the transmission line a time frame delimiter signal (TFD signal) to the signal of the second Mediation with the exact boundary between the two timeframes to synchronize.

each the data packets are encoded as a data stream, in response on the selection buffer control unit a time frame delimiter in the data stream added becomes. How later explained this can be realized using redundant serial codewords become.

The Communication connections can take off Fiber, copper and wireless communication links, e.g. B. between a ground station and a satellite and between two Satellites that are in orbit around the earth to be. The communication connection between two nodes does not need any to be a serial communication link. It can be a parallel one Communication link used - such a transmission path can have multiple data bits, the associated clock signal and associated Transmit control signals simultaneously.

The Data packets can Internet Protocol (IP) data packets and asynchronous transfer mode cells (ATM cells) and can be over the same forwarded virtual pipe with an associated pipe identifier (PID) become. The PID can have an Internet Protocol (IP) address, an internet protocol group multipoint connection address, an asynchronous transfer mode (ATM), a virtual connection identifier (VCI) and a virtual one Waypoint (VPI) (or used in conjunction as VCI / VPI).

The Routing controller determines three possible assignments of an incoming data packet: (i) the output port, and (ii) the time of arrival (ToA). The ToA will then used by the scheduling controller to determine when a data packet from the selection buffer controller to the nearest switch should be forwarded in the virtual pipe. The routing control unit uses for the mapping from an input port to an output port at least one of the Internet protocol addresses, version 4, (IPv4 addresses), the Internet Protocol Address, Version 6, (IPv6 Addresses), the Internet Protocol Group Multipoint Address, the Internet MPLS Label Address (Multi-Protocol Label Switching Address or multi-protocol tag switching address), the address of the virtual ATM Line Identifier and Virtual Route Identifier (VCI / VPI) and the IEEE 802 MAC address (IEEE 802 Media Access Control Address).

each The data packets comprise a header which has an associated one Contains timestamp. For every There are pictures of the routing control unit an associated map of each of the data packets between the respective ones associated timestamp and associated routing time, which is assigned to one of the predefined time frames by which Planning control unit. The timestamp can record the time to which a package was generated by its application.

In an embodiment the time stamp is determined by an Internet Real-Time Protocol (RTP) and generated by a predefined one of the switches. The timestamp can be used by a scheduling controller to determine the forwarding time of a data packet from an output port.

each The data packets originate from an end station, with the time stamp at the respective end station for inclusion in the respective original data packet is produced.

These Generation of a timestamp can be done by the UTC either by directly from the GPS or using the Internet network time protocol (NTP) are derived.

1 synchronous virtual pipe

In accordance with the present invention, a system for transmitting data packets over a data network is provided, while for reserved traffic, a constant, limited jitter (or constant, limited delay uncertainty) and congestion-induced loss of data packets are not maintained. These features are useful for many multimedia applications such as telephony and video teleconferences essential.

In accordance with the design, the method, and the illustrated implementation of the present invention, one or more of these as in 1 - 2 shown virtual pipes 25 created over a data network with general topology. This data network can span the globe. Every virtual pipe 25 is about one or more in 1 shown negotiations 10 constructed using communication links 41 are connected in a way.

1 illustrates a virtual pipe 25 from the starting port 40 a switch A via switches B and C. This virtual pipe ends at the output port 40 of the node D. The virtual pipe 25 transmits data packets from at least one cell to at least one destination.

2 Figure 3 illustrates three virtual pipes: a virtual pipe 1 from the switch A output to the switch D output, a virtual pipe 2 from the switch B output to the switch D output, and a virtual pipe 3 from the switch A output the output of the switch C.

The data packet transfers over the virtual pipe 25 about the negotiations 10 are determined to occur during a plurality of predefined time intervals, each of the predefined time intervals including a plurality of predefined time frames. The timely transmissions of data packets are made by coupling a signal of the common time reference 002 (CTR signal) with each of the switches 10 achieved.

3 illustrates the structure of a pipeline switch 10 , The mediation 10 includes one or a plurality of input ports 30 , one or a plurality of output ports 40 , a switching fabric 50 and a time receiver 20 the global positioning system (GPS time receiver) 20 with a GPS antenna 001 , The GPS time receiver provides a common time reference signal (CTR) to all input and output ports 002 ,

1.1 The common time reference (CTR) 002

As in 4 is shown, secures the common time reference 002 that with the negotiations 10 coupled, the following property: If the in 4 shown tick points 004 of the local clock on all pipeline switches (for example, on the 1 and 2 at switches A, B, C, and D) to the real-time axis 005 projected, they all occur within predefined synchronization envelopes 003 on. In other words, the local tact tick points 004 occur within the synchronization envelope 003 so that outside of the synchronization envelope, all local clocks have the same clock value.

As in 4 is shown, the common time reference is divided in a predefined manner into time frames Tf of equal duration, usually Tf = 125 microseconds. The time frames are grouped into time cycles. Each time cycle has a predefined number of time frames.

Based on 5 There are k time frames in each time cycle. Contiguous time cycles are grouped together into related super cycles, where, as in 5 is shown in each super cycle i gives time cycles.

6 illustrates how the common time reference signal can be synchronized with the UTC standard (the Coordinated Universal Time standard). In this illustrated example, the duration of each supercycle, measured according to the UTC standard, is exactly one second. As in 6 In addition, the beginning of each supercycle coincides with the beginning of one UTC second. Consequently, periodic cycle and supercycle scheduling is not affected when leap seconds are inserted or dropped for UTC corrections (due to changes in the earth rotation period).

The Timeframes, time cycles, and supercycles become all respective exchanges assigned in the virtual pipe at all times in the same way.

1.2 Pipeline forwarding

The pipeline forwarding refers to data packets that are in each stage with a predefined delay over a virtual pipe 25 (either via a communication link 41 or via a mediation 10 from the entrance port 30 to the starting port 40 ) to get redirected. The data packets enter a virtual pipe from one or more sources 25 and are forwarded to one or more destinations.

This type of pipeline forwarding used in accordance with the present invention is disclosed in U.S. Patent Nos. 5,377,966, 5,395,074 7 illustrated. The data package 41A is forwarded out of switch A during the time frame t-1. This data package 41A after a delay T-ab the switch B reaches this data packet 41A becomes a data packet during the time frame t + 1 41B forwarded from the switch B and reaches after a delay T-bc the switch C. This data packet 41B becomes a data packet during the time frame t + 5 41C forwarded from the switch C. The data package 41C after a delay T-cd reaches switch D. Thus, the delay from the output of switch A to the output of switch C is 6 = t + 5 - (t-1) time frames. As in 7 3, all data packets routed through this virtual pipe have a delay of six time frames from the output of switch A to the output of switch C.

Again, based on 1 is the timely pipeline forwarding of data packets over the virtual pipe 25 illustrated. At time frame 1, a data packet is received from one of the input ports 30 receive the switch A and in the following manner along this virtual pipe 25 forwarded: (i) the data packet 41A becomes at time frame 2 of the time cycle 1 from the output port 40 forwarded to switch A, (ii) the data packet 41B is after 18 Timeframe, timeframe 10 of the time cycle 2, from the output port 40 forwarded to switch B, (iii) the data packet 41C is after 42 Timeframe, at time frame 2 of the time cycle 7, from the output port 40 relayed to switch C and (iv) the data packet 41D is after 19 Timeframe, at time frame 1 of the time cycle 9, from the output port 40 the mediation D forwarded.

• • Alle Datenpakete treten periodisch im zweiten Zeitrahmen eines Zeitzyklus in die virtuelle Pipe 25 ein (d. h., werden aus dem Ausgangsport 40 der Vermittlung A weitergeleitet) und werden nach 79 Zeitrahmen aus dieser virtuellen Pipe 25 ausgegeben (d. h. aus dem Ausgangsport 40 der Vermittlung D weitergeleitet.
• • Die Datenpakete, die in die virtuelle Pipe 25 eintreten (d. h. aus dem Ausgangsport 40 der Vermittlung A weitergeleitet werden), können von einer oder von mehreren Quellen kommen und können über eine oder über mehrere Eingangsübertragungsstrecken 41 die Vermittlung A erreichen.
• • Die Datenpakete, die aus der virtuellen Pipe 25 austreten (d. h. aus dem Ausgangsport 40 der Vermittlung D weitergeleitet werden), können über eine Mehrzahl von Ausgangsübertragungsstrecken 41 zu einem von einer Mehrzahl von Zielen weitergeleitet werden.
• • Die Datenpakete, die aus der virtuellen Pipe 25 austreten (d. h. die aus dem Ausgangsport 40 der Vermittlung D weitergeleitet werden), können zu mehreren Zielen gleichzeitig weitergeleitet werden (d. h. Mehrpunktverbindungs-Datenpaketweiterleitung (Eines-zu-vielen-Datenpaketweiterleitung)).
• • Die Kommunikationsverbindung 41 zwischen zwei angrenzenden der Vermittlungen 10 kann von wenigstens zwei der virtuellen Pipes gleichzeitig verwendet werden.

As in 1 is illustrated, the following applies:

• • All data packets enter the virtual pipe periodically in the second time frame of a time cycle 25 a (ie, be out of the output port 40 the mediation A forwarded) and are after 79 Timeframe from this virtual pipe 25 output (ie from the output port 40 the mediation D forwarded.
• • The data packets that are in the virtual pipe 25 enter (ie from the output port 40 can be forwarded to the switch A), may come from one or more sources and may be via one or more input transmission links 41 reach the mediation A.
• • The data packets coming from the virtual pipe 25 exit (ie from the output port 40 the switch D) can be routed through a plurality of output links 41 be forwarded to one of a plurality of destinations.
• • The data packets coming from the virtual pipe 25 exit (ie those from the output port 40 can be forwarded to switch D) can be forwarded to multiple destinations simultaneously (ie, multipoint connection packet forwarding (one-to-many packet forwarding)).
• • The communication connection 41 between two adjacent exchanges 10 can be used by at least two of the virtual pipes simultaneously.

• • Die drei virtuellen Pipes können über dieselben Kommunikationsverbindungen multiplexieren (d. h. ihren Verkehr mischen).
• • Die drei virtuellen Pipes können während derselben Zeitrahmen und auf beliebige Weise multiplexieren (d. h. ihren Verkehr mischen).
• • Derselbe Zeitrahmen kann von einer oder von mehreren virtuellen Pipes von mehreren Datenpaketen verwendet werden.

In 2 where there are three virtual pipes, the following applies:

• The three virtual pipes can multiplex (ie mix their traffic) over the same communication links.
• The three virtual pipes can multiplex (ie mix their traffic) during the same time frame and in any way.
• • The same time frame can be used by one or more virtual pipes of multiple data packets.

1.3 Capacity allocation of the virtual pipes

For every virtual Pipe there are predefined timeframes in which respective data packets into their respective exchanges, and separate predefined timeframes in which the respective data packets are made their respective negotiations become. Although the timeframes of each virtual pipe at each their mediations along the common time reference to any Can be assigned to a way it is convenient and practical the time frames periodically in time cycles and super cycles assign.

