Two-Stage Multiple Access for Many Devices of Unique Identifications over Frequency-Selective Fading Channels

In this paper, we consider sparse index multiple access for uplink random access in a wireless system of a number of devices when a fraction of them are active. This multiple access scheme is suitable for the case that an access point (AP) needs not only to receive data symbols, but also to identify active devices when there are a number of devices with unique identification sequences (the number of devices can be easily more than a million) with low signaling/control overhead. We propose a two-stage transmission scheme for random access and derive computationally efficient methods to estimate the channel state information (CSI) of active devices over frequency-selective fading channels in the first stage and to perform joint active device identification and data detection in the second stage using a well-known sparse signal estimation method in compressive sensing. Simulation results demonstrate that the proposed approach can successfully estimate the CSI of active devices under reasonable conditions and identify the unique identification sequences or vectors of active devices with a high probability. For example, when 6 out of 64 devices become active, the AP can identify all six devices (using estimated CSI) with a probability higher than 1 - 10-2over frequency-selective fading channels.