This application is the US National Stage of International Application No. PCT/EP02/11013, filed Oct. 1, 2002 and claims the benefit thereof. The International Application claims the benefits of European application No. 01126009.8 filed Oct. 31, 2001, both of the applications are incorporated by reference herein in their entirety.
The invention relates to a method, receiving device, and transmitting device for determining the fastest message path without clock synchronization.
Secured transport protocols normally measure the round-trip delay, which is to say the overall delay occurring when a message is transmitted from a transmitter to a receiver and when a further message, such as a confirmation message, is transmitted from the receiver to the transmitter, in order to derive therefrom when messages have to be repeated. Typical examples of such secured transport protocols are Transmission Control Protocol TCP according to IETF RFC 793 and Stream Control Transmission Protocol SCTP according to IETF RFC 2960.
To measure said round-trip delay, the transmit time of a message is noted in the transmitter and the receive time of a confirmation message arriving from the recipient of the transmitted message is determined. The round-trip delay is determined from the difference between said times. No clock synchronizing is required for this method.
Since, however, the messages can be transmitted from the transmitter to the receiver on message paths different from those on which the confirmation messages are transmitted from the receiver to the transmitter, the determined value of the round-trip delay allows no inferences to be made about the one-way delays from the transmitter to the receiver and vice versa. The determined value of the round-trip delay allows no inferences to be made about the one-way delays from the transmitter to the receiver and vice versa even when the messages are transmitted from the transmitter to the receiver on the same message path as that on which the confirmation messages are transmitted from the receiver to the transmitter as the message path can be highly asymmetric.
If multiple message paths exist for a transmitter for transmitting the messages to a receiver, then it can be advantageous to send the messages on that message path on which they are transmitted to the receiver the quickest. For this, information must be available to the transmitter as to which message path is the quickest. The fastest message path can be determined using the one-way delays of all message paths from the transmitter to the receiver.
Known solutions for measuring the one-way delays require clock synchronism of the clocks or, as the case may be, clock generators of the transmitter and receiver. Clock synchronism here means that the clocks or, as the case may be, clock generators run at the same rate but without necessarily also having the same absolute times. If clock synchronism is present, any message, or a message specially provided for test purposes and furnished by the transmitter with a time stamp is sent from the transmitter to the receiver. The one-way delay for the relevant message path can be ascertained by the receiver from the difference between the time of reception and the time stamp. In order to ascertain which path from among a number of existing message paths is the fastest, the method must be suitably repeated for all message paths. By comparing the one-way delays ascertained by the receiver for all the message paths, the fastest message path can be easily identified in the receiver and signaled to the transmitter.
Clock synchronism between the transmitter and receiver does not, however, exist in many cases.
The object of the present invention is to improve the known methods, receiving devices, and transmitting devices for determining the fastest message path.
Said object is achieved by means of the features of claims 1, 9, and 15.
Preferred embodiments are the subject of the dependant claims.
A major advantage of the method according to the invention lies in not requiring clock synchronism of the transmitter and receiver for ascertaining the fastest message path between the transmitter and receiver. The underlying principle is that in order to ascertain the fastest message path it is not necessary to know the absolute delay times on the various message paths but that it instead suffices to transmit a message on each of several message paths under investigation and to determine the fastest message path by means of the sequence of reception.
According to an advantageous development of the method it is also possible to ascertain, alongside the fastest message path, the relative message propagation times, which is to say the delays caused by each individual message path compared to the fastest message path claim 2. This can be applied to, for example, distributing the load between two or more message paths if the propagation times of the message paths differ only slightly, which is to say the ascertained relative propagation times of said message paths, referred to the fastest message path, are adequately short.
The messages sent by the transmitter for determining the fastest message path or, as the case may be, for measuring the relative propagation times are advantageously sent simultaneously or a very short time apart—claim 4. If the communication protocol and/or transmitting device employed do/does not provide for the simultaneous sending of messages on multiple message paths, it can be provided for said messages to be sent a very short time apart.
The method according to the invention is explained in more detail below with the aid of two drawings.