8th FIG. 12 illustrates the timing of a virtual pipe switch wherein there is a predefined subset of time frames (i, 75, and 80) of each time cycle during which data packets are transmitted into this switch, and for this virtual pipe, a predefined subset of time frames (i +3, 1 and 3) of each time cycle during which the data packets are transmitted from this switch. If each of the three data packets has 125 bytes or 1000 bits and if so in each time cycle 80 Time frame of 125 microseconds (ie, 10ms time cycle duration), the bandwidth allocated to this virtual pipe is 300,000 bits per second.

in the Generally, the for calculates a bandwidth or capacity allocated to a virtual pipe, that the number of during each of the time cycles transmitted Is divided by the time cycle duration. In the case of a supercycle the bandwidth allocated to a virtual pipe is calculated by that the number of during each of the super cycles transmitted Bits is divided by the supercycle duration.

Since the like in 3 shown structure of mediation 10 allows a network comprising these switches to operate as a large distributed pipelined architecture, as commonly found in digital systems and computer architectures, it may also be referred to as a pipelined switch.

Each pipeline switch 10 includes a plurality of addressable input ports 30 and output ports 40 , As in 12 is illustrated includes the input port 30 further a routing control unit 35 for mapping each of the data packets arriving at one of the input ports to a respective one of the output ports. As in 16 illustrated includes the output port 40 a scheduler and a send buffer 45 , As in 9 shown is an exit port 40 via a communication connection 41 with an input port 30 connected. The communication link can be realized using various technologies consistent with the present invention.

As in 3 shown is the common time reference 002 to the entrance ports 30 and to the starting ports 40 from the GPS time receiver 20 Delivered its timing signal from the GPS antenna 001 receives. GPS time receivers are available from a variety of manufacturers such as from TrueTime, Inc., (Santa Rosa, CA). With such equipment, it is possible anywhere in the world to maintain a local clock with an accuracy of ± 1 microsecond from the UTC (Coordinated Universal Time) standard.

1.4 The communication link and Time frame encoding

The communication links used for the system disclosed in this invention 41 can be different types: fiber optic, wireless, etc. The wireless transmission links can, for. Between at least one ground station and a satellite, between two satellites moving in orbit around the earth, or between two ground stations.

Based on 9 are a serial transmitter 49 and a serial receiver 31 illustrates how they interact with each link 41 are coupled. For a transmission link 41 In the context of this invention, a plurality of encoding schemes such as SONET / SDH, 8B / 10B fiber channel, 4B / 5B FDDI (4B / 5B distributed fiber data interface) may be used in a serial line. In addition to encoding and decoding data sent over the serial link, the serial transceiver transmits / receives ( 49 in 12 and 31 in 16 ) Control words for a variety of control purposes, which are usually not related to the description of the present invention. However, in accordance with the present invention, a control word, the time frame delimiter (TFD), is used. The TFD identifies the boundary between two consecutive time frames and becomes a serial transmitter in a manner to be described below as part of the output port operation 49 sent when a CTR-002 tact tick occurs. It is necessary to clearly distinguish between the data words conveying the information and the control signal or words (e.g., the TFD is a control signal) over the serial link 41 to distinguish. There are many ways to do this. One possibility is to use the known 4B / 5B coding scheme (used in the FDDI). In this scheme, each 8-bit character is divided into two 4-bit parts, and then each part is encoded into a 5-bit codeword over the serial link 41 is sent.

10 Figure 4 illustrates a coding table of 4-bit data in 5-bit serial codewords. 4B / 5B is a redundant coding scheme, ie that there are more codewords than data words. Thus, some of the unused or redundant serial codewords may be used to convey control information.

11 is a table of 15 possible coded control codewords that can be used to transmit the time frame delimiter (TFD) over the serial link. The TFD transfer is completely transparent to the data transfer, so it can be sent non-destructively in the middle of the data packet transfer.

When the communication links 41 SONET / SDH, the time frame delimiter can not be embedded as redundant serial codewords because SONET / SDH serial coding relies on scrambling without redundancy. Thus, the TFD is realized using the following SONET / SDH frame control fields: Transport Organization Data (TOH) and Path Organization Data (POH). Although the SONET / SDH uses a 125 microsecond frame, it is noted that at the moment it can not be used directly in accordance with the present invention, since the SONET / SDH frames are not globally synchronized and also not synchronized with the UTC are. In contrast, if the SONET / SDH frames are globally synchronized, the SONET / SDH can be used compatible with the present invention.

1.5 The input port

As in 12 is shown has the input port 30 three parts: a serial receiver 31 , a routing control unit 35 and separate queues to the output ports 36 , The serial receiver 31 transmits the data packets and the time frame delimiter to the routing control unit 35 ,

The routing control unit 35 is composed of a central processing unit (CPU), a random access memory (RAM) for storing the data packet, a read only memory (ROM) for storing the routing control unit processing program, and the routing table used therefor; to determine the output port to which the incoming data packet should be switched.

As in 13 is illustrated, the header of the incoming data packet contains a virtual pipe ID, PID 35C which is used in the routing table 35D the address 35E the queue 36 to look up to which the incoming data packet should be transmitted. Before the package in his queue 36 is transmitted, the arrival time (ToA) 35T as in the 15A and 15B is attached to the packet header. The ToA 35T is from the planning control unit 45 of the starting sport 40 used in calculating the forwarding time from the output port and is in 16 shown.

The Data packet can be various formats such as internet protocol, Version 4 (IPv4), Internet Protocol, Version 6 (IPv6), Cells of the asynchronous transfer mode (ATM cells) etc. to have. The PID of the data packets can determined by one of the following: an internet protocol address (IP address), an identifier of a virtual wire of the asynchronous transfer mode (ATM), a virtual route identifier (VCI / VPI), addresses of the Internet protocol, Version 6 (IPv6), Internet MPLS labels (Internet Multi-Protocol Label Switching Addresses or Internet Multi-Protocol Tag Switching Addresses) and IEEE 802 MAC Address (IEEE 802 Media Access Control Address), etc.

14 illustrates the flowchart for the routing control unit 35B executed processing program for the routing control unit 35 , The program responds to two basic events from the serial receiver 31 out

12 : the received time frame delimiter TFD in step 35-0 and the receive data packet in step 35-02 , Upon receipt of a TFD, the routing control unit calculates 35 in step 35-03 the value of the arrival time (ToA) 35T which is attached to the incoming data packets. For this calculation, it uses a constant Dconst, which is the time difference between the tick point of the common time reference (CTR) 002 and receiving the TFD at time t2 (which is at an adjacent switch by the CTR 002 is generated at this node). This time difference is caused by the fact that the delay from the serial transmitter 49 to the serial receiver 31 is not an integer number of timeframes. If the data packet in step 35-02 is received, the routing control unit performs 35B three operations off, as in step 35-04 setting out the ToA, looking up the address of the queue 36 using the PID and storing the data packet in this queue 36 ,

1.6 The switching fabric

There are several ways to implement a switching fabric. However, the switching fabric for the present invention is peripheral and thus will be described only briefly. The main feature that should ensure the switching fabric is that the priority for packets for which the priority bit P ( 35P in the 15A and 15B ) is set high (ie reserved traffic) in which output port is switched to a constant limited delay measured in time frames.

In accordance with the present invention, in which the packets in the input ports already queued to their respective output ports this is possible. Thereupon the delay can be using the Clos set in the time domain (see J.Y. Hui, "Switching and Traffic Theory for Integrated Broadband Networks ", Page 65) by two time frames, one time frame at the input port and a timeline to over the switching fabric come to be limited. There can be more Realizations such as based on a shared Bus with a round-robin service of the data packets with high priority or one Crossbar switch, to be used.

One further possible Mediation design is a shared memory that has a deterministic delay limit from an input port to an output port. packet switches with shared memories are commercially different Suppliers, eg. From MMC Networks Inc. (Santa Clara, Calif.).

The 15A and 15B illustrate data packages without or with an attached timestamp.

1.7 The starting port

The starting port 40 is in 16 illustrating a scheduler with a send buffer 45 and with a serial transmitter 49 (as described hereinbefore). The planning control unit 45 maintains a map of each of the data packets between the associated ones Time of arrival (ToA) and associated forwarding time from the output port via the serial transmitter 49 out. The forwarding time is compared to the common time reference (CTR) 002 certainly.

Three output port configurations are illustrated here: a dual-buffer scheduler, as described in US Pat 17 and 18 is shown, a general planning control unit, as in the 19 . 20 and 21 is shown, and a general planning control unit with timestamp, as shown in the 22 and 23 is shown.

The dual buffer scheduling controller 46 as stated in the block diagram 17 and in the schedule 18 is constructed of a central processing unit (CPU), a random access memory (RAM) for storing the data packet, and a read only memory (ROM) for storing the controller processing program. Each time frame, as defined by the common time reference 002 is considered to be an even tick point or an odd tick point. The determination of an even tick point versus an odd tick point is relative to the beginning of a time cycle. In the preferred embodiment, the first time frame of a time cycle is determined to be an odd tick point, if the second time frame of the time cycle is determined to be an even tick point, the third time frame of the time cycle is determined to be an odd tick point, and so on, with the determination of even versus odd Tick point, as shown, changes for the duration of the time cycle. In an alternative embodiment, the first time frame of a time cycle is determined to be an even tick point, if the second time frame of the time cycle is determined to be an odd tick point, the third time frame is determined to be an even tick point, and so forth, determining the even tick point versus an odd tick point as shown changes for the duration of the time cycle. The actual sequence of even tick points versus odd tick points of time frames in a time cycle can be arbitrarily started without loss of generality.

The dual buffer scheduling controller 46 works in the following way. From the switching fabric 50 come over the transmission line 51 Data packets. If the priority bit 35P is activated (ie reserved traffic), the packet (during odd tick points of the common time reference) is sent via the packet DMUX (demultiplexer). 51S over the transmission line 51-1 to the buffer Ba and during even tick points of the common time reference over the transmission link 51-2 mediated to the buffer Bb. Data packets in which the priority bit 35P is not activated (ie unreserved traffic), are sent via the packet DMUX (demultiplexer) 51S over the transmission line 51-3 to the buffer Bc of "best effort" mediates. The transmit buffer selection operation is performed by the selection signal 46A Controlling the Dual Buffer Scheduling Controller with the Packet DMUX (Demultiplexer) 51S combines.

The data packets are routed through the line during odd tick points of the common time reference 46-2 from the buffer Bb and during even tick points of the common time reference over the connection 46-1 from the buffer Ba via the packet MUX (multiplexer) 47S and the connection 47C in 17 to the serial transmitter 49 forwarded. If the buffer Bb is empty during odd tick points of the common time reference, data packets are forwarded from the best effort buffer Bc to the serial transmitter. If the buffer Ba is empty during even tick points of the common time reference, data packets are forwarded from the best effort buffer Bc to the serial transmitter. The transmit buffer selection operation is performed by the selection signal 46B controlled by the double buffer scheduler 46 with the packet MUX (Multiplexer) 47S combines.

In the 19 . 20 and 21 is the operation of a more general planning control unit 45 described a send buffer 45C and a selection buffer controller 45D contains. The data packet planning control unit 45A performs together with the select buffer controller 45D using the PID 35C and the data packet arrival time (ToA) 35T the map to determine the particular time frame at which a particular packet should be forwarded out of the egress port. The two control units 45A and 45D are constructed of a central processing unit (CPU), a random access memory (RAM) for storing the data and a read only memory (ROM) for storing the control unit processing program.

