Secure cell-free massive MIMO under active eavesdropping

As a result, securing CF-mMIMO networks against active eavesdropping has become a critical research priority. In Part 1, we focus specifically on pilot spoofing attacks, where a single active eavesdropper targets one legitimate user by transmitting forged pilot signal during uplink training. We formulate a power control problem that aims to maximize the signal-to-interferenceplus-noise ratio (SINR) of the target user (i.e., the user under attack), subject to: (i) an upper bound on the eavesdropper’s SINR, (ii) per-AP transmit power constraints, and (iii) quality-of-service (QoS) SINR requirements for other users.

This nonconvex optimization problem is solved using a path-following iterative algorithm. To further enhance secrecy, we propose a large-scale fading-based greedy AP selection strategy, which chooses a subset of APs for transmission to improve the secrecy spectral efficiency (SSE). We also introduce an efficient detection method to identify the presence of an active eavesdropper in the system.

Extensive simulations and analytical evaluations demonstrate that the proposed approach significantly improves secrecy rates compared to baseline methods. In Part 2, we extend our framework to a more challenging scenario involving multiple active eavesdroppers, each targeting a different legitimate user. We develop a joint AP selection and power allocation iii strategy that maximizes the sum SE of all legitimate users while ensuring that each attacked user achieves a positive secrecy rate.

This mixed-integer optimization problem is addressed using a low-complexity accelerated projected gradient (APG) algorithm. Additionally, we propose practical methods for detecting multiple active eavesdroppers and estimating their large-scale channel gains. Simulation results confirm that jointly optimizing power allocation and AP selection under secrecy constraints significantly enhances network security and throughput.

In summary, this thesis addresses secure transmission in CF-mMIMO systems for next-generation wireless networks by developing power control and AP selection techniques in the presence of active pilot spoofing attacks. The proposed solutions effectively improve secrecy performance while satisfying the high-rate, reliable communication demands of future networks.<br>