Quick discovery of mobile devices in the many-user regime — carrier sensing or simultaneous detection?

We consider the problem of detecting the active wireless stations among a very large population. This problem is highly relevant in applications involving passive and active RFID tags and dense IoT settings. The state of the art mainly utilizes interference avoiding (e.g., CSMA-based) approaches with the objective of identifying one station at a time. We first derive basic limits of the achievable delay with interference avoiding paradigm. Then, we consider the setting in which each station is assigned a signature sequence, picked at random from a specific alphabet and active stations transmit their signatures simultaneously upon activation. The challenge at the detector is to detect all active stations from the combined signature signal with low probability of misdetection and false positives. We show that, such an interference embracing approach can substantially reduce the detection delay, at an arbitrarily low probability of both types of detection errors, as the number of stations scale. We show that, under a randomized activation model the collision embracing detection scheme achieves Θ(log2(n)/log(log(n))) delay while the expected delay of existing CSMA schemes are Ω(log2(n)) for a population of n stations. Finally, we discuss large-scale implementation issues such as the design of low-complexity detection schemes and present numerical investigations.