GB2406026A

GB2406026A - Synchronizing transmitter and receiver clocks using a synchronisation signal, and measuring round trip time

Info

Publication number: GB2406026A
Application number: GB0418114A
Authority: GB
Inventors: Nigel Gaylard; Scott Burke
Original assignee: MICROWAVE DESIGNS Ltd
Current assignee: MICROWAVE DESIGNS Ltd
Priority date: 2003-08-15
Filing date: 2004-08-13
Publication date: 2005-03-16
Anticipated expiration: 2024-08-13
Also published as: GB0319206D0; GB0418114D0; GB2406026B

Abstract

A method of reporting timing information in a network is disclosed. The method comprises: at a sending node, initiating transmission of a packet at a predetermined part of a clock cycle. The packet includes a synchronisation signal. At a receiving node, the method aligns a clock to the received synchronisation signal, and sends an acknowledgement to the sending node. At the sending node, the round-trip time to send a signal to the receiving node is measured, and reporting the measured time to a software system intended for processing data packets. Typically, the time is reported to a higher network layer that is responsible for using the timing information to process data packets at a time after they are received so as to maintain synchronisation with other nodes in the network.

Description

Time synchronisation of interconnected devices This invention relates to

time synchronisation of interconnected devices. Specifically, it relates to the time synchronisation of interconnected nodes in a network.

Typically, a synchronous system with a plurality of nodes maintains its synchronicity by either continually streaming data to all the nodes or by all nodes synchronizing off an external clock. In a packet-based network where the timing of transmitted and received signals is not predictable, the ability for multiple receiving nodes to be able to detennine precisely when to act upon incoming messages presents a considerable technical challenge.

Present networking technology uses time stamping within messages and reference to an external time standard for synchronisation. Such methods are suitable for synchronizing nodes to microsecond tolerances. Packetbased networking is finding applications in fields considerably removed from those contemplated by its designers. Some of these new applications require more precise synchronisation than can be offered by conventional techniques. For example, synchronisation of nodes in distributed audio or video equipment demands a synchronisation accuracy considerably greater than that is possible using an external reference. Moreover, the cost of accessing such time reference servers is unattractive to many users.

An aim of this invention is to provide a protocol for time synchronisation in a network at a physical layer level that does not require an external time reference.

From a first aspect, this invention provides a method of reporting timing information in a network comprising: at a sending node, initiating transmission of a packet at a predetermined past of a clock cycle, the packet including a synchronisation signal; at a receiving node, aligning a clock to the received synchronisation signal, and sending an acknowledgement to the sending node; at the sending node, measuring the round-trip time to send a signal to the receiving node' and repotting the mcasui-ed time to a software system intended for processing data packets.

A; ce .e The protocol reports synchronization and timing enror between nodes to the software that must process the data. Knowing the timing enror, applications running on each node may process the information for synchronization of their clocks. No central time server is required because no attempt is made to synchr5nise al] nodes to a common time base. The physical layer protocol determines the time delay in communication between nodes; it is the responsibility of other software components, most typica]]y, higher layers in the network stack, to act upon this infonnation when processing data received over the network.

The method may be executed within a network stack on the sending or receiving nodes.

Typically, it is performed in a physical layer of the network stack. In such a case, the software system intended for processing data packets is typically the higher-level protocol. The higher-]eve] protocol may then process data packets received over the network, the time at which the packet is processed after its receipt being determined as a function of the measured time reported by the lower-level protocol. Thus, timing data is passed up the network stack from the low-level physical protocol.

The measured time may be reported by the sending node to the receiving node. This depends upon the requirements of the particular application. For example, this step is typically taken in audio and video reproduction, where it is used by the receiving node to generate synchronized output. Most usually, the measured time is repotted to the receiving node by a higher-level protocol in the network stack.

From a second aspect, the invention provides a method of synchronization of devices on a network, the network including a master node and one or more receiving nodes, in which the receiving nodes are operative to report timing information by a method according to the first aspect of the invention upon receipt of a broadcast packet forth the master node, on receipt of each packet, or periodically.

The protocol also provides a mechanism for reporting between nodes the time difference between the transmitting and receiving clocks, which is useful in a variety of diagnostic and user applications, SUC]1 as distance measurement between nodes for security on a network, rea]-time monitoring of the networks cable quality if used on a wired network.

eee ee An embodiment of the invention will now be described in detail, by way of example, and with reference to the accompanying drawings, in which.

Figure] is a flow diagram illustrating a sequence of events in determining synchronization information in an embodiment of the invention; Figure 2 is a diagram of a data packet used in the method of Figure 1; and Figure3 is a timing diagram illustrating the timing of data flows in the method illustrated in Figure 1.

The method of this embodiment facilitates synchronised communication between a master node and a peripheral node. For example, the method might be applied in an audio system with a central server and two or more remote speakers, connected either wired, wireless or optically. The central server transmits the audio for all channels as an encoded stream in a sequence of packets with audio and synchronization information.