The data packets come over the transmission link 51 from the switching fabric 50 at. The data packets where the priority bit 35P is activated (ie reserved traffic), by the scheduling control unit 45A to one of the k send buffers 45C (B-1, B-2, ..., Bk). As in 5 is shown, each of the k buffers is determined to store packets forwarded in each of the k time frames in each time cycle.

In 20 the flowchart for the program executed by the scheduling controller is illustrated. When in step 45-03 the data package received from the fabric becomes the PID 35C used in the data packet header to, as in step 45-04 is specified, the forwarding parameter 45F in the forwarding table ( 45B out 19 ) look up. The following will be in step 45-05 by subtracting the arrival time ToA 35T from the common time reference CTR 002 and adding the forwarding parameter 45F the index i of the send buffer between B-1 and Bk 'is calculated and then the incoming data packet as in step 45-06 is specified to the transmit buffer Bi mediated.

Incoming data packets in which the priority bit 35P is not activated (ie unreserved traffic) are transmitted by the scheduling control unit via the connection 45 -be to the send buffer BE of "best effort".

21 illustrates the flowchart for the operation of the selection buffer controller 45D , The control unit 45D reacts in step 45-11 on the tick point 002 the common time reference (CTR) and incremented in step 45-12 transmit buffer index i (ie, i: = i + 1 mod k ', where k' is the number of scheduled traffic buffers) and sends a time frame delimiter TFD to the serial transmitter. If the transmission buffer Bi then, as in step 45-14 is not empty, send it in step 45-13 a data packet from the send buffer bi, while otherwise, as in step 45-15 is specified, sends a "best effort" data packet from the "best effort" buffer B-be.

The 22 and 23 illustrate a system with a scheduling controller, wherein each of the data packets comprises a header containing an associated timestamp. The timestamp is generated by an internet real-time protocol (Internet RTP) in which its data packet format is stored in 22 is illustrated. Alternatively, the timestamp may be defined by a predefined one of the exchanges 10 be generated in the system or the time stamp can be generated at a respective end station for inclusion in the respective original data packet.

23 illustrates the operation of the scheduling controller in the event that the packet header is timestamped 35TS contains. From the switching fabric 50 come over the transmission line 51 Data packets. Data packets in which the priority bit 35P is set (ie, reserved traffic), become one of the k send buffers by the scheduling controller 45C (B-1, B-2, ..., Bk). As in 5 is shown, each of the k buffers is determined to store packets forwarded in each of the k time frames in each time cycle. In 23 the flowchart for the program executed by the scheduling controller is illustrated. When in step 45-21 a data packet is received from the fabric, as in step 45-22 is specified, the PID 35C in the data packet header used in the routing table 45B the forwarding parameter 45F look up. The following will be in step 45-23 by subtracting the arrival time ToA 35T from the common time reference CTR 002 and adding the forwarding parameter 45F the index i of the send buffer between B-1 and Bk is calculated and then the incoming data packet as in step 45-24 is specified to the transmission buffer Bi mediated.

2 Time-based routing

In This variant of this invention is a package that at an input port a mediation arrives on the basis of (i) its position in the predefined time interval and (ii) the unique address of the incoming input port to an output port. Any mediation along a route from a source to a destination routes packets at periodic time intervals using the common time reference, to a goal further. The time interval duration may be longer than that for sending duration of a package.

24 shows a schematic description of a switch 10 , The mediation 10 is constructed of four components: a plurality of uniquely addressable input ports 30 (in 24 there are N such ports), a plurality of uniquely addressable output ports 40 (in 24 there are N such ports), a switching fabric 50 and a time receiver 20 Global Positioning System (GPS) with a GPS antenna 001 , The GPS time receiver provides a common time reference (CTR) for all input and output ports 002 ready. The common time reference is divided into time frames. Each of the time frames also consists of predefined positions so that the input port can uniquely identify the positions. The time and position at which a data packet arrives at the input port is provided by the routing control unit 35 in 12 used to determine the output port to which this incoming data packet should be switched.

In 24 each of the time frames t = i and t = i + 1 has four predefined positions: a, b, c and d. At each of the positions, a data packet can be stored. As will be explained below, the locations between the variable size data packets may be explicitly marked with position delimiters (PDs) or implicitly marked. The implied position in a time frame can either be by measuring time delays - this is suitable for transmitting ATM cells (cells of the asynchronous transfer mode) of a fixed size - or by arranging variable size data packets at the predefined positions in each of the time frames - wherein if the output port 40 If no data packet has to be sent at a predefined position, an empty data packet or null data packet should be sent.

25 shows an axis of the common time reference (CTR) 002 that is divided into time cycles. Each time cycle is divided into predefined frames. Each of the time frames has predefined positions: a, b, c, and d of either fixed size (in terms of time duration) or variable size (in terms of time duration), and thus the predefined position can either have fixed size data packets or variable size data packets ,

2.1 The time-based input port

The in 12 shown input port 30 has three parts: a serial receiver 31 , a time-based routing control unit 35 and separate queues 36 to the plurality of output ports 40 , The serial receiver 31 transmits data packets, time frame delimiter (TFD) and position delimiter (PD) to the time-scheduled routing control unit 35 ,

• 1. Der Parameter 35-1 in Tabelle 35D (27 und 29) – der Ausgangsport 40, zu dem das ankommende Datenpaket vermittelt werden sollte – dieser Parameter wird verwendet, um das Datenpaket zu der Warteschlange 36 zu vermitteln, die die vorderste zu dem entsprechenden Ausgangsport ist;
• 2. der Parameter 35-2 in Tabelle 35D (27 und 29) – der abgehende Zeitrahmen, in dem das Datenpaket aus dem Ausgangsport weitergeleitet wird – dieser Parameter wird an dem Datenpaket-Anfangsblock in 27 angebracht, und
• 3. der Parameter 35-3 in Tabelle 35D (27 und 29) – die Position in dem abgehenden Zeitrahmen, in dem das Datenpaket aus dem Ausgangsport weitergeleitet wird – dieser Parameter wird an dem Datenpaket-Anfangsblock in 28 angebracht.

The routing control unit is constructed of a central processing unit (CPU), a random access memory (RAM) for storing the data packets, and a read only memory (ROM) for storing the time-based routing control unit processing program the determination of the following parameters (see 35D in 26 . 27 and 29 ) a time-based routing table is used:

• 1. The parameter 35-1 in table 35D ( 27 and 29 ) - the output port 40 to which the incoming data packet should be switched - this parameter is used to route the data packet to the queue 36 to mediate, which is the foremost to the corresponding output port;
• 2. the parameter 35-2 in table 35D ( 27 and 29 ) - the outgoing time frame in which the data packet is forwarded from the output port - this parameter is entered at the data packet header in 27 appropriate, and
• 3. the parameter 35-3 in table 35D ( 27 and 29 ) - the position in the outgoing time frame in which the data packet is forwarded from the output port - this parameter is entered at the data packet header in 28 appropriate.

• 1. Lokale Zeit und Position unter Verwendung (1) des Zeitrahmens der Ankunft (ToA) 35T – des Zeitrahmens unter Verwendung der gemeinsamen Zeitreferenz 002 und (2) des Positionswerts 35P in diesem Zeitrahmen, gemessen durch den Positionszähler 35PC. Dies ist in 12 gezeigt.
• 2. Den Zeitstempel 35TS und die Position 35PC unter Verwendung (1) des Zeitstempels 35TS in dem Datenpaket-Anfangsblock in 28B und (2) des Positionswerts 35P in diesem Zeitrahmen, gemessen durch den Positionszähler 35PC
• 3. Den Zeitstempel PID (der in den Paketanfangsblöcken in 28 gezeigt ist) und die Position 35PC durch (1) den Zeitstempel 35TS in dem Datenpaket-Anfangsblock in 28B, (2) die ID der virtuellen Pipe (PID) 35C in dem Datenpaket-Anfangsblock in 28B (die virtuelle Pipe wird am Ende dieser Beschreibung ausführlich diskutiert) und (3) den Positionswert 35P in diesem Zeitrahmen, gemessen durch den Positionszähler 35PC. Dies ist in 29 gezeigt.

The time-based routing control unit 35B determines the entry for the time-based routing table 35D in the 27 and 29 in different ways like:

• 1. Local time and position using (1) the time frame of arrival (ToA) 35T The time frame using the common time reference 002 and (2) the position value 35P in this time frame, measured by the position counter 35PC , This is in 12 shown.
• 2. The timestamp 35TS and the position 35PC using (1) the timestamp 35TS in the data packet header in 28B and (2) of the position value 35P in this time frame, measured by the position counter 35PC
• 3. The timestamp PID (which is in the packet headers in 28 shown) and the position 35PC by (1) the time stamp 35TS in the data packet header in 28B , (2) the virtual pipe ID (PID) 35C in the data packet header in 28B (the virtual pipe is discussed in detail at the end of this description) and (3) the position value 35P in this time frame, measured by the position counter 35PC , This is in 29 shown.

The data packets, see z. B. 28 , various formats, such as Internet Protocol Version 4 (IPv4), Internet Protocol, Version 6 (IPv6), may have asynchronous transfer mode (ATM) cells. The PID 35C The data packets may be determined by one of the following: an Internet Protocol (IP) address, an Asynchronous Transfer Mode (ATM), a Virtual Line Identifier, and a Virtual Route Identifier (VCI / VPI), an Internet Protocol address, Version 6 (IPv6), Internet Multi-Protocol Label Switching Addresses (Internet Multi-Protocol Tag Switching Addresses), and IEEE 802 (IEEE 802) MAC Addressing. 802 Media Access Control address).

The timestamp 35TS in the packet header in 28B can be generated by an application using the Internet Real-Time Protocol (RTP) and is also used in the ITU-T H.323 standard. Such data packets use the in 22 shown format. Alternatively, the time stamp may be generated by a predefined one of the exchanges in the system, or alternatively the time stamp may be generated at the respective end node for inclusion in the respective original data packet.

• 1. Empfange den Zeitrahmenbegrenzer TFD 35-1 – wie in 135-04 aus 30 angegeben ist, setzt die Leitweglenkungs-Steuereinheit in Reaktion auf dieses Ereignis den Positionszähler zurück (35P: = 0 in 135-04 aus 30) und berechnet den Wert des Ankunftszeitrahmens (ToA) 35T. Für diese Berechnung verwendet sie eine Konstante Dconst, die die Zeitdifferenz zwischen dem Tickpunkt der gemeinsamen Zeitreferenz (CTR) 002 und dem Empfang des TFD zum Zeitpunkt t2 ist (es wird angemerkt, dass der TFD in einer angrenzenden Vermittlung durch die CTR 002 an diesem Knoten erzeugt wurde). Diese Zeitdifferenz wird durch die Tatsache verursacht, dass die Verzögerung von dem seriellen Sender 49 zu dem seriellen Empfänger 31 keine ganze Anzahl von Zeitrahmen ist.
• 2. Empfange den Positionsbegrenzer PD 135-02 – inkrementiert in Reaktion auf dieses Ereignis den Positionszähler, 35P := 35P + 1, 135-05 aus 30.
• 3. Empfange das Datenpaket 135-03 – wie in 135-06 aus 30 gezeigt ist, werden in Reaktion auf dieses Ereignis drei Operationen ausgeführt: (1) an dem Paketanfangsblock wird der Parameter 35-2 des abgehenden Zeitrahmens angebracht, (2) an dem Paketanfangsblock wird der Parameter 35-3 der Position in dem abgehenden Zeitrahmen angebracht und (3) das Datenpaket wird unter Verwendung des Ausgangsportparameters 35-1 in Tabelle 35D in den 27 und 29 in der Warteschlange 36 gespeichert.

