Random Matrix Based Physical Layer Security Technologies And Application

Physical layer security is a promising approach to information confidentiality.The cooperation between physical layer security and modern cryptography achieves comprehensive information confidentiality from the physical layer to the upper layers.Signal processing is used to improve the difficulty of the decoding ability of eavesdroppers,then,Wyner coding(physical layer security coding)eliminates the few reserved abilities of decoding at eavesdropper sides,which is the general idea of physical layer security.For signal processing methods,the artificial noise(AN)technologies can improve the secrecy capacity as they do not interfere with desired users but only impair eavesdropped channels.However,the ANN technologies have a requirement that the number of transmit antennas is larger than these of receivers to generate nullspaces for transmitted AN signals,and the dimensionality of AN spaces is not optimized.Hence,the secrecy capacity of ANN is not large enough.For Wyner coding,the high dynamic characteristics of 5G networks make the wireless channels change rapidly,thus,the fixed Wyner code rates will reduce the transmission efficiency of physical layer security communications and result in a large secrecy outage probability.To address the aforementioned problems,we design a novel AN technology based on statistical optimization,then,use the proposed AN technology for secure communications in MEC,and design corresponding Wyner coding for data dissemination of offloading of computation tasks and downloading of results.The AN technology based on eigenspaces of Wishart matrices of channels(ANEW):messages are encoded in the first to the s1-th strongest eigenspaces based on ordered eigenvalues of Wishart matrices of legitimate channels,while AN signals are generated in remaining {Nt-s1} spaces(Ntis the number of transmit antennas).A part of eigenspaces reserved for messages can be used for AN signals,instead of that only nullspaces are selected for AN signals in ANN.We deduce the expression of ergodic secrecy rates under uncorrelated MIMO Rayleigh fading based on m PDF of the first to the s1-th largest eigenvalues of an uncorrelated central Wishart matrix,and the expression of ergodic secrecy rates is a function with respect to s1.We further optimize s1 in the natural number space with the objective of maximization of ergodic secrecy rates.The simulations show that the proposed ANEW has a better ergodic secrecy rate than ANN.ANEW under receiving-side-correlated MIMO Rayleigh fading: the correlations among receiving-side-antennas in legitimate receivers(eavesdroppers)usually exist when the antenna space at receiving-side-antennas is limited or the scatters are not enough.As ANEW is statistical optimization,whose core is the objective function of ergodic secrecy rates,we specify the characteristics of the random model of the row-correlated complex Gaussian matrix and Wishart matrix,then deduce the expression of ergodic secrecy rates under receiving-side-correlated MIMO Rayleigh fading based on this random model,which has the variance s1.We also optimize s1 with the objective of maximization of ergodic secrecy rates.In addition,we identify the properties of ergodic secrecy rates in terms of spatial correlation parameters,such as Ao A and antenna spacing.The simulations also show that the proposed ANEW has a better ergodic secrecy rate than ANN,and the receiving-side-correlation of eavesdroppers increases the ergodic secrecy rates.Secure transmission in MEC based on ANEW: Computation offloading in MEC can reduce computing latency of users,while keeping the information delivered in offloading and feedback phases confidential by an ANEW-assisted physical layer security transmission technology.Existing offloading schemes usually assume the wireless channels are time-invariant where computation task partition does not consider time-variant secrecy rates affected by changing channels,thus the ratio of task partition between local computing and edge computing is not reasonable.We utilize the ergodic secrecy rates instead of secrecy rates of the initial time in offloading processes to determine how many tasks are offloaded to the edge,where ergodic secrecy rates are the average secrecy metric over all transmission bursts.During transmission phases,the delivery of large amounts of data encoded by Wyner codes is through multiple fading distortions of wireless channels.To overcome the time-varying characteristic,the concept of effective secrecy rates is presented to adaptively adjust the Wyner code rate that can improve effective secrecy rates with the constraint of a secrecy outage probability.In addition,we present an SBS-assisted AN scheme to protect the uplink in the case of single-antenna users that are difficult to generate AN signals.Simulations demonstrate that the computation task partition based on ergodic secrecy rates can provide lower latency for users compared to the schemes based on secrecy rates of the initial time in offloading processes.