Optimization algorithms for the proactive configuration of elastic optical networks under jamming attacks and demand uncertainty

Abstract

This work presents optimization algorithms on attack-aware routing and spectrum allocation (Aa-RSA) in conjunction with spectrum selective switch (SSS) placement, taking into account variations in traffic demands. The objectives of the algorithms are to minimize the required number of SSSs, as well as the number of lightpath re-allocations between different demand scenarios, taking into account the crosstalk interactions that could potentially be utilized during a jamming attack to facilitate the spread of the attack to additional lightpaths throughout the network. To solve the offline network design problem, first the proposed Aa-RSA algorithm is modeled as a mathematical formulation that jointly considers all the design parameters. Then, integer linear program (ILP)-based decomposition techniques and graph-based heuristics are also proposed to address large-sized networks. Additionally, it is shown through a dynamic time-varying Aa-RSA scenario, that the offline pre-calculated network configurations are a good fit for the traffic demand variations, eliminating the computational time required in the path computation element (PCE) for dynamically reconfiguring the network. The benefits of each of the proposed algorithms with respect to the network's operating and capital expenditures (measured in terms of performance metrics such as the number of SSSs deployed, the spectrum utilization, and the number of re-allocations) are reported and validated through extensive simulation results.