Physically Securing Energy-Based Massive MIMO MAC via Joint Alignment of Multi-User Constellations and Artificial Noise

This paper investigates the artificial noise (AN) aided physical layer security (PLS) for energy-based massive multi-input multi-output multi-access channels with finite alphabet data inputs, where both the legitimate base station (Bob) and the passive eavesdropper (Eve) are equipped with large antenna arrays and each user has multiple antennas. For such system, the main challenge is how to allocate power between the transmitted constellation of finite size and AN for both interference management and PLS enhancement. To this end, we first characterize a distance-optimal (DO) constellation structure that maximizes the minimum Euclidean distance of the received signals with energy detection. It turns out that the DO multiuser constellations constitute a commonly-used pulse amplitude modulation (PAM) constellation at the receiver side. Then, we specifically design an energy-efficient PAM-constellation alignment to form a received sizeable PAM constellation by adaptively aligning the power of multi-users' PAM sub-constellations. This design provides us with a new finite-alphabet non-orthogonal multi-access scheme as well as with a stepped water filling (SWF) power allocation between the constellation and AN for each user, which can provide the power-domain freedom for PLS. To spatially exploit this freedom, a SWF-based product-constant AN generation algorithm is developed such that the product of Bob's channel and the AN is a constant which can be inferred by Bob, and the channel-noise product for Eve cannot be uniquely determined, even for noise-free channels. Simulations indicate that for SWF, Bob's error rate will vanish when the number of the receiver antennas goes to large, but Eve's error rate has a non-vanishing lower-bound, even with an unlimited number of antennas. In addition, PAMA, as a new finite-alphabet NOMA scheme, has significant advantages in both security and communication error performance over the time division multiaccess.