Item-level tagging sees more tags: Analyzing the performance of EPC Gen-2 protocol in large-scale RFID systems

Radio Frequency IDentification (RFID) system is a highly sought-after technology prominently used in the automation, logistics, and supply-chain industries. EPC Gen-2 is the current standard RFID air-interface protocol in the industry. Underlying this standard is the Q-adaptive MAC protocol, which is used to identify and read the tags, by stochastically arranging them over time. The Q-adaptive protocol's performance is sensitive to the number of tags placed under the vicinity of the reader. With the current era of `item-level tagging' witnessing an increased number of tags in a given area; mathematically understanding the performance of the protocol in these large-scale RFID systems becomes essential. Recent works have established the importance of the Q-adaptive protocol, by analyzing its performance with the help of Markov-chain models. However, these analyses suffer from the common `state-space explosion' problem, as the state-space increases quadratically with an increasing number of participating tags. Hence it is essential to come up with a scalable analysis, whose computation model is insensitive to the number of tags. To this end, in this paper, we propose a scalable bound-based solution to study the delay performance of Q-adaptive protocol. We derive the delay bound by exploiting the fact that, in the large number of tags regime, the Q-adaptive protocol rapidly reaches to theoretical maximum performance and stays reasonably close in that optimal region for most of the time. Our extensive simulation results validate our bound-based solution with reasonable accuracy.