On three shades of partial protection in elastic optical networks

Gone have been the days when the allocation of spectrum resources for a demand is over-provisioned due to worst-case based designs and the optical transmission technologies are based on the fixed spectral grid, modulation format and/or line-rate schemes. Elastic optical networks technologies featuring adaptive spectrum allocation mechanisms have been becoming the architecture and technology of choice for future core networks to meet explosive traffic growths in a cost and energy-efficient manner. Partial protection strategies based on the observation that in failure events, a service can tolerate a certain amount of degradation and therefore by reducing the protection traffic in the network, better spectrum utilization could be attained. Such concept has been widely studied in the traditional wavelength division multiplexing (WDM) context and yet has been somehow faded due to the fact that fixed transmission technologies provide limited room for spectral improvements. This paper aims to renew the interest in partial protection and re-adapt it to the context of elastic optical networks. The evolutionary perspective we lay out in this paper identifies a new route for achieving greater spectral efficiency in a pragmatic way by simply differentiating the protection services for each demand rather than the uniform treatment for all demands. As such, we present a new research problem entitled, routing, modulation level, spectrum and protection service assignment which is essentially an extension of the well-established one, that is, routing, modulation level, and spectrum assignment as the (partial) protection service for each demand is taken into account and optimized. Three variants of that problem reflecting shades of applying partial protection are covered in details. Specifically, the first one considers the intuitive case as the amount of protection traffic for each demand is provided as the input to the network planning process while the second one is dedicated to the special case of enforcing the same level of partial protection for all demands. More interesting case emerges in the third variant where given the service level agreement for the total protected network traffic, we provide the optimal solution that determines the protection service for each individual demand so as to minimize the spectral occupancy. Extensive numerical simulations have been conducted in the realistic COST239 network topology and it has been revealed that a remarkable spectral saving could be achieved thanks to introducing partial protection services. Moreover, our results have demonstrated that the demand-wise approach far outperforms the one-size-fits-all one and the difference could be up to 25% in our studied cases.

Automated summary

Elastic optical networks featuring adaptive spectrum allocation mechanisms are becoming the architecture and technology of choice for future core networks. This paper aims to renew the interest in partial protection and adapt it to the context of elastic optical networks by differentiating the protection services for each demand. Extensive simulations in a realistic network topology have shown that introducing partial protection services can achieve significant spectral savings.