30 is a detailed description of the time-based routing control unit 35B executed program. The program responds to three events from the serial receiver 31 and on the position value 35P within this timeframe, measured by the posi tion counter 35PC , The time-based routing controller program 30 which contains the three parameters in Table 35D 27 and 29 used to which this incoming packet is assigned works as follows:

• 1. Receive the time frame delimiter TFD 35-1 - as in 135-04 out 30 is specified, the routing control unit resets the position counter (35P: = 0 in 135-04) in response to this event 30 ) and calculates the value of the arrival time frame (ToA) 35T. For this calculation, it uses a constant Dconst, which is the time difference between the tick point of the common time reference (CTR) 002 and receiving the TFD at time t2 (note that the TFD is in an adjacent switch by the CTR 002 was generated at this node). This time difference is caused by the fact that the delay from the serial transmitter 49 to the serial receiver 31 is not an integer number of timeframes.
• 2. Receive the position limiter PD 135-02 Increments the position counter in response to this event, 35P : = 35P + 1, 135-05 out 30 ,
• 3. Receive the data packet 135-03 - as in 135-06 out 30 3, three operations are performed in response to this event: (1) at the packet header becomes the parameter 35-2 the outgoing time frame is attached, (2) at the packet header becomes the parameter 35-3 the position in the outgoing time frame and (3) the data packet is determined using the output port parameter 35-1 in Table 35D in the 27 and 29 in the queue 36 saved.

2.2 The time-based output port

The output is in 16 and has two parts, a scheduling control unit with a transmit buffer 45 and a serial transmitter as described above 49 , The data packet planning control unit 45A in 19 transmits the data packet to the send buffer containing a read-write memory (RAM) as described below. 45C is.

In the 19 . 31 and 32 becomes the operation of the data packet planning control unit 45 described a send buffer 45C and a selection buffer controller 45D contains. The planning control unit 45A performs together with the select buffer controller 45D using the two parameters 35-2 and 35-3 through the routing control unit 35B attached to the data packet, the figure out. Both control units are constructed of a central processing unit (CPU), a random access memory (RAM) for storing the data, and a read only memory (ROM) for storing the control unit processing program.

From the switching fabric 50 over the transmission line 51 incoming data packets containing their priority bit 35P is activated (ie reserved traffic), by the data packet scheduling control unit 45A to one of the k 'send buffers 45C conveys: B-1, B-2, ..., Bk '(if k' = k, where k is the time cycle size measured in time frames, there is a special case). Each of the k 'buffers is destined to store a packet which is in a cyclically repeating order in each of the like 5 and 6 shown k time frame is forwarded in each time cycle. The actual program executed by the data packet scheduling control unit is in 31 described. If from the fabric 145-01 a data packet is received (in 31 ), as in 31 in 145-02 is specified, the two parameters 35-2 and 35-3 in the data packet header to determine in which of the transmit buffers between B-1 and Bk 'and at which position this data packet is to be stored.

Incoming data packets containing their priority bit 35P , please refer 28 is not activated (ie unreserved traffic) are switched by the data packet scheduler control unit over the connection 45-be to the best effort send buffer BE.

32 shows the operation of the selection buffer control unit 45D that, as in 145-11 ( 32 ) is indicated on the tick point 002 the common time reference (CTR) responds. Consequently, the selection buffer control unit increments the send buffer index i 145-12 (ie i: = i + 1 mod k 'where k' is the number of buffers for scheduled traffic) sends a time frame delimiter TFD 47A to the serial transmitter 145-12 and reset the position pointer to one, p: = 1, 145-12 , While the send buffer Bi is not empty then 145-13 , send it as in 145-14 . 145-15 and 145-16 otherwise, if the send buffer Bi is empty, as in 145-17 from the best effort buffer B to the end of the time frame (until the next tick point of the CTR 002) or until the buffer BE becomes empty, sends best effort data packets.

If the send buffer Bi is not empty 145-13 , sends the selection buffer control unit, as in 145-14 is specified, data packets from all non-empty predefined positions in this buffer. After sending a data packet or if the position p in buffer Bi is empty, 145-15 , sends the selection buffer control unit, as in 145-16 is specified, a position limiter (PD) 47B to that serial transmitter and increments the position counter p: = p + 1.

3 traffic integration

3.1 The integrated input port

The in 12 shown input port 30 has three parts: serial receiver 31 , Routing control unit 35 and separate queues to the output ports 36 , The serial receiver 31 transmits the data packets and the time frame delimiter to the routing control unit 35 ,

The routing control unit is constructed of a central processing unit (CPU), a random access memory (RAM) for storing the data packet and a read only memory (ROM) for storing the routing control unit processing program, wherein for determining of the output port to which the incoming data packet should be switched, a routing table is used. The header of the incoming data packet contains a virtual pipe ID - the PID 35C in 34 which is used in the routing table 35D the address 35E in the queue, to which the incoming data packet in its queue 36 should be transferred, look up. Before the package in his queue 36 is transmitted, the arrival time (ToA) 35T in 34 attached to the package header. The ToA 35T is from the planning control unit 45 in 16 used in calculating the forwarding time from the output port.

The Data packet may have various formats such as internet protocol, Version 4, (IPv4), Internet Protocol, Version 6, (IPv6), Cells the asynchronous transfer mode (ATM cells). The data packet PID may be through one of the following be determined: the Internet Protocol address (IP address), a asynchronous transfer mode (ATM), a virtual service identifier and a virtual route identifier (VCI / VPI), Internet Protocol 6 addresses (IPv6), Internet MPLS (Internet Multi-Protocol Label Switching Addresses) labels or Internet multi-protocol tag switching addresses) and the IEEE 802 MAC address (IEEE 802 Media Access Control Address).

35 is a table for defining two bits, P1 and P2, in the packet headers in 34 , The two bits classify three types of data packets: P1 / P2 are "00", constant bit rate (CBR) traffic; P1, P2 are 01, variable bit rate (VBR) traffic; and P1, P2 are "10", "Best Effort" traffic. As in 36 is shown, the above classification of the by the routing control unit 35B executed program to determine in which of the three parts of the queue to the in 33 shown output port 36 the data packet should be taught.

36 is a detailed description of the routing control unit 35B executed program. The program responds to two basic events from the serial receiver 31 : Receive time frame delimiter TFD 235-01 and receive data packet 235-02 , After receiving a TFD, 235-03 , the routing control unit calculates the value of time of arrival (ToA) 35T in 34 which is attached to the incoming data packets. For this calculation he uses a constant, Dconst, which is the time difference between the tick point of the common time reference (CTR) 002 and the reception of the TFD at time t2 (note that the TFD is sent by the CTR 002 was created at this node in an adjacent switch). This time difference is caused by the fact that the delay from the serial transmitter 49 to the serial receiver 31 is not an integer number of timeframes. When a data packet is received, 235-02 , performs the routing control unit 35B three operations out, 235-04 in 36 : attach the ToA, hit the address of the queue using the PID 36 and save the data packet in the queue 36 in the output port 37 while in the header in 34 P1 / P2 is used to determine in which part, CBR / VBR / best effort, to store this queue the incoming data packet.

3.2 Integrated Planning Control Unit in the output port

In the 37 - 39 is the operation of a more general planning control unit 45 described that, as in 37 is shown, a planning and replanning control unit 145A , a send buffer 145C and a selection buffer and stowage control unit 145D contains. The planning and replanning control unit 145A performs together with the select buffer controller 145D the mapping of the data packet into the time frame. The mapping is done in the scheduling and replanning controller using the PID 35C and the data packet arrival time (ToA) 35T to determine the particular time frame in which the particular packet should be forwarded from the output port. Details are shown below. Both control units 145A and 145D are constructed of a central processing unit (CPU), a random access memory (RAM) for storing the data and a read only memory (ROM) for storing the control unit processing program.

In the in 37 more general configuration shown will be packets of data over the transmission link 51 from the switching fabric 50 arrive where their priority bits 35P (P1 / P2) are either "00" or "01" (ie reserved CBR traffic) by the control unit 145A to one of the k send buffers in 145C B-1, B-2, ..., Bk. Each of the k buffers is intended to store a packet that is in each of the k time frames in each time cycle included in the 5 and 6 are redirected. Another possible operation is to separately map the incoming packets to each of the time frames of a supercycle. When a super-cycle map is realized, there are 145C k * l transmit buffers: B-1, B-2, ..., Bk * 1, ie, k buffers for each of the 1 cycles of a supercycle.

In 38 the actual program executed by the scheduling and replanning control unit is described. If, as in 145-03 is specified, either via the transmission path 51 from the fabric or over the connection 45R from the selection buffer and stowage control unit 145D a data packet is received that will 35C . 35T and 35P in the data packet header, as in 145-04 is specified, used the forwarding parameter 45F in the routing table 145B look up. Below is in 145-05 by subtracting the arrival time ToA 35T from the common time reference CTR 002 and adding the forwarding parameter 45F the index i of the send buffer is calculated between B-1 and Bk and then, as in 145-05 specified, the incoming data packet to the transmission buffer Bi mediates.

Incoming data packets containing their priority bits 35P P1 / P2 are either "10" (ie, unreserved traffic or "best effort") or "11" (ie, re-scheduled packet) are sent by the scheduling and replanning controller via connection 45-be to the transmit buffer BE " of the best effort "mediates.

• 1. Sende Zeitrahmenbegrenzer-Steuersignal (TFD-Steuersignal) 47A zu dem seriellen Sender 49;
• 2. leite alle ungesendeten Pakete im Sendepuffer B-f, in denen das Feld 35P in ihrem Anfangsblock nicht "11" ist, d. h., die kein neu geplantes Paket sind, zur Steuereinheit 145A zurück weiter und stelle das Feld 35P auf "11", d. h. auf ein neu geplantes Paket, ein (es wird angemerkt, dass ein Paket nur einmal neu geplant werden kann);
• 3. falls die Anzahl der Datenpakete, die neu geplant werden müssen, eine vordefinierte Anzahl, angenommen n, übersteigt, wähle zufällige n Datenpakete aus und sende sie über die Verbindung 45R zu der Paketplanungs- und Neuplanungssteuereinheit 145A zurück und verwirf den Rest der Pakete.
• 4. Inkrementiere den Sendepufterindex f (d. h. f: = f + 1 mod k', wobei k' die Anzahl der Puffer für geplanten Verkehr ist). Falls der CBR-Teil des Puffers B-f nicht leer ist, 245-12, sendet sie, wie in 245-16 und 145-13 angegeben ist, ein Datenpakt vom Sendepuffer B-f, zunächst CBR-Pakete und daraufhin VBR-Pakete, während sie andernfalls, wie in 245-14 angegeben ist, von dem Puffer B-E "des besten Bemühens" ein Datenpaket "des besten Bemühens" sendet.

