Introduction of Dynamic Virtual Force Vector in Particle Swarm Optimization for Automated Deployment of RFID Networks

A scheme for automated planning of passive RFID network is proposed. The scheme comprises two parts. The first part creates a fast site-specific probabilistic propagation model for successful identification from the reader of any possible tag antenna. The materials of surrounding walls as well the tag antennas radiation pattern, the geometry and the polarization of both reader and tag are taken into account. In the second part, a hybrid form of particle swarm optimization (PSO) algorithm is applied. The proposed approach selects a subset of tag antenna configurations to be installed so that a given cost function is satisfied. By clustering problematic areas during each iteration and moving the swarms towards them, we imitate the acts of a human-planner. The combinatorial performance of all active readers is evaluated at each tag location; this reveals that good identification performance is recorded at overlapping regions, where no single reader- tag antenna operates adequately. The proposed clustering approach greatly improves the convergence-time of the standard PSO and greatly reduces equipment, cutting down the cost of the network accordingly. Comparison with standard PSO reveals that the overall equipment can be reduced by a factor of two, satisfying the same quality constraints.