A study on test nodes selection problem with test capacity constraint in QoE measurement network

Test nodes selection problem in QoE (Quality of Experience) measurement network is to select a few number of nodes in service network to test users' experience information as much as possible. In this paper, test nodes selection problem with test capacity constraint is considered. To begin with, we describe it as a constrained set covering problem and develop a 0-1 integer nonlinear programming model based on a penalty factor. We propose a sub procedure to solve test capacity constraint, and propose Binary Particle Swarm Optimization-Basic Algorithm (BPSO-BA) to solve the constrained set covering problem. Simulation results show that the BPSO-BA can obtain good results with fast convergence and has almost no dependence on the initial values. However, it is sensitive to the penalty factor. Then we improve the basic algorithm for its adjustment difficulty of penalty factor. In addition, combining with 2-opt neighborhood searching algorithm, coverage rate for user nodes can be further enhanced. Finally, comparing the results of greedy heuristics, basic algorithm and improved algorithm, we find that the improved algorithm can get the highest coverage rate among the three algorithms for the same number of test nodes. Experimental results demonstrate that both the constrained set covering model and improved algorithm proposed in this paper are efficient and superior for solving test nodes selection problem with test capacity constraint in QoE measurement network.