Summary: | 博士 === 國立中興大學 === 資訊科學與工程學系所 === 99 === In this dissertation, we address and explore two important performance issues of optical packet switching networks. The first issue we address is that an optical packet switch need to periodically reconfigure its switching fabric for moving packets through the switch. The reconfiguration overhead is not negligible with respect to the packet transmission time. And this has a significant impact on the switch performance. The overhead increases the average waiting time of packets and worsens throughput performance, so scheduling packets requires additional considerations on the reconfiguration frequency. The second issue is on the performance of routing packets through optical switches with Fiber Delay Lines (FDLs). Switch buffers that use FDLs have a volatile nature due to signal loss and noise accumulation, because the packets are recirculated in FDLs for storage. Packets suffer from excessive recirculation through FDLs, and they may be dropped eventually in their routing paths. Because of this, packet scheduling becomes more difficult in FDL buffers than in RAM buffers, and requires additional design considerations for reducing packet loss.
In this dissertation, we firstly intend to find analytically the optimal reconfiguration frequency that minimizes the average waiting time of packets. We proposes an analytical model to facilitate our analysis on reconfiguration optimization for input-buffered optical packet switches with the reconfiguration overhead. The analytical model is based on a Markovian analysis and used to study the effects of various network parameters on the average waiting time of packets. Of particular interest is the derivation of closed-form equations that quantify the effects of the reconfiguration frequency on the average waiting time of packets. Quantitative examples are given to show that properly balancing the reconfiguration frequency can significantly reduce the average waiting time of packets. In the case of heavy traffic, the basic round-robin scheduling scheme with the optimal reconfiguration frequency can achieve as much as 30\% reduction in the average waiting time of packets, when compared with the basic round-robin scheduling scheme with a fixed reconfiguration frequency.
Secondly, we propose a latency-aware scheduling scheme and an analytical model for all-optical packet switching networks with FDL buffers. The latency-aware scheduling scheme is intended to minimize the packet loss rate of the networks by ranking packets in the optimal balance between latency and residual distance. The analytical model is based on non-homogeneous Markovian analysis to study the effect of the proposed scheduling scheme on packet loss rate and average delay. Furthermore, our numerical results show how various network parameters affect the optimal balance. We demonstrate quantitatively how to achieve the proper balance between latency and residual distance so that the network performance can be improved significantly. For instance, we find that under a given latency limit and light traffic load our scheduling scheme achieves a packet loss rate 71% lower than a scheduling scheme that ranks packets simply based on latency.
|