On the basis of the time differences between receipt of the identical messages over the different message paths P1, P2, P3 it is moreover possible to ascertain the relative differences in propagation time ΔT2, ΔT3, which is to say the delay times occurring on all message paths P2, P3 compared to the fastest message path P1. Expressed as a formula, the relative difference in propagation time ΔT2 for the message path P2 is: ΔT2=T2-T1. For the message path P3, the relative difference in propagation time ΔT3 is obtained according to the following formula: ΔT3=T3-T1. The relative propagation time ΔT1 of the fastest message path P1 represents a special case. It can be expedient, depending on the application, to set the relative propagation time ΔT1 of the fastest message path P1 to zero: ΔT1=0—not shown. If the relative message propagation times ΔT1, ΔT2, ΔT3 are transmitted to the transmitting device S, it will have been ensured that the transmitting device S recognizes the fastest message path P1 without the need to signal this separately.
In order to determine the fastest message path P1 or, as the case may be, relative message propagation times, the receiving device E only needs to be able to recognize that the messages were sent at the same time. This can be derived from, for instance, a time stamp in the messages, a specific message type with sequence numbers, or simply from the protocol itself.
When the fastest message path has been determined by the receiving device E, it is only necessary to signal to the transmitting device which message path is the fastest and is to be used henceforth for transmitting messages from the transmitting device S to the receiving device E.
If the communication relationship is bi-directional, then the described method will be employed independently in both directions—not shown.
Message elements embedded in customary messages, notifications or parameters can be used instead of complete messages in order to implement the method according to the invention. A request message element DelayReq will then be used instead of a separate request message DelayReq. Either the already cited confirmation messages DelayRes can be sent as a reply, or a different message type will be used in which an appropriate confirmation message element DelayRes is embedded.
The method according to the invention can be advantageously applied with the Stream Control Transmission Protocol SCTP according to IETF RFC 2960. This is shown schematically in
It is further assumed that a part of the expansions which are described in IETF Internet Draft draft-ietf-tsvwg-addip-sctp-02 has been implemented. It is possible with this part for an SCTP endpoint E to set the primary path or, as the case may be, primary message path of a communication partner S, meaning that E can determine on which path the communication partner S sends messages to the endpoint E. This expansion is used for signaling which message path is the fastest.
An SCTP message consists of a common header and a number of chunks. Said number of chunks can vary with different SCTP messages; the use of messages without chunks is also permissible. Each chunk is classified by a chunk type—a number between 0 and 255. The method described here uses special new chunk types (DelayReq(SN), DelayRes(SN)) with a sequence number SN. SCTP allows such expansions. In doing this it is possible to maintain interoperability with SCTP endpoints for which said new chunks have not been implemented. The receiving device E sends out a request message DelayReq(SN) in order to initiate a measurement. SN+1 is used as the sequence number for the next measurement, which is to say that an additional counter is maintained by the receiving device E. If a request message DelayReq(SN) is received by the transmitting device S, then in each case one confirmation message DelayRes(SN) will be sent on some or all available message paths P1, P2, P3. The receiving device E will then be able to ascertain which message path P1, P2, P3 is the fastest. The fastest message path P1 is specified by the destination IP address of the confirmation message which contains the DelayRes(SN) chunk and is received first. Said IP address will then be signaled with the message “Set Primary IP Address” (described in 3.2.5 and 4.4 of the cited IETF Internet Draft draft-ietftsvwg-addip-sctp02, reproduced in extract form below) to the transmitting device S, which will thereupon always use said message path P1 as the first message path for transmitting data. The frequency of said measurements depends on the stability of the network. An available possible option is, however, to perform such measurements periodically.
As an alternative to the above described method employing DelayReq and DelayRes chunks it is also possible to use the Heartbeat and Heartbeat-Ack chunk in order to determine the fastest message path P1, P2, P3. In response to the reception of a Heartbeat chunk, Heartbeat-Ack chunks will then be sent on some or all message paths instead of the DelayRes chunks in the above exemplary embodiment. This method also ensures interoperability with other implementations. It is moreover advantageously possible to dispense with introducing new chunk types, although it must be added that the behavior of the protocol will have to be adjusted correspondingly.
Extract from IETF Internet Draft draft-ietf-tsvwg-addip-sctp-02
Number | Date | Country | Kind |
---|---|---|---|
01126009.8 | Oct 2001 | EP | regional |
Filing Document | Filing Date | Country | Kind |
---|---|---|---|
PCT/EP02/11013 | 10/1/2002 | WO |