Multicasting in WDM Single-Hop Local Lightwave Networks

In modem networks, the demand for bandwidth and high quality of service (QoS) requires the efficient utilisation of network resources such as transmitters, receivers and channel bandwidth. One method for conserving these resources is to employ efficient implementations of multicasting wherever possible. Using multicasting, a source sending a message to multiple destinations may schedule a single transmission which can then be broadcasted to multiple destinations or forwarded from one destination to another, thus conserving the source transmitter usage and channel bandwidth. This thesis investigates the behaviour of single-hop WDM optical networks when they carry multicast traffic. Each station in the network has a fixed-wavelength transceiver and is set to operate on its own unique wavelength as a control channel. Each station also has a tuneable wavelength transceiver in order to transmit or receive signals to or from all the other stations. A transmission on each channel is broadcasted by a star coupler to all nodes. Multicasting in single-hop WDM networks has been studied with different protocols. This thesis studies the multicasting performance adopting receiver collision avoidance (RCA) protocol as a multicasting protocol. This study takes into consideration the effect of the tuneable transceiver tuning time which is the time required to switch from one wavelength to another, and the propagation time required by a packet to propagate from one node to another. The strategy in RCA protocol is that nodes request transmission time by sending a control packet at the head of their queues. Upon receipt of this information all nodes run a deterministic distributed algorithm to schedule the transmission of the multicast packet. With the control information, nodes determine the earliest time at which all the members of the multicast group can receive the packet and the earliest time at which it can be transmitted. If a node belongs to the multicast group addressed in the control packet, its receiver must become idle until all nodes in the group have tuned to the appropriate wavelength to receive the packet. This problem leads to poor transmission and consequently low channel utilisation. However, throughput degradation due to receiver conflicts decreases as the multicast size increases. This is because for a given number of channels, the likelihood of a receiver being idle decreases as the number of intended recipients per transmission increases. The number of wavelengths available in a WDM network continues to be a major constraint. Thus in order to support a large number of end users, such networks must use and reuse wavelengths efficiently. This thesis also examines the number of wavelengths needed to support multicasting in single-hop optical networks.

Text FK_2000_19.pdf Download (1MB)

Actions (login required)

View Item