39 shows the operation of the selection buffer and stowage control unit 145B pointing to the tick point 002 the common time reference (CTR) responds. When the CTR signal is received, 245-11 , be through the control unit 45D in 245-15 the following operations are performed:

• 1. Send time frame delimiter control signal (TFD control signal) 47A to the serial transmitter 49 ;
• 2. forward all unsent packets in the send buffer Bf containing the field 35P in its header is not "11", ie, which are not a newly scheduled packet, to the control unit 145A go ahead and set the field 35P to "11", ie a newly scheduled package (it is noted that a package can only be rescheduled once);
• 3. if the number of data packets to be rescheduled exceeds a predefined number, say n, select random n data packets and send them over the connection 45R to the package planning and replanning control unit 145A back and discard the rest of the packages.
• 4. Increment the transmit buffer index f (ie f: = f + 1 mod k ', where k' is the number of buffers for scheduled traffic). If the CBR part of the buffer Bf is not empty, 245-12 , send it as in 245-16 and 145-13 is specified, a data packet from the transmit buffer Bf, first CBR packets and then VBR packets, while otherwise, as in 245-14 from which "Best Effort" buffer BE sends a "Best Effort" data packet.

4 Operation with variable transmission links delay

The The present invention further relates to a system and a method for sending and forwarding packets via Packet switching network in which some of its communication links dynamically changing Have delays. Such changes the transmission link delay thereby caused to have a mobile switching node (e.g. B. satellites).

40 illustrates a virtual pipe 25 from the starting port 40 switch A via switches B and C. This virtual pipe ends at the output port 40 of the node D. The virtual pipe 25 transmits data packets from at least one source to at least one destination. The communication connection in 40 , which connects the switch B to the switch C, may have a delay that changes in time with a defined delay. Such communication links can be found in various network architectures, such as mobile wireless networks, satellite networks, and self-healing SONET rings. In satellite networks, the delay between a satellite in space and a base station on the earth changes in the following way. As the satellite appears above the horizon and moves to the base station, the delay initially decreases while the delay increases as the satellite moves away from the base station until it disappears below the horizon.

The 41 and 42 describe the delay changes on the link between nodes B and C as projected onto a common time reference (CTR) detailed in the above 4 . 5 and 6 dis is cut. 41 describes a communication link between nodes B and C, with the arrival time to node C decreasing, ie, the delay between nodes B and C becoming shorter. In 41 is the delay of the data packet 0a longer than the data packet 1b , is the delay of the data packet 1b longer than the data packet 2c etc. 42 describes a communication link between nodes B and C, with the arrival time to node C increasing, ie, the delay between nodes B and C becoming longer. In 42 is the delay of the data packet 0a shorter than the data packet 1b , is the delay of the data packet 1b shorter than the data packet 2c etc. A complete description of the above resynchronization operation is part of the following in the 46 . 47 . 48 . 50 and 51 described output operation.

The 43 . 44 and 45 describe delay changes that are a consequence of routing successive data packets on alternative routes or routes in the network. 43 shows two (hereinafter defined) virtual pipes, p (from node A to B to C to D and E) and p '(from node A to B' to D 'and to E), with the requirement that data packets be independent from which virtual pipe, p or p ', they were forwarded to are forwarded out of node E at the same predefined time. 44 shows the scenario in which the data packets arrive in the virtual pipe p 'at node E before the time they would have arrived at node E in the virtual pipe p. Consequently, the data packets on the path p ', as in 44 is shown, delayed and forwarded in time as if they had arrived in the virtual pipe p. 45 shows how such resynchronization can be achieved using a resync buffer at node E. In 45 if a data packet emerges from the virtual pipe p ', 1p ' , in a resync buffer 10R so that this data packet, if it is out of this buffer 2p ' exits, is passed from the output of the node E, as if this packet would be forwarded in the virtual pipe p. A complete description of the above resynchronization operation as part of the output port operation will be described in the following 16 . 46 . 47 . 48 . 49 . 50 and 51 described.

The starting port 40 is in 16 illustrates and includes a scheduling controller with a transmit buffer 45 and with a serial transmitter 49 (as described earlier). The planning control unit 45 maintains an mapping of each of the data packets between the associated respective time of arrival (ToA) and an associated routing time out of the output port via the serial transmitter 49 out. The forwarding time is compared to the common time reference (CTR) 002 certainly.

• 1. eine Verzögerungsanalyse- und Planungssteuereinheit 245A, die ferner eine Weiterleitungstabelle 245B umfasst,
• 2. einen Sendepuffer 245C, der üblicherweise als ein Schreib-Lese-Speicher (RAM) realisiert ist, und
• 3. eine Auswahlpuffer-Steuereinheit 245D, die Datenpakete zu dem seriellen Sender weiterleitet.

The scheduling controller and the send buffer 45 have different modes of operation in the 46 . 47 . 48 . 49 . 50 and 51 are described. The various modes correspond to some of the possible changes in the communication link delay as described in US Pat 40 - 45 were discussed. The scheduling controller and the send buffer 45 in 46 contain three parts:

• 1. a delay analysis and planning control unit 245A , which further comprises a forwarding table 245B includes,
• 2. a send buffer 245C commonly implemented as a random access memory (RAM), and
• 3. a selection buffer control unit 245D which forwards data packets to the serial transmitter.

The delay analysis and planning control unit 245A performs together with the select buffer controller 245D using the PID 35C , the time stamp 35TS and the data packet arrival time (ToA) 35T the map to determine the particular time frame at which a particular packet should be forwarded out of the egress port. Both control units 245A and 245D are constructed of a central processing unit (CPU), a random access memory (RAM) for storing the data and a read only memory (ROM) for storing the control unit processing program.

From the switching fabric 50 come over the transmission line 51 Data packets. Data packets where the priority bit 35P is activated (ie reserved traffic) are determined by the delay analysis and scheduling control unit 245A to one of the l · k send buffers 45C (B-1, B-2, ..., Bl · k). As in the 5 and 6 is shown, each of the lxk buffers is destined to store packets forwarded in each of the lxk time frames in each supercycle. The presence of 1 × k transmit buffers allows the delay analysis and scheduling control unit 245A Plans data packets in a wide range of delay changes. If the supercycle is one second, the scheduling capability is the delay analysis and scheduling controller 245A up to a second. However, this is an extreme case where, in most practical scenarios, even in the case of transmission paths with changing delay, the scheduling requirements are only a small number of time frames.

The send buffer 245C contains an additional buffer BE for "best effort" data packets. The priority bit 35P in the data packets "best effort" is not enabled, and this is how the delay analysis and scheduling engine determines that such data packets should be stored in the "best effort" buffer. The "best effort" data packets each time become the serial transmitter 49 forwarded if there are no more scheduled data packets (if the priority bit 35P is activated).

47 illustrates the flowchart for the operation of the selection buffer controller 45D , In step 245-11 the control unit reacts 45D on the tick point 002 the common time reference (CTR) and then incremented in step 345-12 transmit buffer index i (ie i: = i + 1 mod l '* k', where l '* k' is the number of buffers for scheduled traffic) and sends a time frame delimiter TFD to the serial transmitter 49 , If the transmission buffer Bi in step 345-13 is not empty, he sends, as in step 345-14 is specified, a data packet from the transmit buffer Bi, while otherwise, as in step 345-15 is specified, sends a "best effort" data packet from the "best effort" buffer BE.

• I. 49 – der wie in den 40-42 beschriebene Fall kontinuierlicher Verzögerungsänderungen, wie er in 45-051 angegeben ist:
• 1. Es sei <s1, s2, s3, ..., sj> die Menge der Zeitrahmen mit einer PID = p, die sich in jedem Superzyklus wiederholt, wie in der Weiterleitungstabelle 245B bei dem Eintrag p angegeben ist.
• 2. Die Steuereinheit 245A durchsucht die Menge <s1, s2, s3, ..., sj>, um den ersten möglichen Zeitrahmen si zu bestimmen, der nach ToA 35T) + CONST auftritt (wobei CONST eine konstante Schranke an die Verzögerung über die Switching Fabric ist, und
• 3. si ist der Zeitrahmen, zu dem das Datenpaket für die Sendung über den seriellen Sender geplant wird – und i ist der Indexsendepuffer B-i.

In 48 the flowchart for the program executed by the data analysis and scheduling control unit is illustrated. The main task of the program is to compute the index i of the transmit buffer Bi between B-1 and B-l '* k' in step 245-05 , To perform the calculation in step 245-05 There are several possible methods, depending on the type of delay changes that can occur on the communication links. In the 49 and 50 three possible calculation methods are described:

• I. 49 - like in the 40 - 42 described case of continuous delay changes, as described in US Pat 45-051 is indicated:
• 1. Let <s1, s2, s3, ..., sj> be the set of time frames with a PID = p that repeats in each supercycle, as in the routing table 245B at the entry p is specified.
• 2. The control unit 245A search the set <s1, s2, s3, ..., sj> to determine the first possible time frame si, which after ToA 35T ) + CONST (where CONST is a constant bound on the delay across the switching fabric, and
• 3. si is the time frame at which the data packet for transmission is scheduled via the serial transmitter - and i is the index transmission buffer Bi.

• II. 50 – der Fall des Mehrfachwegs mit Resynchronisation, wie er in den 43, 44 und 45 beschrieben ist: Wenn in Schritt 245-03 das Datenpaket von der Fabric empfangen wird, wird die PID 35C in dem Datenpaket-Anfangsblock verwendet, um den Resynchronisationsparameter 45R in der Weiterleitungstabelle (245B aus 46, wie in Schritt 45-151 angegeben ist, und daraufhin in Schritt 45-152: Berechne den Index i des Sendepuffers 245C: i = [(ToA 35T) + 45A] mod l'·k' (wobei l'·k' die Anzahl der Puffer für geplanten Verkehr ist) nachzuschlagen. Im Fall zweier virtueller Pipes p und p', wobei eine eine Alternative zu der anderen ist, wie in den 43, 44 und 45 gezeigt ist, wird die obige Resynchronisation an beiden Enden benötigt. Genauer wird die Resynchronisation, wie in den 43, 44 und 45 gezeigt ist, sowohl am Knoten A als auch am Knoten E benötigt.
• III. 51 – der Fall, wenn in dem Paketanfangsblock ein Zeitstempel verwendet wird, 15 und 22: Wenn in Schritt 45-03 das Datenpaket von der Fabric empfangen wird, wird die PID 35C in dem Datenpaket-Anfangsblock, wie in Schritt 45-251 angegeben ist, verwendet, um den Weiterleitungsparameter 45F in der Weiterleitungstabelle (245B aus 46) nachzuschlagen und daraufhin in Schritt 45-252 den Index i des Sendepuffers 245C zu berechnen: i = ((Zeitstempel 35TS) + 45F] mod l'·k' (wobei l'·k' die Anzahl der Puffer für geplanten Verkehr ist).

The set <s1, s2, s3, ..., sj> forms a plurality of time frames in which a data packet for transmission from the output port of a switch can be scheduled.

