Research On Power Allocation In Full-duplex Wireless Communication Network

The fifth generation mobile communication will provide higher data rate and better QoS.However,to achieve higher capacity in the limited spectrum resource,the spectrum have to be utilized more efficiently.In the current mobile communication networks,the system always operates in the time-division duplex mode or frequency-division mode to prevent the interference between downlink and uplink.No matter which kinds of duplex operation is deployed,half of the spectrum resource always is wasted.The main target of full-duplex is to transmit and receive simultaneously with the same frequency,and all the spectrum resource will be fully utilized.The premise of the utilization of the full-duplex technique is that the device can successfully eliminate and mitigate the self-interference generated by the full-duplex transmission.The complete self-interference cancellation mechanism consists of three steps: self-interference mitigation in propagation domain,self-interference cancellation in analog domain,and self-interference cancellation in digital domain.Although the selfinterference cancellation and mitigation technology has guaranteed the applicability of the full-duplex,the non-idealities in the device such as the non-linearity in RF circuits and the phase noise in the oscillator will always result in the remnant of the self-interference.Besides,for the network with multiple full-duplex nodes,the co-channel interference can also cause a severe problem.Concerning these issues,this paper investigates the power allocation problem in fullduplex wireless networks.By designing the reasonable power allocation algorithms,the influence of the interference can be reduced and the system performance of full-duplex networks can be improved greatly.The relay network and the secure communication network are two typical application scenarios for the full-duplex technology.Towards this end,the power allocation in the full-duplex Decode-and-Forward(DF)relay network and the full-duplex secure network are handled and investigated in detail.For the full-duplex DF relay network,both single-and multi-carrier relay transmission schemes are considered.For the single carrier relay with single antenna,the endto-end outage probability is derived and chosen as the objective function of the power allocation problem.Considering the individual power and global power joint constraint,the optimal power allocation algorithm is achieved,and it is verified that the optimal allocation result locates at the boundary of the constraints.Additionally,a lower bound of the achieved outage probability is obtained.For the single-carrier relay with multiple antennas,the end-to-end capacity achieved by the match filter receiver is utilized as the cost function.Since the inter-node power allocation is coupled with the intra-node beamforming,to handle this problem,a inter-node phase and intra-node phase iterative allocation is designed.First,given the intra-node beamforming vectors,the optimal internode power allocation can be solved.Then,given the inter-node allocation result,a signal to leakage and noise ratio based transmitting beamforming is proposed,and according the Rayleigh-Ritz Ratio,the beamforming vectors with closed-form expression are obtained.Finally,iterate the two phase allocation until the convergence or the maximum iteration is achieved.For the full-duplex DF relay with multi-carrier,both the carrier-wise decode-forward(CDF)and group-wise decode-forward(GDF)are considered.For the CDF protocol,the optimal power allocation problems subject to total power and individual power constraints are shown to be equivalent to convex problems.The optimal power allocation results are obtained after solving the corresponding KKT(Karush-Kuhn-Tucker)conditions by bisection search based algorithms.The upper bound of system capacity at high SNR regime is obtained.Besides,the scenario without direct link and the problem to minimize the transmitting power are also investigated.For the GDF protocol,a source and relay iterative allocation method is developed to maximize the system capacity.In the source allocation or the relay allocation,the sequential convex programming is used to approximate the non-convex constraints to be convex by the first order Taylor expansion,then the non-convex optimization will approximate to convex one.For the full-duplex secure communication network with multi-carrier transmission,the full-duplex destination will transmit the artificial noise to the eavesdropper when receiving target signal from sources.The target of power allocation is to maximize the secrecy capacity.Due to the non-differential characteristic of the objective function,an equivalent differential expression is derived and utilized.A necessary condition of determining whether a subcarrier at destination needs to be allocated with nonzero power is proposed.Since the distributive property of the network topology,a source and destination iterative allocation strategy is deployed.Although both the source allocation and the relay allocation problems are non-convex when only considering the power constraint,combining the corresponding KKT conditions and the property of objective function,the optimal allocation results at source allocation and destination allocation are able to be obtained.When the target data rate is considered,the source allocation can be solved by two-dimensional bisection search.While,the destination allocation is non-convex at this scenario,and the sequential convex programming is able to be used to solve the locally optimal solution.Furthermore,the proposed power allocation algorithms can be expanded to multiple destination and multiple source topologies by the assist of recursive processing.The results in this work show that the performance of the full-duplex DF relay network and the full-duplex secure communication network can be greatly improved by deploying effective power allocation methods.Thus,for full-duplex wireless networks,power allocation is an indispensable technique to guarantee the system performance.