Helping interferer physical layer security strategies for M-QAM and M-PSK systems

Physical layer security encompasses information theoretic approaches that could guarantee perfect secrecy in wireless communication systems. In this framework, helping interferer strategies rely on intentionally creating confusion at a potential eavesdropper by injecting a jamming signal. In cases where the information signal has a Gaussian probability density function (pdf) it has been demonstrated that the optimal jamming signal, under an overall power constraint, should also be Gaussian. However, in practical communication systems where data symbols are typically drawn from discrete uniform probability mass functions (pmf), commonly M-ary Quadrature Amplitude and M-ary Phase Shift Keying modulation schemes, the structure of the optimal jamming signal is still an open question. In the present work we aim at shedding light into this question. Our approach is based on formulating a secrecy capacity maximization problem by expressing the optimal arbitrary helping interferer pdf as a mixture of unknown Gaussians. The proposed approximation is well-suited for jamming signals of practical interest, i.e. Gaussian or M-QAM interferers and reveals that in certain scenarios it is advantageous to use jamming signals whose statistical structure resembles the data rather than the noise.