Aiming at the inherent correlation of azimuths and tilts of the antennas installed on the base stations, a quaternion-based particle swarm optimization algorithms is proposed to optimize the coverage performance for mobile networks, in which the quaternions represents the antenna orientations, and the rotations formed by the multiplication in the quaternion field ensures that the feasible solutions, that is the particles, update along the shortest paths. Moreover, the author considers the process to perform infinitesimal movements towards different directions in turn and thus introduced a blending quaternion-based particle swarm optimization algorithm to overcome the problems that the destination of the particle movement depends upon the order of directions including along the inertial velocity, towards the individual historical optimal solution and towards the global optimal solution, which is indeed caused by the anti-commutative law of quaternions. Simulations show that the proposed algorithms, especially the latter, perform better than the canonical particle swarm algorithm, firefly algorithm and genetic algorithm both in the convergence efficiency and the final optimized coverage.