Each speaker receives the packets and must reproduce the audio data at in synchronism with the other speakers. The timing infonnation within the packet indicates how long after receipt of the packet it should be played. This is required to maintain the relative phase of the audio output of the various speakers. Methods embodying the invention can apply to other technologies as well, such as sync}onising audio and video, or for 3D Video projection, among many other potential applications.

The synchronization procedure makes use of a packet having a structure shown in Figure 2. The packet has a header that includes a synchronization pulse, followed by a payload and an acknowledgement section.

The method commences with the central server transmitting a packet of the type described above to the remote node. The shape of the synchronization pulse is shown in Figure 3. The packet is transmitted such that the leading edge of the synchronization pulse is synchronized with a rising edge of the server's main clock, as shown in Section A in Figure 3.

The receiver, on completion of receiving a packet, performs en or checking upon it. If the packet passes the tests, processing continues as follows. The node receives the pulse sequence at the start of the packet, it aligns its transceiver clock to the incoming eee ese eee synchronization pulse or pulses, as shown in Section B in Figure 3. The receiver then decodes and acknowledges the remainder of the incoming packet. The physical layer of the receiving node can then align its master clock with the transceiver clock and is then synchronised with the transmitting unit. After synchronizing with the packet, and synchronizing with the sequence, the node sends an acknowledgement back to the sender unit.

The receiving clock as shown at Section B has a static time offset in relation to the clock of the sending unit, due to natural start-up conditions, temperature and frequency drift. This offset is equa] to It -tO in Figure 3. The time of flight; that is, the time taken for the synchronization sequence to anrive at the receiver is indicated by t: and shown in Section B of Figure 3.

The physical layer of the receiver locks to the incoming synchronization sequence thereby pulling its master clock's rising edge into time alignment with the incoming sequences rising edge as shown in Section C of Figure 3.

The sender unit then measures the time offset between the synchronization sequences contained within the acknowledgement. This difference of t3 - to represents twice the time of flight for the transmission. The measured difference can then be reported to higher layers in the network protocol stack. Depending upon the application, the sending node may then send a packet back to the receiving node, the packet containing the measured timing information. It will be noted that this packet is at a network layer that is higher in the protocol stack than the physical layer at which the timing measurement is perfonned.

The exchange of packets described above is summarised in the flowcharts of Figure 1.

Unlike other methods of providing time alignment, the resolution that can be achieved by methods embodying the invention is limited to the best available digital delay units, which are used to provide the synchronization mechanism within the physical layer. At the time of making the invention, this is in the order of I Ops.

In an enhancement to this method, slave nodes in master-slave type systems can use the synchronization pulses to resynchronise their master clock in order to minimise drift and temperature variations. Alternatively there could be a node on the system that ee ae e ë eece b oadcasts a message instructing all nodes to resynchronise their master clocks and thus ensure that all the nodes in the system are fully synchronous.

Another such implementation would have nodes using the synchronization pulse to provide a reference offset with which to apply to the received message. Thus, the node could resynchronise on a packet-by-packet basis to the exact time reference of the master node. .

Claims

:: .. .. . Hi. Claims 1. A method of reporting timing infonnation in a

network comprising: at a sending node, initiating transmission of a packet at a predetermined part of a clock cycle, the packet including a synchronization signal; at a receiving node, aligning a clock to the received synchronization signal, and sending an acknowledgement to the sending node; at the sending node, measuring the round-trip time to send a signal to the receiving node, and reporting the measured time to a software system intended for processing data packets.
2. A method of reporting timing information according to claim I which is performed in a network stack on the sending or receiving node.
3. A method of repotting timing infonnation according to claim 2 which is perfonned in a physical layer of the network stack.
4. A method of repotting timing infonnation according to claim 3 in which the measured time is reported to a higher layer of the network stack.
5. A method of reporting timing infonnation according to any preceding claim in which the measured time is reported to the receiving node.
6. A method of reporting timing infonnation according to any preceding claim in which the measured time is reported to the receiving node by a higher level protocol in the network stack.
7. A method of repotting timing infonnation according to claim 6 in which the software system intended for processing data packets is the higherlevel protocol.
8. A method of reporting timing infonnation according to claim 7 in which the higher-level protocol processes data packets received over the network, the time at which the packet is processed after its receipt being detennined as a function of the measured time repotted by the lower- level protocol.

ce ë e
9. A method of synchronization of devices on a network, the network including a master node and one or more receiving nodes, in which the receiving nodes are operative to report timing infonnation by a method according to any preceding claim upon receipt of a broadcast packet forth S the master node.
] O. A method of synchronization of devices on a network, the network including a master node and one or more receiving nodes, in which the receiving nodes are operative to report timing infonnation by a method according to any of claims 1 to periodically.
11. A method of synchronization of devices on a network, the network including a master node and one or more receiving nodes, in which the receiving nodes are operative to report timing infonnation by a method according to any of claims I to 8 on each packet r eceived.