Q-factor assessment of SOA-based ultrafast nonlinear interferometer

On the feasibility of full pattern-operated all-optical XOR gate with single semiconductor optical amplifier-based ultrafast nonlinear interferometer

“…On the other hand the extent of the permissible values implies that the SOA is biased well into deep saturation because in the most extreme case it is subject to a pair of control pulses whose energy content within the specific dynamic range is significant. This results in a strong accumulative effect on the SOA nonlinear response unlike the simpler case in which this active element is driven by one only such signal, as it happened in [34] where the working condition required to ensure a similar performance standard was indeed relaxed. Therefore, the competence of the considered scheme to perform modulo-2 arithmetic on ultrafast bitwise level is stressed on the edge and inevitably a heavier operational burden is imposed on the erbium doped fiber amplifiers (EDFAs) that must be employed for this purpose.…”

“…In order to model the operation of the SOA-based UNI in its dual rail switching mode it is necessary to mathematically describe the output at port O1, which is given by the following basic interferometric equation [32][33][34] …”

“…The aforementioned approach for finding the values of h A (t) and h B (t) can be adopted provided that the appropriate initial conditions are defined and applied in 6 and 7. These can be derived following the rationale explained in [34] but adapted to the working principle of the considered scheme and the demanding situation where two control signals drive one after the other the SOA, thus affecting its dynamic characteristics in an alternating manner. More specifically, the first temporal segment of the leading pulse in control A encounters an unsaturated SOA because the gain of this active element is not modified before this moment.…”

“…This in turn makes less easy the adjustment of the setup and compromises its overall practicality regarding the issues of versatility, synchronization and retiming, reconfigurability and ''on-the-fly" processing, thereby limiting its potential for unobstructed use in more complicated circuits or subsystems, especially in those where the situation may already be operationally difficult, for example due to the existence of feedback paths [5]. Moreover, although the XOR gate has attracted considerable theoretical interest when implemented with the SOA-MZI [25][26][27][28][29] or SOA-assisted Sagnac [30,31], this does not also hold for the SOA-based UNI since the relevant studies have focused on the case where there is solely one intense signal that is responsible for switching (control) [32][33][34] and hence the strain imposed on the SOA dynamics is weaker as opposed to two such beams being simultaneously present [35]. The above reasoning points out the imperative need for methodically investigating the feasibility of obtaining the truth table of the XOR function with a single SOAbased UNI, which is the ultimate performance challenge given the inherent difficulties related to the SOA saturation properties.…”

“…Consequently, their recombined polarization at the output PBS is the same with that at the front-end and hence the clock pulse leaves entirely from port O2, which corresponds to this polarization orientation. Things change, however, when a pulse pertaining to one of the data streams is launched in the SOA since a depletion of its carriers and hence an alteration of its refractive index is caused, provided that the former is energetic enough (at least 10 times stronger than the clock [34]) to dominate the dynamical behaviour of the latter. In fact, if such pulse is arranged to temporally access one, but always the same, clock replica, then owing to the manifestation of the cross-phase modulation effect inside the SOA a phase difference is incurred between the orthogonally polarized clock counterparts so that they are no more balanced in this respect.…”