A localized fault tolerant load balancing algorithm for RFID systems

Radio frequency identification (RFID) is a unique scientific invention that comprises individually recognizable, low-cost tags and readers where the readers monitor the tags using frequencies from the radio spectrum. Uniform distribution of the tags for gaining a balanced load of the readers is a significant concern to ensure successful collection of data from all of the tags of an RFID system with multiple readers. Moreover, some of the readers in an RFID network may become defective during operation and stop working. As a result, information would not be collected from those tags which were associated with the defective readers and the network would operate with partial information. We target to maintain a balance among the load of the readers by placing the tags as evenly as possible to address the fast tag reading problem. We convert the addressed issue as a load balancing problem and introduce a cellular automaton inspired localized algorithm as a solution to this problem. Our proposed algorithm utilizes the local information of the readers to relocate tags from a heavily loaded reader to a lightly loaded reader. We develop our proposed algorithm as a fault tolerant one so that all of the tags in the network are always under surveillance even if some of the readers become defective. Numerical analysis and comparison results suggest that the proposed localized load balancing algorithm outperforms the existing localized solution and gives a competitive result compared to the centralized algorithm. Finally, we implement our proposed algorithm in the parallel programming platform Compute Unified Device Architecture that greatly improves the runtime of the proposed algorithm.