چكيده به لاتين
Massive multiple-input multiple-output (mMIMO) systems are considered as the key technology for high data traffic management in the next generation of cellular networks. Since mMIMO communications are destroyed or interrupted by the interference signals, in recent studies, more attention has been paid to fortify these communications against interferers. Some interference signals are unintentionally generated due to broadcast nature of the radio waves and the others are intentionally created in the MIMO systems. Intentional interference is generated either through the limitation of the resources, known as ``legitimate intentional interference'', or by a jammer reducing the spectral efficiency (SE) which is known as ``illegitimate intentional interference''. This work studies the management of intentional interference in mMIMO systems for both legitimate and illegitimate forms. Intentional interference at the pilot phase is known as ``pilot contamination''. The SE of mMIMO system is saturated when the pilot contamination exists in the system.
In this thesis, the management of the illegitimate intentional interference is considered in two steps: Jamming detection and jamming suppression. For jamming detection, a new method is proposed based on a generalized likelihood ratio test (GLRT), exploiting intentionally unused pilot sequences in the network over some coherence blocks. The performance of the proposed detector improves by increasing the number of antennas at the base station, the number of unused pilots and also by the number of the coherence blocks that are utilized. Simulation results confirm our analyses and show that in the mMIMO regime, perfect detection is achievable even with small number of the unused pilots. For jamming suppression, a new framework is proposed including a novel minimum mean-squared error based jamming suppression (MMSE-JS) estimator for channel training and a linear zero-forcing jamming suppression (ZFJS) detector for uplink combining. The MMSE-JS exploits some intentionally unused pilots to reduce the pilot contamination caused by the jammer. The ZFJS suppresses the jamming interference during the detection of the legitimate users' data symbols. The analysis shows that the proposed framework is robust against jamming attacks, so that its SE is about 42% rather than the SE of the framework presented in the literature.
The former works are proposed for uncorrelated mMIMO systems. Since the channels might be correlated, in this thesis, jamming suppression is also considered for spatially correlated mMIMO systems. To this end, a novel framework is proposed including a new optimal linear estimator in the training phase and a bilinear equalizer in the data phase. The proposed estimator is optimal in the sense of the SE of the legitimate system attacked by a jammer. It is demonstrated that optimized power allocation at the legitimate
users can improve the performance of the proposed framework, and the gain remains even if the jammer optimally adjusts its power. The proposed framework can be exploited to combat jamming in scenarios with either ideal or non-ideal hardware at the legitimate users and the jammer. It is shown that the proposed framework performs 6 times better than the conventional MMSE-ZF framework.
Furthermore, in this work, two adaptive algorithms based on the normalized least mean
square (NLMS) and the recursive least square (RLS) are suggested for channel estimation, with the goal of legitimate intentional interference management in uncorrelated mMIMO system. The implementations of the proposed algorithms do not require orthogonal pilot sequences. While the NLMS has lower complexity and also gives an acceptable SE in channels with low attenuation, the RLS is more complex and provides the desirable SE in mMIMO system. In simulations, the performances of the proposed algorithms are evaluated and compared with the performance of the conventional MMSE estimation. Although the SE of the system equipped with the MMSE estimator is saturated in the presence of the legitimate pilot contamination, the SEs of the proposed algorithms unlimitedly grow.