• II. 50 - The case of multiple path with resynchronization, as in the 43 . 44 and 45 is described: when in step 245-03 the data packet received by the fabric becomes the PID 35C in the data packet header uses the resync parameter 45R in the forwarding table ( 245B out 46 as in step 45-151 is specified, and then in step 45-152 : Compute the index i of the send buffer 245C : i = [(ToA 35T ) + 45A ] mod l '* k' (where l '* k' is the number of scheduled traffic buffers). In the case of two virtual pipes p and p ', where one is an alternative to the other, as in the 43 . 44 and 45 is shown, the above resynchronization is needed at both ends. Specifically, the resynchronization, as in the 43 . 44 and 45 is shown required at both node A and node E.
• III. 51 The case when a timestamp is used in the packet header, 15 and 22 : When in step 45-03 the data packet received by the fabric becomes the PID 35C in the data packet header, as in step 45-251 is specified to the forwarding parameter 45F in the forwarding table ( 245B out 46 ) look up and then in step 45-252 the index i of the send buffer 245C to calculate: i = ((timestamp 35TS ) + 45F ] mod l '· k' (where l '* k' is the number of buffers for scheduled traffic).

5 monitoring, Cell flow control and accounting

Further The present invention relates to a system and a Procedure for The supervision, Cell flow control and accounting of the shipment and forwarding of data packets via a packet-switched network. The network's mediations hold one common time reference, either from an external time reference Source (such as from a GPS global Positioning system) is obtained or generated internally and is distributed.

The in 16 illustrated output port 40 includes a scheduling controller with a send buffer 45 , with a serial transmitter (as described above) 49 and with the monitoring and cell power control units. The planning control unit 45 maintains an mapping of each of the data packets between the associated respective time of arrival (ToA) and an associated routing time out of the output port via the serial transmitter 49 out. The forwarding time is compared to the common time reference (CTR) 002 certainly.

In the 19 - 21 was previously the operation of a general planning control unit 45 described a send buffer 45C and an off optionally buffer controller 45D contains. The data packet planning control unit 45A performs together with the select buffer controller 45D using the PID 45C and the data packet arrival time (ToA) 35T the mapping to determine the particular timeframe at which a particular packet should be forwarded out of the egress port. Both control units 45A and 45D are constructed of a central processing unit (CPU), a random access memory (RAM) for storing the data and a read only memory (ROM) for storing the control unit processing program.

5.1 The monitoring and cell power control units

• 1. die Verzögerungsüberwachungs-Steuereinheit 65D – zur Sicherstellung des richtigen Zeitgebungsverhaltens durch die PID = p (52, 53).
• 2. die Zellstromkontrollen- und Laststeuereinheit 65P – zur Sicherstellung der richtigen Kapazitätsnutzung durch die PID = p (54, 55).

The monitoring and cell power control units 65 ( 52 - 55 ) are both part of the entrance port in 12 as well as the starting sport in 16 , The monitoring and cell power control units 65 are of two basic types:

• 1. the delay monitoring control unit 65D To ensure the correct timing behavior by the PID = p ( 52 . 53 ).
• 2. The cell current control and load control unit 65P To ensure the correct use of capacity by the PID = p ( 54 . 55 ).

The two control units 65D and 65P are constructed of a central processing unit (CPU), a random access memory (RAM) for storing the data and a read only memory (ROM) for storing the control unit processing program.

5.2 The delay monitoring control unit 65D

• 1. Ein Datenpaket ist in zwei vordefinierten Verzögerungsparameterbereichen (siehe Kasten 65D-05): Zwischen den zwei Verzögerungsparametern: 65-par-L und 65-par-H, bei denen der Eintrag PID = p 35C in der Parametertabelle 65-Tab festgestellt wurde (siehe Kasten 65D-02). Genauer berechnet die Verzögerungsüberwachungs-Steuereinheit 65D die tatsächliche Verzögerung, die das Datenpaket bereits erfahren hat, 65-Del = Ankunftszeit 35T – Zeitstempel 35TS (siehe Kasten 65D-03), und vergleicht daraufhin, dass sie in dem vordefinierten Verzögerungsbereich liegt: (65-Del < 65-Par-L und 65-Del < 65-Par-H) (siehe Kasten 65D-04).
• 2. Das Datenpaket ist verspätet (siehe Kasten 65D-07): Seine Verzögerung ist größer als 65-par-H, d. h. 65-Del > 65-Par-H (siehe Kasten 65D-06), und
• 3. das Datenpaket ist vorzeitig (siehe Kasten 65D-08): seine Verzögerung ist kleiner als 65-par-L, d. h. 65-Del < 65-Par-L.

The 52 and 53 describe the operation of a delay monitoring control unit 65D , This control unit checks data packets in which its reserved priority bit, 35P , in its starting blocks is activated for three cases:

• 1. A data packet is in two predefined delay parameter ranges (see box 65D-05 ): Between the two delay parameters: 65-par-L and 65-par-H, where the entry PID = p 35C was found in the parameter table 65-Tab (see box 65D-02 ). More specifically, the delay monitoring control unit calculates 65D the actual delay that the data packet has already experienced, 65-Del = Arrival time 35T - Time stamp 35TS (see box 65D-03 ), and then compares that it lies in the predefined delay range: ( 65-Del < 65-Par -L and 65-Del < 65 Par-H ) (see box 65D-04 ).
• 2. The data packet is late (see box 65D-07 ): Its delay is greater than 65-par-H, ie 65-Del> 65-Par-H (see box 65D-06 ), and
• 3. the data packet is premature (see box 65D-08 ): its delay is less than 65-par-L, ie 65-Del <65-par-L.

The three cases have meaning for Ensuring the correct network operations and compliance the user service quality requirements (User QoS requirements). About that In addition, those collected by the delay monitoring control unit become Information reported to upper-layer protocols, the outside the scope of this invention.

5.3 The cell current control and load control unit 65P

• 1. Zellstromkontrolle – Erfassen einer PID = p, die ihre reservierte Kapazität übersteigt, und
• 2. Rechnungslegung – Aufzeichnen der tatsächlichen Kapazitätsnutzung der PID = p.

The 54 and 55 describe the operation of an alternative embodiment of the cell current control and load control unit 65P that checks and ensures that a data packet containing a reserved priority bit 35P (in its header), the predefined load of the PID does not exceed its virtual pipe = p. The operation of this control unit can be used for both:

• 1. cell current control - detecting a PID = p exceeding its reserved capacity, and
• 2. Accounting - recording the actual capacity utilization of the PID = p.

The two cases have meaning for ensuring proper network operations and compliance the user service quality requirements (User QoS requirements).

When a data packet is received (see box 65P-01 ), calculates the cell current control in the load control unit 65P using the load table 65L , which stores previous values of L (p), first the instantaneous load L (p) for the PID = p (see box 65P-02 ) by L (p): = L (p) + 1 (see box 65P-02 ). The load can be calculated in several ways: (i) per time frame the PID = p, (ii) per cycle PID = p or (iii) per super cycle the PID = p.

Subsequently, the cell current control and load control unit fails 65P using the PID = p 35C parameter 65-Par in the table 65 Tab to. If 65-par> L (p) (see box 65P-03 ), the data packet is then dropped (see box 65P-05 ) while the data packet is otherwise forwarded (see box 65P-04 ).

In the above two cases, the information of the load (L (p)) is recorded via the PID = p and reported usage accounting for use of PID = p at upper layer protocols. In addition, the cell flow control and load information is also used to ensure proper network operations and to meet user quality of service (user QoS) requirements. The information collected by the cell flow control and load control unit is reported to upper layer protocols that are outside the scope of this invention.

6 connection of a synchronous with a asynchronous switching network

Further The present invention relates to a system and a Method for sending and forwarding packets via heterogeneous packet switching network in which some of its exchanges synchronous and some switches are asynchronous. More precisely the invention ensures that the synchronous exchanges data packets forward in predefined time intervals, although they are at the synchronous negotiations with a relatively large but limited delay uncertainty Arrive. This delay uncertainty is the result of being asynchronous on the way of a data packet Exchanges before it reaches the synchronous switch.

• 1. Wie in 56 gezeigt ist, von einer Endstation 100 über asynchrone LAN-Vermittlungen 20 zu einer synchronen Vermittlung 10 einer virtuellen Pipe. In dieser Ausführungsform sendet eine Endstation 100 Datenpakete von ihrer Netzschnittstelle 102. Das Datenpaket wird über eine oder mehrere asynchrone Vermittlungen 20 (z. B. Knoten B und C in 56) vermittelt, bis es die synchrone Vermittlung 10 einer virtuellen Pipe erreicht. Wie im Folgenden erläutert wird, resynchronisiert die synchrone Vermittlung 10 das ankommende Datenpaket in dieser Konfiguration.
• 2. Wie in 57 gezeigt ist, von einer Netzschnittstelle 102 einer Endstation 100 über asynchrone LAN-Vermittlungen 20 zu einem Asynchron-Synchron-Gateway 15, das den asynchronen Datenpaketstrom in einen synchronen Strom umsetzt und daraufhin die Datenpakete zu einer synchronen Vermittlung einer virtuellen Pipe weiterleitet.
• 3. Wie in 58 gezeigt ist, von einer ersten synchronen Vermittlung 10 einer virtuellen Pipe über asynchrone Vermittlungen oder Router 20 zu einer zweiten synchronen Vermittlung 10 einer virtuellen Pipe. In dieser Konfiguration wird das ankommende Datenpaket zu der zweiten synchronen Vermittlung 10 einer virtuellen Pipe, wie im Folgenden erläutert wird, resynchronisiert.

A system is provided for managing the transmission of data packets over asynchronous switches in a timely manner with three configurations:

• 1. As in 56 is shown by an end station 100 over asynchronous LAN switches 20 to a synchronous exchange 10 a virtual pipe. In this embodiment, an end station sends 100 Data packets from their network interface 102 , The data packet is transmitted via one or more asynchronous switches 20 (eg Node B and C in 56 ) until it synchronizes the mediation 10 reached a virtual pipe. As will be explained below, the synchronous switch resynchronizes 10 the incoming data packet in this configuration.
• 2. As in 57 shown from a network interface 102 a terminus 100 over asynchronous LAN switches 20 to an asynchronous synchronous gateway 15 which converts the asynchronous data packet stream into a synchronous stream and then forwards the data packets to a synchronous switch of a virtual pipe.
• 3. As in 58 is shown by a first synchronous exchange 10 a virtual pipe over asynchronous switches or routers 20 to a second synchronous exchange 10 a virtual pipe. In this configuration, the incoming data packet becomes the second synchronous switch 10 a virtual pipe, as explained below, resynchronizes.

6.1 Resynchronization by the router control unit at the entrance port

A data packet through one or more asynchronous switches 20 , that is, exchanges without shared time reference, has been forwarded, can either be resynchronized to its original plan at the input port or at the output port. The 59 - 61 describe the resynchronization operation on the input port of either a synchronous switch 10 a virtual pipe or a gateway 15 ,

To the resynchronization by the end station 100 send data packets sent by the synchronous exchange 10 a virtual pipe, like in 59 is shown periodically timing messages M002 to the end station 100 , The timing messages M002 represent the value of the common time reference (CTR) 002 as they do through the synchronous mediation 10 Will be received.

