We.B1.9 58 ICTON 2005 ⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯ The work reported in this paper was supported in part by the IST-NOBEL project. 0-7803-9236-1/05/$20.00 ©2005 IEEE Experimental Characterization of Impairments Induced by Link-Control-Channels in DWDM Systems Marco Presi *, Luca Giorgi, G. Contestabile, Stefan Herbst ** and Ernesto Ciaramella Scuola Superiore Sant'Anna, Via G. Moruzzi 1, 56127 -Pisa, Italy Tel: (+39) 050 5492269, Fax: (+39) 050 5492194, e-mail: marco.presi@cnit.it *: also with Physics Department, University of Pisa, Italy **: Marconi ONDATA, Backnang, Germany ABSTRACT To reduce in-line amplifier power transients induced by optical power variations, Link Control Channel might be considered. This technique can lead to significant transmission degradations. We present a numerical and experimental characterization of such degradation effects. We determine which control channels power levels and grid spacing are compatible with common DWDM links specifications. These results could be used for design rules. Keywords: link control channel, transients suppression, DWDM systems. 1. INTRODUCTION Optical systems and networks widely use Erbium-Doped Fibre Amplifiers (EDFAs) that are known to be sensitive to input power excursions when working in the saturation regime. It is well known that such input power variations produce EDFAs gain fluctuations, thus producing undesirable performance degradation, and possible components breaking[1]. Gain fluctuations response time is directly related to Erbium excited states life-time, and thus it is in the order of several μs. Power control is therefore a key-point, not only for systems performance, but also for components protection. In a typical optical link, composed of several cascaded EDFAs, the transient response time is shortened at each span-link and induced power fluctuations are enhanced [2]: however in most cases, controlling the first amplifier in the link can prevent transient effect propagation [3]. In transparent optical networks, a non- controlled transient could propagate in the various links that constitute a light-path: since lightpaths are cross- connected in the optical nodes transient-induced degradations may "propagate" to affect even disjoint lightpaths. Considering an optical node, in case of a link failure (all channels at one input port are missing) it is expected that sub-sets of channels will be missing on the various output ports: thus amplifiers stabilization should be implemented at the output ports of each node. Various solutions could be used to prevent transients. Among them, gain clamping in EFDAs can be accomplished by various schemes [3, 4]. An alternative approach is the dynamic control of the overall link power, by introducing an extra CW channel in the WDM comb [5]. In this case, power compensation is realized monitoring the total power at the link input and dynamically adjusting the power of the control channel. The only point of this scheme would be to realize a circuit with the proper speed, a circuit know as Link Control Unit (LCU) [4]. The Link Control Channel (LCC) implementation has still some drawbacks. While it is able to reduce impairments deriving from power transients, it does not guarantee that global WDM spectral shape is preserved, which may become critical in ultra long haul links (>1000 km ). Indeed it was shown that dropping a particular sub-set of channels from a DWDM comb, even controlling the overall gain produces nonlinear effects (from combinations of intra-band Raman gain and Spectral Hole Burning) that can significantly degrade the system performance [6]. In this paper, we show that a LCU can produce other nonlinear impairments at quite shorter link lengths than in [6], particularly when the LCC should replace a quite high number of channels. We found that when the power ratios between LCC and system channels exceeds a threshold value, the system suffers from nonlinear penalties: this strongly limits range of power variations that a control channel can compensate for. In the following we first numerically analyze which system parameters can be critical and determine an upper limiting value for the LCU. We then show an experimental assessment of these limitations. 2. NUMERICAL ANALYSIS After dropping a sub-set of channels n' from a set of n WDM channels, the power adjustment that the Link Control Channel (LCC) must supply in order to maintain the link EFDAs chain in saturated regime is given by the equation [7]: ( ) ( ) ( ) [ ] n n' n' n' n n n n c c c P G λ P G λ G λ = ∆P 1 1 1 1 − Σ − − Σ − , (1) here ∆P is the power adjustment, λ c and G c are LCC wavelength and gain respectively, and P n P n' are the power of a single channel before and after dropping, and λ n and G n represent each n th data channel. Equation (1), shows