The Research Of Key Tecenologies On Full-duplex Radio Frequency Wireless Communications Systems

Considering the current shortage of spectrum resources, with the novel self-Interference (SI) cancellation technologies which have been proposed by researches, full duplex technology gradually played its potential. However, due to the strict of the SI cancellation technologies and the existence of the residual SI, there are still many challenges in the application of full duplex technology. In order to make full use of full duplex SI cancellations, and optimize the performance of full duplex systems with residual SI, this paper mainly studies the following contents:1.For limited radio frequency SI isolation full-duplex systems with single tap analog cancellation structure, this paper considers the joint impairments of the circuit components of the full-duplex transceiver in limited RF isolation systems as well as the spatial propagation characteristics of SI signals, and presents a complete behavioral model of SI signals which go through the RF isolation to digital domain. According to the new model of the SI, this paper proposes an adaptive filter based widely-nonlinear digital cancellation scheme. The non-orthogonal basis function based on amplification factor has been first used, the responses of the basis function have been proven to be the minimum variance unbiased (MVU) estimate which satisfies the Cramer-Rao lower bound.On this foundation, the close-form expressions of the powers of minimum estimation errors have also been derived in detail. According to the simulation results, when the amplification factor satisfy AF≥ 3 ,parameter learning comes to stable and has the minimum errors, then digital SI cancellation ability will approach the upper bound.2.In order to find the optimal training length of the training-based schemes for full duplex links with non-coherent channels, we compute a capacity lower bound of a MIMO FD channel as a function of the received signal-to-noise ratio p, fading coherence time T, and numbers of link antennas M ,N, and maximize the bound as the training symbols matrix ST. Numerical results show that if T >> max(M,N), the length of training symbols satisfy T0 ≥2max(M,N) with appropriate ρ.Conversely, if T ≈ 2max(M,N), then T0 < max (M,N).3.By combing the adaptive modulations, the adaptive threshold estimation algorithm for discrete variable rate M-ary quadrature amplitude modulation (VR-MQAM) has been proposed in this paper.Firstly, detailed FD system modeling is carried out, taking into account the differences from the half-duplex systems on rate adaptations. Then, in this model, we obtain the closed-form expression of the VR-MQAM spectral efficiency by deriving the distribution of the received signal to interference and noise ratio (SINR). Finally, an adaptive threshold estimation scheme for VR-MQAM, which can adapt to the variation of imperfect CSI, has been proposed. Numerical results reveal that,considering the effects of the residual SI, comparing to the existing variable threshold scheme, our adaptive scheme improves the spectral efficiency without any extra complexity.