The network interface 102 at the terminus 100 integrates the timing information received from the timing message M002 into the timestamp field 35TS ( 15 ) in the header of the data packets, 1DP, 2DP, 3DP, ..., 9DP in 59 and send them over the asynchronous switches 20 to the synchronous exchange 10 a virtual pipe. 60 shows the possible arrival time of the data packets 1DP, 2DP, 3DP, ..., 9DP at the input port 35 , A data packet may experience a delay between a minimum delay and a maximum delay, as in 60 is shown, the delay barrier uncertainty as the maximum delay - minimum delay form, is.

• 1. Nachschlagen des Verzögerungsschranken-Unbestimmtheitsparameters 35DB (d. h. [Maximalverzögerung – Minimalverzögerung]) in der Leitweglenkungstabelle 35B unter Verwendung der PID 35C.
• 2. Verzögern des Pakets bei dem Eingangsport um: 35DB–[(CTR 002)–(Zeitstempel 35TS)].

The 61A and 61B show by the routing control unit 35 at the entrance port 30 using a resync buffer 30R Resync operation that is performed in the following two steps:

• 1. Look up the delay-bound uncertainty parameter 35dB (ie [Maximum Delay - Minimum Delay]) in the routing table 35B using the PID 35C ,
• 2. Delay the package at the input port by: 35dB - [(CTR 002 )-(Time stamp 35TS )].

6.2 The router controller operation with a gateway

If it is in front of the entrance to the mediation a virtual pipe an asynchronous synchronous gateway 15 there, the router control unit works as it did here before

14 was specified. As noted below, the gateway in this case receives the CTR 002 and also directs the forwarding time frame delimiter (TFD) 47A continue to the serial receiver. More specifically, their operation may be the same as the output operation described below.

14 Fig. 10 illustrates the flowchart for the processing program executed by the routing control unit 35B of the router control unit 35 , The program responds to two basic events from the serial receiver 31 out 14 Receiving the time frame delimiter TFD in step 35-01 and receiving the data packet in step 35-02 , Upon receipt of a TFD, the routing control unit calculates 35 in step 35-03 the arrival time value (ToA) attached to the incoming data packets 35T , For this calculation, it uses a constant Dconst, which is the time difference between the tick point of the common time reference (CTR) 002 and the receipt of the TFD at time t2 (which is determined by the CTR 002 is generated in this node in an adjacent exchange). This time difference is caused by the fact that the delay from the serial transmitter 49 to the serial receiver 31 is not an integer number of timeframes. If the data packet in step 35-02 is received, the routing control unit performs 35B three as in step 35-04 outlined operations: attach the ToA, look up the address of the queue 36 using the PID and storing the data packet in this queue 36 ,

6.3 Operation of the router control unit without a gateway

If the mediation 10 a virtual pipe is connected directly to the asynchronous network, the switch can 10 a virtual pipe does not receive a TFD, so the operation of the routing control unit as described hereinbefore in 14 should be changed in the following way. The operations in 35-01 and 35-03 are picked out, with some being one of 35-04 in 14 clearly indicate that the arrival time attached to the packet header is directly from the common time reference (CTR). 002 should be derived.

6.4 The output port

The output operation described here concerns both mediation 10 the synchronous virtual pipe in the 56 and 58 as well as the asynchronous synchronous gateway 15 in 57 ,

The like here before in 16 specified output port 40 includes (as described hereinbefore) a scheduling controller with a transmit buffer 45 and with a serial transmitter 49 , The planning control unit 45 maintains an mapping of each of the data packets between the associated respective time of arrival (ToA) and an associated routing time out of the output port via the serial transmitter 49 out. The forwarding time is compared to the common time reference (CTR) 002 certainly. As described herein, this map resynchronizes the stream of data packets forwarded through the intermediary of a virtual pipe.

• 1. eine Verzögerungsanalyse- und Planungssteuereinheit, die ferner eine Weiterleitungstabelle 45B umfasst;
• 2. einen Sendepuffer 45C, der üblicherweise als ein Schreib-Lese-Speicher (RAM) realisiert ist; und
• 3. eine Auswahlpuffer-Steuereinheit 45D, die Datenpakete zu dem seriellen Sender weiterleitet.

The scheduling controller and the send buffer 45 have different modes of operation in the 19 . 48 - 49 and 62 are described. The different modes of operation correspond to some of the possible configurations with the asynchronous switches as they are described in US Pat 56 - 58 were discussed. The scheduling controller and the send buffer 45 in 19 contain three parts:

• 1. a delay analysis and scheduling control unit, which further includes a routing table 45B includes;
• 2. a send buffer 45C commonly implemented as a random access memory (RAM); and
• 3. a selection buffer control unit 45D which forwards data packets to the serial transmitter.

The delay analysis and planning control unit 45A performs together with the select buffer controller 45D using the PID 35C , the time stamp 35TS and the data packet arrival time (ToA) 35T the mapping for determining the particular time frame to which a respective packet should be forwarded from the output port. The two control units 45A and 45D are constructed of a central processing unit (CPU), a random access memory (RAM) for storing the data and a read only memory (ROM) for storing the control unit processing program.

From the switching fabric 50 come over the transmission line 51 Data packets. The data packets whose priority bit 35P is activated (i.e., reserved traffic) are determined by the delay analysis and scheduling control unit 45A to one of the k send buffers 45C (B-1, B-2, ..., Bk). As in the 5 and 6 is shown, in each time cycle, each of the k buffers is designated to store packets forwarded in each of the k time frames. The presence of k send buffers allows the delay analysis and Planning control unit 45A Plans data packets in a wide range of delay changes.

The send buffer 45C contains an additional buffer BE for "best effort" data packets. The priority bit 35P in the "best effort" data packets is not activated, and this is how the delay analysis and scheduling control unit determines that these data packets should be stored in the "best effort" buffer. Every time there are no more scheduled data packets (the priority bit 35P is activated), the data packets of "best effort" become the serial transmitter 49 forwarded.

6.5 Operation of Selection Buffer Control Unit:

21 illustrates the flowchart for the operation of the selection buffer controller 45D , The control unit 45D reacts in step 45-11 on the tick point 002 the common time reference (CTR) and then incremented in step 45-12 transmit buffer index i (ie, i: = i + 1 mod k ', where k' is the number of scheduled traffic buffers) and sends a time frame delimiter TFD to the serial transmitter 49 , If the transmission buffer Bi in step 45-13 is not empty, he sends, as in step 45-14 is specified, a data packet from the transmit buffer Bi, while otherwise, as in step 45-15 is specified, sends a "best effort" data packet from the "best effort" buffer BE.

6.6 The delay analysis and planning control unit

• I. 49 – der Fall beliebiger, aber beschränkter Verzögerungsänderungen, wie er in den 56–58 beschrieben ist, wie er in 45-051 angegeben ist:
• 1. Es sei <s1, s2, s3, ..., sj> die Menge der Zeitrahmen mit einer PID = p, die sich in jedem Zeitzyklus wiederholt, wie in der Weiterleitungstabelle 45B bei dem Eintrag p angegeben ist.
• 2. die Steuereinheit 45A durchsucht die Menge <s1, s2, s3, ..., sj>, um den ersten möglichen Zeitrahmen si zu bestimmen, der nach (ToA 35T) + CONST auftritt (wobei CONST eine konstante Schranke an die Verzögerung über die Switching Fabric ist); und
• 3. si ist der Zeitrahmen, zu dem das Datenpaket für die Sendung über den seriellen Sender geplant wird – und i ist der Indexsendepuffer B-i.

The schedule for the program executed by the delay analysis and scheduling control unit is in 48 illustrated. The main task of the program is to compute the index i of the transmit buffer Bi between B-1 and Bk in step 245-05 , There are several possible methods for performing the calculation in step 245-05 which depends on the type of delay changes that can occur on the communication links. In the 49 and 62 two possible calculation methods are described:

• I. 49 The case of arbitrary but limited changes of delay, as described in the 56 - 58 is described as he in 45-051 is indicated:
• 1. Let <s1, s2, s3, ..., sj> be the set of time frames with a PID = p that repeats every time cycle, as in the routing table 45B at the entry p is specified.
• 2. the control unit 45A search the set <s1, s2, s3, ..., sj> to determine the first possible time frame si, which after (ToA 35T ) + CONST (where CONST is a constant bound on the delay across the switching fabric); and
• 3. si is the time frame at which the data packet for transmission is scheduled via the serial transmitter - and i is the index transmission buffer Bi.

• II. 62 – der Fall, wenn in dem Paketanfangsblock ein Zeitstempel verwendet wird (15 und 22): Wenn in Schritt 45-03 das Datenpaket von der Fabric empfangen wird, wird die PID 35C in dem Datenpaket-Anfangsblock wie in Schritt 45-351 angegeben verwendet, um den Weiterleitungsparameter 45F in der Weiterleitungstabelle (45B aus 19) nachzuschlagen und daraufhin in Schritt 45-352 den Index i des Sendepuffers 45C zu berechnen: i = ((Zeitstempel 35TS) + 45F] mod k', wobei k' die Anzahl der Puffer für geplanten Verkehr ist.

The set <s1, s2, s3, ..., sj> forms a plurality of time frames in which a data packet for transmission from the output port of a switch can be scheduled.

• II. 62 The case when a timestamp is used in the packet header ( 15 and 22 ): When in step 45-03 the data packet received by the fabric becomes the PID 35C in the data packet header as in step 45-351 specified used to the forwarding parameter 45F in the forwarding table ( 45B out 19 ) look up and then in step 45-352 the index i of the send buffer 45C to calculate: i = ((timestamp 35TS ) + 45F ] mod k ', where k' is the number of buffers for scheduled traffic.

Claims (44)

