Scalable approaches for path computation

Optimal and fast approaches for path computation are of paramount importance especially for the path computation elements (PCE) of generalized multi-protocol label switching (GMPLS) networks. To overcome the computational complexity, optimality must be traded for the execution time, enabling scalable computation. This paper considers the typical routing problems addressed by PCE and compares different optimization approaches based on integer linear programming (ILP) formulation and its relaxation, with the objective of assessing optimality and scalability of the approaches.