An Opportunistic Thresholding Detector for IoT Random Access in Massive MIMO

Introduction of internet of things (IoT) into the massive MIMO random access scheme has generated many complications. Specifically, the number of IoT devices is significantly greater than mobile broadband. On the other hand, their traffic is intermittent. Given the different nature of IoT traffic, on-off random access channel can model it accurately. An extension of this access model has been recently applied to massive MIMO scenarios where it was demonstrated to entail common support recovery in a jointly sparse multiple measurement vector (MMV) problem. Several sparse recovery algorithms have also been proposed. However, we are not particularly bothered by the choice of algorithm but pose a new question which has been previously overlooked. While measurements size is severely constrained and dictated by the environment and the physical nature of communication channel, which is out of our control, the number of antennas is at our disposal to increase at will. The question is can we relieve the burden on number of measurements by increasing number of antennas and still maintain a satisfactory recovery performance. The so-called trivial pursuit (TP) gives a positive answer for independent sensing matrices. Motivated by TP, we introduce a novel opportunistic thresholding detector (OTD) which exploits partial de-correlation of sensing matrices to improve support recovery performance with minimum modifications and thus keeping the low-complexity nature of the algorithm. Extensive simulations has corroborated the superior performance of OTD compared to ordinary thresholding.