Research On Intelligent Reflecting Surface Assisted Wireless Communication Technology

In recent years,intelligent reflecting surface(IRS)regarded as a revolutionary technology has drawn a great amount of attention in both industry and academia due to its capability of remarkably improving spectral and energy efficiency in communication systems.IRS is a man-made metasurface composed of numerous low-cost passive elements,each of which can reflect the incident signal with independent amplitude offset and phase shift adjusted by the smart controller.By appropriately tuning the passive phase shifts according to the wireless environment,signals reflected by the IRS can be added constructively/destructively with those directly transmitted by the base station to strengthen/surpass the received signal power at the target user.However,the high IRS phase shift matrix design complexity erodes its practicability.In this thesis,joint active base station beamforming and IRS phase shift matrix optimization algorithms are developed so as to achieve near optimal performance in IRS-assisted communication systems with low complexity.Firstly,the IRS-aided multiple-input-single-output(MISO)transmission system is investigated.An efficient deep reinforcement learning(DRL)based algorithm jointly optimizing the active base station beamforming and IRS phase shift matrix based on the received signal-to-noise ratio(SNR)maximization criterion is proposed.In the developed algorithm,the optimal transmit beamforming can be acquired using maximum ratio transmission(MRT),while the IRS phase shift matrix is optimized exploiting the DRL approach.Assuming the received SNR and IRS passive phase shifts attained at the previous time step as state,the IRS passive phase shifts and the received SNR currently as action and reward,respectively,the deep neural network(DNN)parameters are updated employing the deep deterministic policy gradient(DDPG)algorithm.Numerical results demonstrate that the DRL based algorithm can achieve near optimal performance with low time complexity.Next,the IRS-aided downlink secure transmission system is investigated.Based on secrecy rate(SR)maximization criterion,an efficient alternating optimization(AO)based algorithm jointly designing the active base station beamforming and IRS phase shift matrix is developed.Specifically,the optimal active base station beamforming under fixed IRS phase shifts can be determined in closed-form using the generalized Rayleigh quotient theorem.Then,a quadratic transform(QT)is exploited to decouple the optimization variable from the numerator and denominator of the objective function.Meanwhile,the non-convex IRS phase shifts optimization problem is converted into a more solvable quadratic programming(QP)form,which can be addressed leveraging low complexity manifold optimization(MO)or alternating direction method of multipliers(ADMM)algorithm.Simulation results reveal that the proposed algorithm can almost achieve the performance upper bound with low time complexity.Furthermore,it is proved that transmitting artificial noise(AN)combined with the source signal is meaningless when only one eavesdropper is activated in the system.Finally,the IRS-enabled wideband orthogonal frequency division multiplexing(OFDM)system is investigated.A low-complexity AO based joint active base station beamforming and IRS phase shift matrix optimization algorithm is proposed based on the system achievable rate maximization criterion.Different from narrowband systems,in wideband systems,the IRS phase shifts need to cater to all signal paths at different gains,for which the underlying optimization problem is thus more challenging to solve.In the proposed algorithm,the optimal closed-form transmit beamforming vectors corresponding to different subcarriers can be attained exploiting the Cauchy-Schwarz inequality and water filling algorithm under fixed IRS phase shifts.Sequentially,under fixed active beamforming vectors,IRS phase shift matrix can be effectively optimized via fractional programming(FP),MO or majorization-minimization(MM)techniques.To further lower the required iterations,a heuristic IRS phase shift matrix optimization algorithm based on the sum of subcarrier gain maximization(SSGM)criterion is proposed.Simulation results unveil that both proposed algorithms can almost achieve the performance upper bound with low complexity.