부산대학교 도서관

Cellular IoT networks are expected to enable connecting massive smart devices reliably while fulfilling diverse requirements and are facing critical challenge for optimizing the coverage by adjusting the azimuths and the downtilts of antennas installed in the base stations. Particle swarm optimization (PSO) has shown outstanding performance in solving many high-dimensional parameter optimization problems. We proposed an improved PSO algorithms, Q-PSO, based on quaternion. The proposed Q-PSO method changes the manner that the particles move towards the best candidate by the quaternion operations, increasing the convergence speed, thereby improving the efficiency of the algorithm. Simulation results show that the Q-PSO method significantly edges out the standard PSO in terms of the convergence rate, the optimized coverage ratio and the efficiency for most cases. The proposed PSO methods not only deepen our understanding of swarm intelligence but also provide some guidelines for addressing other network optimization problem with huge number of configuration parameters.