A Bandwidth-Balanced RMLSA Solution for Static Elastic Optical Network: A Two Stages Approach

This paper addresses one of the main tasks in transparent static elastic optical networks (EONs), known as the routing, modulation level, and spectrum assignment (RMLSA) problem. We present, for the first time, a two-stage RMLSA solution, focusing on reducing spectrum consumption (or network capacity). The first macro stage, called least demand bandwidth balance (LDBB), relies on a physical layer impairment (PLI) model to jointly compute the connections' route and modulation level (RML). We use a new balancing criterion that effectively distributes each network link's frequency slot unit (FSU) demands, exemplifying three different balancing functions based on the maximum number of FSUs on the links and the total number of FSUs demanded and the cost of the route. In the last macro stage, using all connections chosen paths, we perform the spectrum assignment (SA) process using two specific connections prioritization criteria. We propose two SA algorithms, called sliding-fit (SF) and parcel-fit (PF), reducing the spectrum consumption. These algorithms change the SA paradigm by searching connections for a given subset of the frequency spectrum, contrary to the search of FSUs for a given connection in standard approaches. In all cases, our solution exhibited a lower total network capacity than the commonly used strategies found in the literature, with an average network capacity reduction of 5.7 % FSUs. In addition, our proposal may be used to easily dimension network capacity and determine how many extra resources may be needed to attend to all network users.