Research On Wireless Transmission Technique For Secure Spatial Modulation

To meet the requirement of the rapid growth in mobile communication services recently,the wireless communication that could endure the internet of everything requires the high data transmission supporting the peak rate up to 10 Gbps per second and guarantees less than1 ms delay in local area networks.As a powerful technique to improve the transmission rate,multiple-input multiple-out(MIMO)technique has attracted tremendous attention in the past 20 years.However,it suffers from the drawbacks of the expensive hardware cost,the elimination of channel interferences,the low energy efficiency and the accurately estimation of channels,as the number of the transmit antennas(TAs)tends to be large-scale.Due to these predicaments,MIMO technique is restricted on the application for mobile devices.Spatial modulation(SM),one of the derivatives in the context of MIMO techniques that improves the spectrum efficiency upon using both the transmit antenna index and the modulated symbol to convey binary bits,where only one radio frequency(RF)chain is equipped at the transmitter(Alice).Compared to MIMO technique,SM plays a crucial role in balancing the transmit rate and the hardware cost.On the other hand,the broadcast nature of the wireless communication incurs that the desired receiver is vulnerable to hostile users,thus establishing a set of systematically secure transmission strategies becomes an imperative demand.To this end,this thesis aims for enhancing the anti-risk capability for SM-based networks by designing a class of effective methods from the physical layer security(PLS)perspective,in which Rayleigh fading channel is taken into account.More explicitly,these anti-eavesdropping strategies include the transmit antenna selection(TAS),power allocation(PA),as well as the design of both the precoder and the AN beamformer.The main contributions of this work can be listed as follows:(1)In order to improve the secrecy rate(SR)performance of SM systems with finitealphabet inputs,three low-complexity schemes are proposed for solving the TAS problem.In the low signal-to-noise ratio(SNR)region,we analyze the asymptotic characteristic of the ergodic SR performance and find that,the optimum TA set for maximizing the SR performance is to choose the one having the maximum difference between the channel qualities of the intend receiver(Bob)and the unintended receiver(Eve).Correspondingly,another scheme is also proposed to reduce the computation complexity of selecting the TA set for the high-SNR region,where the objective TA set is the one having the maximum difference between the minimum Euclidean distance(Min-ED)of Bob and the minimum Euclidean distance of Eve.Furthermore,a compromise solution that suits for the whole SNR region is further devised,which has the capability of achieving a close SR performance when compared with the exhaustive search(ES)method.Simulation results show that the three proposed methods significantly reduce the computational complexity,and harvest appreciable SR gains compared with the traditionally Euclidean distance based TAS method.(2)As the total power is constrained,elaborately splitting the power allocated to the confidential message and allocated to the AN has a great impact on improving the SR performance.With this in mind,the cut-off rate is invoked to substitute the non-closed SR expression for secure SM systems,thus optimizing the PA factor becomes possible.It is worth mentioning that the cut-off rate based scheme is capable of achieving a similar profit as the original problem in terms of the SR performance attained.To obtain the optimum PA factor,we provide a pair of solutions: the iterative solution and the analytical solution.To elaborate,the iterative solution to the PA factor is arrived by formulating a concave maximization SR problem with the aid of the invoked cut-off rate.To further reduce the computational complexity,a novel criterion that maximizes the ratio of Bob’s signal-to-leakage-noise ratio(SLNR)and of Eve’s AN-to-leakage plus noise ratio(ANLNR)is conceived.Upon carefully analyzing the obtained solutions,only one is eventually confirmed as the feasible solution to the optimum PA factor.Simulation results reveal that the achievable SR performances of the two proposed schemes are close to that of the one-dimensional search method but with considerably low complexities.(3)When a rough channel state information(CSI)of Eve’s channel is obtained at Alice,we study the effect of both TAS and AN beamformer on the SR performance,where a pair of solutions,joint design and separate design,are provided to solve this problem.To be proceeded,a glorious alternative is invoked to substitute the non-closed SR expression.It should be noted that the TAS is a integer programming problem while the design of the AN beamformer is a continuous optimization problem,hence obtaining the optimal solution of this mixed programming problem is generally intractable.In order to address this difficulty,the simulated annealing(SA)mechanism and a convex optimization technique are invoked to manage the TAS and to optimize the AN beamformer,respectively.Compared to the separate design,the joint design performs close to the ES plus gradient descent(GD)method in terms of the SR performance but imposes a considerably high computational complexity.That means,the separate design strikes a performance versus complexity tradeoff.Moreover,an interesting conclusion that maximizing the SR performance can be reduced to a ratio maximization problem between Bob’s signal-to-interference-noise ratio(SINR)and Eve’s SINR,is derived when the number of TAs tends to infinite.Correspondingly,a Dinkelbach-based algorithm is proposed to solve this optimization problem within a few number of iterations.Finally,simulation results confirm the efficiency of the proposed schemes in terms of gathering the SR profit.(4)In precoding-aided SM networks,we investigate the security performance affected by the AN covariance matrix(ANCM),where a multiple-antenna Eve is taken into account.To improve the SR performance,we devise an iterative solution to the ANCM.Explicitly,we first analyze the property of the SR expression when a rough CSI of Eve’s channel is acquired at Alice.Then,a closed form which can be regarded as a lower bound on the SR expression is derived.Notably,the obtained solution based on this derived bound is a strictly suboptimal solution due to the approximation error.To solve this non-concave optimization problem over the ANCM,several slack variables are invoked to handle this problem,where both the non-concave objective function and the non-convex constraint are converted into concave ones.According to the simulation results,the constructed optimization algorithm is capable of approaching the GD-based method in terms of the SR performance attained.(5)For hybrid generalized SM(Gen SM)systems,we present a PLS scheme to enhance the SR performance when a statistical CSI of Eve’s channel is acquired at Alice.Considering the fact that both the analog precoder(Ana-P)and the digital precoder(Dig-P)play nonnegligible roles in improving the security,hence we propose a novel algorithm to separately optimize these two variables because of the predicament of the excessive computational complexity of the joint optimization.As the objective function is non-concave,we conceive a pair of new objective functions to optimize the Ana-P and the Dig-P,respectively.For the former,a lower bound on the derived objective function is conceived to optimize the module-one variables of the analog precoder with the aid of invoking a rank-one matrix.As for the latter,an elongated vector is further devised to substitute a large number of digital precoding vectors by extracting those effective elements of the sparse Dig-Ps.Simulation results suggest that the proposed alternating iterative method has a compelling ability in boosting the security performance.