|dc.description.abstract||Synchronisation deals with the distribution of time and/or frequency across a network
of nodes dispersed in an area, in order to align their clocks with respect to time and/or frequency. It remains an important requirement in telecommunication networks, especially in Time Division Duplexing (TDD) systems such as Ultra Wideband (UWB)
and Digital Enhanced Cordless Telecommunications (DECT) systems. This thesis explores three di erent research areas related to clock synchronisation in communication networks; namely algorithm development and implementation, managing Packet Delay Variation (PDV), and coping with the failure of a master node.
The first area proposes a higher-layer synchronisation algorithm in order to meet the specific requirements of a UWB network that is based on the European Computer
Manufacturers Association (ECMA) standard. At up to 480 Mbps data rate, UWB
is an attractive technology for multimedia streaming. Higher-layer synchronisation
is needed in order to facilitate synchronised playback at the receivers and prevent distortion, but no algorithm is de ned in the ECMA-368 standard. In this research area, a higher-layer synchronisation algorithm is developed for an ECMA-368 UWB network. Network simulations and FPGA implementation are used to show that the new algorithm satis es the requirements of the network.
The next research area looks at how PDV can be managed when Precision Time
Protocol (PTP) is implemented in an existing Ethernet network. Existing literature
indicates that the performance of a PDV ltering algorithm usually depends on the
delay pro le of the network in which it is applied. In this research area, a new sample-mode PDV filter is proposed which is independent of the shape of the delay profile. Numerical simulations show that the sample-mode filtering algorithm is able to match or out-perform the existing sample minimum, mean, and maximum filters, at differentlevels of network load.
Finally, the thesis considers the problem of dealing with master failures in a PTP
network for a DECT audio application. It describes the existing master redundancy
techniques and shows why they are unsuitable for the specific application. Then a
new alternate master cluster technique is proposed along with an alternative BMCA
to suit the application under consideration. Network simulations are used to show
how this technique leads to a reduction in the total time to recover from a master