Adaptive hierarchical artificial immune system and its application in RFID reader collision avoidance

This paper proposes an efficient adaptive hierarchical artificial immune system (AHAIS) for complex global optimization problems. In the proposed AHAIS optimization, a hierarchy with top-bottom levels is used to construct the antibody population, where some antibodies with higher affinity become the top-level elitist antibodies and the other antibodies with lower affinity become the bottom-level common antibodies. The elitist antibodies experience different evolutionary operators from those common antibodies, and a well-designed dynamic updating strategy is used to guide the evolution and retrogradation of antibodies between two levels. In detail, the elitist antibodies focus on self-learning and local searching while the common antibodies emphasize elitist-learning and global searching. In addition, an adaptive searching step length adjustment mechanism is proposed to capture more accurate solutions. The suppression operator introduces an upper limit of the similarity-based threshold by considering the concentration of the candidate antibodies. To evaluate the effectiveness and the efficiency of algorithms, a series of comparative numerical simulations are arranged among the proposed AHAIS, DE, PSO, opt-aiNet and IA-AIS, where eight benchmark functions are selected as testbeds. The simulation results prove that the proposed AHAIS is an efficient method and outperforms DE, PSO, opt-aiNet and IA-AIS in convergence speed and solution accuracy. Moreover, an industrial application in RFID reader collision avoidance also demonstrates the searching capability and practical value of the proposed AHAIS optimization.