Secure Low-power IoT Uplink Communication via Unsupervised Signal Alignment

This paper considers the problem of unsupervised detection in IoT uplink communication over fading channels with multiple antennas at the access point (AP). Instead of requiring long training sequences and coordination between the sensor nodes for training or power control, this work proposes using a simple transmission scheme consisting of repetition coding followed by random interleaving together with a signal alignment strategy at the AP. The key innovation is that pseudo-random (de-)interleaving is used here to simultaneously align the signal of interest, misalign (scramble) the interfering signals, and provide physical layer security. The combination allows reliable detection of the signal of interest at the AP via canonical correlation analysis, even under low-power / high interference conditions. The end-to-end approach provides rigorous recovery guarantees, allows for secure communication between the sensor nodes and the AP, and is computationally cheap. Simulations demonstrate the efficacy of the proposed approach compared to widely used multiplexing methods.