RFID reader-to-reader collision avoidance model with multiple-density tag distribution solved by artificial immune network optimization

Radio frequency identification (RFID) is an emerging non-contact technique where readers read data from or write data to tags by using radio frequency signals. When multiple readers transmit and/or receive signals simultaneously in a dense RFID system, some reader collision problems occur. Typically, in a modern warehouse management system, the warehouse space is partitioned into blocks for storing different goods items on which RFID tags are affixed. The goods items with the equal size are placed in the same block. Because the sizes of goods items are possibly different among blocks, the density values of tags that are affixed on the goods items are different from each other. In this case, tags in each block are distributed randomly and uniformly while tags in the whole warehouse space (i.e., all blocks are considered as a whole) follow a non-uniformly random distribution. For the sake of academic research, this situation is defined as a multiple-density tag distribution. From the viewpoint of resource scheduling, this article establishes an RFID reader-to-reader collision avoidance model with multiple-density tag distribution (R2RCAM-MTD), where the number of queryable tags is used as the evaluation index. Correspondingly, an improved artificial immune network (AINet-MTD) is used as an optimization method to solve R2RCAM-MTD. In the simulation experiments, four cases with different blocks in a warehouse management system are considered as testbeds to evaluate the effectiveness of R2RCAM-MTD and the computational accuracy of AINet-MTD. The effects of time slots and frequency channels are investigated, and some comparative results are obtained from the proposed AINet-MTD algorithm and the other existing algorithms. Further, the identified tags and the operating readers are graphically illustrated. The simulation results indicate that R2RCAM-MTD is effective for reader-to-reader collision problems, and the proposed AINet-MTD algorithm is more efficient in searching the global optimal solution of R2RCAM-MTD than the existing algorithms such as genetic algorithm (RA-GA), particle swarm optimization (PSO), artificial immune network for optimization (opt-aiNet) and artificial immune system for resource allocation (RA-AIS).