A system for scheduling and managing data transmission of data packets, said system comprising: a plurality of switches ( 10 ) with a plurality of input ports ( 30 ) and output ports ( 40 ) each having a unique address for receiving the data packets, the data packets being of at least one type ( 35P ) are; a virtual pipe ( 25 ) comprising at least two of the exchanges connected via communication links ( 41 ) are interconnected in a path to receive data packets of the selected type ( 35P ) from at least one source to at least one destination; and a common time reference signal ( 002 ) coupled to selected ones of the switches; where the common time reference signal ( 002 ) is divided into a plurality of time frames (TF); and wherein each time frame (TF) has a predefined duration; characterized in that when selecting selected ones of the exchanges ( 10 ) Data packets of the selected type ( 35P ) in response to the common time reference signal ( 002 ) during a selected sequence of time frames ( 7 ) from the said selected ones of the exchanges ( 10 ) along the virtual pipe ( 25 ) to be forwarded; the time frames of the selected sequence of time frames have a respective sequence of time frame start times; and wherein there are predefined time intervals among the start times of the respective sequence of time frames. The system of claim 1, wherein if no data packets of the selected type to forward data packets of "best effort" (47-be). System according to claim 1, in which the position of the data packet in the time frame at least one of the following is: arbitrary and predefined. The system of claim 1, wherein there are a plurality of virtual pipes ( 25 ) gives; and at the selected one of the communication links of at least one virtual pipe ( 2 ) are used simultaneously. System according to claim 1, wherein the common time reference signal ( 002 ) is coupled to at least one of the following: GPS (Global Positioning System), Galileo, GLONAS (GLObal Nvigation Satellite System), TWTFT (Two-Way Satellite Time and Frequency Transfer), WWV Radio Station, WWVB Radio Station, IRIG Time Codes ( Inter-range instrumentation group time codes), directly over a single communication link from selected switches, indirectly over a plurality of communication links from selected switches, directly over a single communication link from adjacent network interfaces and gateways ( 15 ) and indirectly via a plurality of communication links of network interfaces and gateways ( 15 ). System according to claim 1, wherein the communication links at least one of the following are: glass fiber, copper cable, coaxial cable, twisted wire pair, wireless channel, 10 megabit / second ethernet, 100 megabit / sec ethernet, 1 Gigabit / sec Ethernet, 10 Gigabit / second Ethernet, wireless IEEE 802.11 channel, wireless UMTS channel (wireless channel of Universal Mobile Telephone System), SONET, SDH and fiber channel. System according to claim 1, in which the communication links wireless communication links between at least one of the following are: a ground station and a satellite and between two satellites that are on orbit around the earth. System according to claim 1, in which the data packets of the selected type at least one of The following are: Internet Protocol data packets (IP data packets), Multi-Protocol Label Switching (MPLS) frames, Ethernet frames, Fiber Channel frames (FC frames) and asynchronous transfer mode cells (ATM cells). The system of claim 1, further comprising a scheduling control unit ( 45 ), wherein the planning control unit ( 45 ) provides for the data packets of the selected type associated forwarding times. A system according to claim 9, wherein the data packets of the selected type ( 35P ), which are forwarded via a selected virtual pipe, at least one assigned pipe identifier (PID) ( 35C ) to have. The system of claim 10, wherein a predefined number of contiguous time frames comprise a time cycle; and where the time cycles are contiguous ( 1 ). A system according to claim 11, wherein the sequence of time frames associated with the switches of a selected virtual pipe are associated with the respective switches for all time cycles ( 8th ). A system according to claim 11, wherein there is a predefined time interval between the input to said selected one of the switches and the output from said selected one of the switches within each of the time cycles ( 8th ). The system of claim 11, wherein a predefined number of time cycles comprise a supercycle; where the supercycle is periodic ( 6 ). System according to claim 14, wherein the common time reference signal in accordance with the UTC standard (Coordinated Universal Time Standard). System according to claim 15, wherein the supercycle duration is at least equal to one of the following is one second, measured using the UTC standard, one Fraction of a second, measured using the UTC standard, one predefined number of seconds, measured using the UTC standards. System according to claim 10, wherein the PID ( 35C ) is at least one of the following: an Internet Protocol (IP) address, an IPv4 address, an IPv6 address, an Internet Protocol Group group address, asynchronous transfer mode (ATM) labels, an identifier a virtual connection (VCI), a virtual route identifier (VPI), a tag as defined by the Multi-Protocol Label Switching (MPLS), a Forwarding Equivalence Class (FEC), an IEEE 802-MAC (Media Access Control ) Address and an IP address along with an IP port number. A system according to claim 9, wherein selected ones of the data packets of the selected type have timestamps ( 35TS ) contain; and where the planning control unit ( 45 ) provides forwarding times for selected ones of the selected type in response to the respective timestamps and common time reference. The system of claim 18, wherein the time stamp ( 35TS ) is generated by at least one of the following: Internet Real-Time (RTP) protocol, a predefined one of the switches, a data packet source, an end station ( 100 ), a gateway ( 15 ) and a network interface. System according to claim 15, where there is a first predefined time frame, within its a selected one Data packet of the selected type transferred to the exchange begins to be, and a second predefined time frame, within whose the selected Data packet of the selected Type starts to be forwarded from said exchange, gives; and wherein the first predefined time frame and the second non-overlapping predefined timeframes are. System according to claim 20, where the first predefined time frame is UTC (Coordinated Universal Time) and in at least one of the following formats is shown: a time frame number within a time cycle and as a time cycle number within a supercycle, a fixed point representation, a floating-point representation and a two-part number: (i) an integer of seconds and (ii) a fraction of a second. System according to claim 20, in which it between the first predefined time frame and the second predefined time frame a predefined time difference gives. System according to claim 22, wherein the predefined time difference for each switch at least one of the following is: a constant number, predefined for each PID, for each Timeframes predefined within each time cycle. The system of claim 10, wherein predefined positions are assigned for selected ones of the input ports within a predefined time frame within which data packets of the selected type are transferred to the respective input ports, and separate predefined positions within a second predefined time frame within which the respective data packets of the selected one Type of selected ones of the output ports are assigned ( 24 ). System according to claim 9, where it is in the planning control unit of each of the selected Placements a first, through the said planning control unit certain scheduled time within a first predefined time frame within which a respective data packet is planned for is to be forwarded from the respective mediation begin, and where the time frame and the first scheduled time in Reaction to the common time reference signal can be determined. System according to claim 25, which is for everyone The specified time frame gives a defined bandwidth, which the capacity for sending the data packets during limited by the time frame; the planning control unit in Reaction that the defined bandwidth for a each time frame exceeded is determined, a congestion; and the planning control unit in response to determining a stowage for a particular of the time frame for selected ones Data packets of the selected Type a second predefined time within a second predefined one Timeframe for transmission said respective data packets from the respective exchange re-plans out. The system of claim 10, wherein the scheduling controller includes a routing table for assigning predefined time frames for the transmission of each of the respective output ports in response to a timestamp field ( 35TS ), the unique address of the input port and the PID ( 35C ) owns. A system according to claim 27, wherein the predefined time frame for the transmission of the data packet of the selected type by adding a predefined number of time frames to the time stamp ( 35TS ) is determined. The system of claim 28, wherein the number of predefined time frames added to the time stamp field to determine the transmission time frame of said data packet in response to the lookup of that number in the routing table in the scheduling controller using the PID ( 35C ) is determined as an index in said routing table. A system according to claim 27, wherein the scheduling control unit matches the time difference in time frames between the common time reference and the time stamp ( 35TS ) and providing means for discarding the respective data packet of the selected type if said time difference is above a predefined time threshold. The system of claim 9, wherein the virtual pipe has a defined maximum delay between any two of the selected ones of the switches; in which the input ports comprise means for receiving the data packets of the selected type and for recording an arrival time (TOA) ( 35T ) for each named data packet; wherein the scheduling control unit schedules a first scheduled time within a first predefined time frame within which each of said data packets is scheduled to be forwarded from a respective output port of the respective switch, and a second scheduled time within a second predefined time frame the said data packet being alternatively scheduled to commence transmission from the respective switch and a third predefined scheduled time within a third predefined time frame for alternatively scheduling the transmission of said data packet from the respective output port of the switch; and wherein the first, second, and third predefined time frames are determined in response to at least one of the following: the common time reference and the TOA ( 35T ). The system of claim 9, wherein for each selected one of the exchanges, there is a predefined time frame within which a respective data packet of the selected type begins to be transmitted to the respective switch and a separate predefined time frame within which the respective packet is transmitted from the respective switch begins to be, gives; wherein each switch has an associated predefined set of time frames, the switch issuing a data packet from said virtual pipe during a time frame of the predefined set of time frames; each of the data packets, when it arrives, in response to an instantaneous value of the common time reference, an arrival time (TOA) ( 35T ), wherein the packet is scheduled to be issued during an unoccupied subsequently available time frame of the respective predefined set of time frames; and wherein the subsequently available time frame is determined in response to the determination that the time elapsed since the TOA is greater than a predefined threshold. The system of claim 10, wherein the PIDs have an associated parameter table containing a predefined set of values specifying the reserved number of data packets of the selected type that can be forwarded in each of the predefined time frames; wherein a cell flow control and load control unit counts and compares the number of data packets with the same subset PID in each of said time intervals and stores a count in a load value table and in response to comparing the count in the load table with respective ones of the predefined set of values provides an output ( 44 . 55 ). System according to claim 33, in which the cell current control and load control unit the Number of data packets of the selected type with the same Subsets PID as within a predefined range or outside of the predefined area. System according to claim 34, where an infringement message is generated and issued if the number of data packets of the selected type with the same Subset of PIDs outside the predefined area lies. The system of claim 9, wherein the switches are at least one synchronous switch and at least one asynchronous switch; wherein an end station ( 100 ) is connected to one of the synchronous switches through communication links in a path via respective asynchronous switches; wherein there is a defined delay between the end station and said synchronous switch; and wherein the scheduling controller determines for each of the synchronous switches a first scheduled time at which a selected one of the data packets from the respective synchronous switch is scheduled to begin forwarding. System according to claim 36 where the scheduler calculates the delay by each the data packets delayed should be, before it is forwarded. A system according to claim 37, wherein a time stamp is (at each said data packet ( 35TS ) is attached when it is from the end station ( 100 ), the time stamp being derived from the common time reference; and wherein the scheduling controller, in response to the timestamp value, calculates the time difference by which said data packets should be delayed before being forwarded from the respective output port of the respective synchronous switch. System according to claim 38, in which the common time difference by at least one the following means is transmitted to the terminal: by a Timing message transmitted in a data packet directly from a UTC source (Source of Coordinated Universal Time) and a timing server. The system of claim 10, wherein the switches are at least one synchronous switch and at least one asynchronous switch; wherein there is a way in the network for connecting a first one of the synchronous switches to a second one of the synchronous switches via a plurality of the asynchronous switches ( 58 ); wherein there is a defined delay in the transmission of respective ones of the data packets between the first synchronous switch and the second synchronous switch; the second synchronous switch, in response to the receipt of the respective data packet from the first synchronous switch, determining the time of arrival (TOA) ( 35T ) records for each separate data packet; and wherein the scheduling controller determines, in response to the PID, the TOA, and the common time reference in the second synchronous switch, a first scheduled time within which a respective data packet is scheduled to begin being forwarded from the second synchronous switch. A system according to claim 40, wherein a respective time stamp is selectively selected on selected ones of the data packets ( 35TS ) when forwarded from the first synchronous exchange; wherein the time stamp value is received from the common time reference; and wherein the scheduling controller, in response to the timestamp value, calculates the first scheduled time at which said data packets should be forwarded from the second synchronous switch. A system according to claim 9, wherein at least one of the switches is an asynchronous synchronous gateway ( 15 ), which is connected to at least one end station ( 100 ) is coupled; wherein at least one of the switches is an asynchronous switch ( 20 ); wherein there are interconnected communication links in a path for connecting the end station to the asynchronous synchronous gateway via asynchronous switching ( 20 ) gives; being between the end station ( 100 ) and the first asynchronous synchronous gateway ( 15 ) gives a defined delay for the transmission of the data packets; wherein the asynchronous synchronous gateway receives data packets from the end station and for each of said data packets an arrival time (TOA) ( 35T ) records separately; and wherein the scheduling control unit in response to the PID ( 35C ) and the TOA ( 35T ) associated with each said data packet, for each said data packet, a first scheduled time within a first predefined time frame within which a respective named data packet is scheduled for it, from the respective asynchronous synchronous gateway ( 15 ) to begin to be transmitted. A system according to claim 42, wherein the scheduling control unit calculates the delay by which the respective said data packet is to be delayed before being sent out of the asynchronous synchronous gateway (16). 15 ) is forwarded. A system according to claim 42, wherein a time stamp (on each said data packet) ( 35TS ) is attached when it is forwarded by the end station; and wherein the time stamp value is derived from the common time reference.