Why Avoid Collisions? Exploit Them! Information Collection in Multi-Tagged RFID Systems

We investigate the problem of target object information collection in multi-tagged RFID systems. Different from its single-tagged peers, the multi-tagged RFID scenario introduces three new challenges: 1) Tags on the same object carry the same information, so reusing single-tagged algorithms causes unnecessary redundancy; 2) The transition of objects from being tagged one to multiple tags leads to an upsurge in slot collisions; 3) Gathering information from more tags necessitates more downlink transmission, making it hard to limit broadcast information while ensuring time-efficient information collection. To tackle these technical challenges, we propose an efficient information collection algorithm, called Backtracking Collision Peeling (BCP), featuring three key techniques. First, BCP selects a single time slot (allowing even collision slots) for each target object to convey its information. This approach bypasses the need for a high-latency collision elimination process, thereby significantly reducing time overhead. Second, by exploiting dependencies among the selected slots, BCP recovers object information in complex collision slots using object information from already resolved slots, offering a novel and effective solution for handling signal collisions. Third, to minimize downlink transmission cost, BCP polls tags by transmitting only incremental changes between polling vectors, rather than the complete vectors. We further improve performance with E-BCP by enhancing the utilization of collision slots, thereby reducing the number of required polling rounds. Experiments show BCP and E-BCP outperform existing methods by at least 35% in aggregate execution time, while also exhibiting stronger stability and robustness.