Performance analysis of homowavelength crosstalk in optical packet networks

Abstract

In this paper, homowavelength crosstalk in optical packet networks is evaluated with the following aspects taken into consideration: device crosstalk coefficient, signal extinction ratio, traffic load,The evolution of packet switching in the optical domain has naturally evolved to address transport-sporadic and bursty IP traffic more efficiently. With advantages such as high speed, data-rate and format transparency, and flexibility, optical packet switching has recently become a popular research topic worldwide [1][2][3][4][5][6]. However, transparency places a stringent constraint on optical components so that the signal quality will not significantly degrade. Crosstalk between optical packet channels, one of the problems associated with transparency, should be investigated both to understand its constraint on the optical components and to guide the design of OPN nodes. Considerable efforts have been devoted to analyzing and suppressing crosstalk in traditional circuit-switched optical networks [7][8][9][10][11][12]. According to whether it occupies the same nominal wavelength of the signal or not, crosstalk in optical networks is generally classified into two types, namely, homowavelength crosstalk (HOC) and heterowavelength crosstalk (HEC) [13]. HEC is not a serious problem, since it can be removed easily by employing filters. On the other hand, HOC, because it occupies the same wavelength of the signal, cannot be removed once coupled and accumulates through the cascade. Thus, HOC potentially poses a serious limitation to the performance of optical packet networks (OPNs). To our knowledge, very few researchers have addressed this issue; therefore, we made a detailed analysis of the HOC in OPNs.By considering device crosstalk coefficient, signal extinction ratio, traffic load, node size, and cascadability, the effect of HOC on the performance of OPN nodes is analyzed in detail. In the following section, a brief introduction of crosstalk in OPNs is made, and then bit error rate and packet error rate arising from HOC are analytically computed. To show the significance of the HOC-induced performance degradation in OPNs, the influence of HOC is evaluated with examples of practical network nodes. In section 3, numerical results are presented with discussions, and conclusions are drawn in section 4. THEORY Crosstalk in OPN NodesIn OPNs, the packet length in the optical domain can be fixed or variable. However, in practical applications, packets with fixed duration are generally adopted, mapped into fixed timeslots, and scheduled in a slotted fashion to simplify the implementation of the nodes. In the input interface of the OPN node, during each timeslot all packets are aligned in phase and separated by demultiplexers according to their wavelengths. Then they are delivered to corresponding output ports, according to their address information by the switch matrix. In the output interface, contention resolution and packet header regeneration are generally performed [1].At the OPN nodes, the components of nonid...

The performance analysis of SOA switches in optical switching networks

“…The power of channel(X,Y) output can be described as [4]: crosstalk. There are generally three ways of generating crosstalk, shown as the dashed lines in Figure 2.…”

“…Case 1 is the leakage from the desired channel at the demultiplexer travels across the switch matrix and is added onto the original channel at the multiplexer; Case 2 is the leakage from the other channels at the demultiplexers; Case3 is the leakage from the other channels is coupled onto the desired channel at the SOA switch matrix. The BER in case of k interfering channels can be calculated with Gaussian approximation as [4]:…”