This paper focuses on the problem of buffering performance optimization for optical packet switch with shared WDM (Wavelength Division Multiplexing) optical buffers. For this purpose, several packet scheduling algorithms are proposed to make full use of buffering resources. Influences of these algorithms on optical packet switch are evaluated by simulation experiments. Simulation results show that under different traffic load, different packet scheduling algorithms should be adopted. In addition, the performance of optical buffers also depends on the granularity of FDL.
In the paper, a new optical packet switch is proposed, which uses shared WDM buffers (fiber delay-lines) and shared tunable wavelength converters (TWCs) as contention resolution. In each fiber delay-lines, many wavelengths can be used to buffer packets at the same time. The shared TWCs have two kinds of functions, either as translators for free wavelengths in destination output fibers, or as translators for free wavelengths in the feedback FDLs. Based on this architecture, a lower packet loss rate can be achieved without the need of deploying a large number of FDLs and TWCs. The performance of the new architecture has been extensively studied by means of simulation experiments.
In this paper, an inner wavelength method (IWM) is proposed to enlarge buffering capacity of shared FDL-buffers. In addition, an optical packet switch called Extended shared buffer type optical packet switch (Extended SB-OPS) is proposed to realize the inner wavelength method. In order to further improve performance of Extended SB-OPS, a greedy algorithm based on inner wavelength method (GA-IWM) is introduced. The performance of Extended SB-OPS is evaluated by simulation experiments.
In this paper, a new type of p-cycles, multi-granularity flow p-cycles (MFP), is presented. A two-step approach is developed for the design of MFP with span failure recovery. Our simulation results show that MFP are more capacity-efficient than conventional span-protecting p-cycles (SP) and flow p-cycles (FP). The impacts of network topology on performance of SP, FP and MFP are evaluated. In this paper, we also propose a new p-cycle pre-selection strategy. Results show that our strategy is more efficient than two existing pre-selection strategy in p-cycle networks.
This paper considers the dynamic routing issues in WDM grooming
networks. Given network resources, a major objective of a routing
algorithm is to minimize the blocking probability of connection
requests. Wavelengths and transceivers are usually limited
resources in a network. The way to use these resources directly
determines the network performance. So the challenge is how to use
wavelength resources efficiently while keeping higher
transceiver-utilization efficiency. To solve this problem, we
develop an approach called Hop Constrain Lightpath
Establishment Approach (HCLEA). The main idea of HCLEA is to
guide the routing algorithms not to create inefficient long
lightpaths. Simulation results show that HCLEA does improve the
network performance in terms of transceiver-utilization
efficiency, wavelength-utilization utilization, and bandwidth
The demand to carry large amount of data as fast and as reliably as possible is continuously increasing, in parallel with formation of the information society. Nowadays, Fiber optic cables using Wavelength Division Multiplexing (WDM) are used to provide huge transport capability to support such demand effectively and economically. Each link is able to carry huge amount of traffic, thus a possible failure causes loss of tremendous data. So network survivability is highly required for such a high speed and huge capacity network. The operation of the modern telecommunication network is based on a multi-layer structure. The major reason of this multi-layer is the multiplexing hierarchy for efficient use of transmission facilities and the convenience of service and facility management. The most common example of multi-layer telecommunication network is the ATM over SDH over WDM configuration. This work investigates to the survivability of such multi-layer networks, and an integrated spare capacity dimensioning scheme, joint network design (JND) method, is proposed. Compared with single layer design (SLD) method, JND is bandwidth efficient and cost effective. The configuration problem of multi-layer fault tolerant networks is formulated as an Integer Linear Programming model. We also present a genetic algorithm approach for large networks whose size makes the ILP model difficult to use.
According to the previous researches, all-optical converters based on FWM and XGM conversion schemes at present are well suited for system use. Both performances of optical networks with FWM and XGM conversion are examined in this paper. From the results, we find that the benefit obtained by full range conversion can be achieved by using FWM conversion with the degree of conversion d=1 or 2, when the number of wavelengths on a fiber is small. But with the increase in number of wavelengths, higher conversion degree is required to achieve the performance of full range. As for the case of XGM conversion, our results show that the performance of XGM is worse distinctly than full range conversion because of its asymmetric conversion, no matter the number of wavelengths is large or small. In order to improve the performances of the networks with limited range conversion in the case of large number of wavelengths, three simple wavelength assignment algorithms (FUWC, LUWC and MSW) are proposed. We apply these algorithms to the case of Ring and Mesh-torus. Through simulation, we find that our algorithms do greatly reduce the blocking probabilities of networks with limited range conversion.