Security enhancements for sparse MIMO systems: A compressive sensing - artificial noise technique

In wireless communication, it is possible to achieve high level of security by exploiting the random characteristics of wireless channels and the transmitted signal structure.Thus, applying security at physical layer gained much attention since this randomness can be utilized to provide infinite set of secret keys; unlike cryptographic techniques that are applied at the network layer and are limited by the finite set of available keys.Not only this, but also the emerging types of networks pose restrictions on applying security at higher level due to the light computational abilities of the devices used such as sensors, radio-frequency identification (RFID) tags and so forth.Thus, physical layer security (PLS) provides a way to utilize the physics of radio propagation to secure such networks.Furthermore, sparsity, that exists in many types of signals, can be used to achieve sampling and compression efficiently.In this thesis, we consider physical layer security through combining both compressive sensing (CS) and artificial noise (AN) to elevate secrecy of MIMO communication systems.Contrary to the classical methods of CS-PLS, this technique does not impose any restriction on the adversary except one.It is assumed that both the adversary and the legitimate receiver have access to the same information with only one imposed postulation that the adversary should possess fewer antennas than the transmitter.This is a valid assumption in many applications that include a powerful base station.First, we lay down the theoretical foundation for physical layer security with emphasis on both compressive sensing and artificial noise.Then, we modify the CS system model to work in MIMO environment to be able to include AN, where the latter requires a MIMO layout to guarantee the presence of a valid null space to inject the artificial noise to the system.Two methods are considered to perform such modification; the first one is CS by repetition and the other one is CS on Air.Furthermore, different techniques are used to inject AN to the CS-based system and the security performance of each method is evaluated.The difference between these techniques lies in the method of creating the artificial noise and where it is injected.Moreover, the secrecy performance is enhanced by using beamforming to direct the information-bearing signal towards the intended receiver and degrades it in other directions.Finally, we investigate the effect of choosing the perturbation parameter (𝜖), due to the presence of AWGN and AN, in ℓ 1 -minimization algorithms on the recovery performance at the eavesdropper and consequently the achieved secrecy.𝜖 is related to the noise level and decides the accuracy of recovering the transmitted signal.Simulation results show that using Abstract iii beamforming, when combining CS and AN, enhances secrecy considerably while imposing no assumptions on the confidentiality of information from the eavesdropper including CSI, seeds (used for generating random keys) or keys.Furthermore, they show that the choice of 𝜖 significantly affects secrecy since it cannot be correctly estimated at the adversary and hence increases the errors in ℓ 1 -minimization techniques that are used for signal reconstruction. List of Notations• ( • ) 𝑇 , ( • ) 𝐻 and ( • ) + refer to matrix transpose, conjugate transpose (Hermitian transpose) and pseudo-inverse, respectively.• 𝑡𝑟( • ) and 𝑑𝑒𝑡 ( • ) denote the trace and determinant of a matrix, respectively.• 𝔼{ • } refers to the mathematical expectation.• | • | returns magnitude of a complex number.• ‖ • ‖ returns the frobenius norm.• ‖ • ‖ 0 counts the number of nonzero entries of its argument (ℓ 0 -norm).• ‖ • ‖ 1 and ‖ • ‖ 2 refer to ℓ 1 -norm and ℓ 2 -norm, respectively.• ℜ{ • } and ℑ{ • } return the real and imaginary parts of its complex argument, respectively.• Pr(𝐴) is the probability of event 𝐴.• 𝑃 𝑒 is the probability of error.• 𝑓( • ) refers to random mapping caused by channel impairments on its argument.• 𝐻( • ) and 𝐻( • | • ) denote the entropy and the equivocation (conditional entropy), respectively.• 𝐼( • ; • ) denotes the mutual information.• 𝑝 𝐴 denotes the probability distribution of 𝐴.• 𝑝 𝐴𝐵 denotes the joint probability distribution of 𝐴 and 𝐵. List of Notations xiv• ≼ denotes "less or equal to" in the positive semidefinite partial ordering between real symmetric matrices.• Alice, Bob and Eve are the names given to transmitter, legitimate receiver and Eavesdropper, respectively.• Bold upper-case variables refer to matrices.• Bold lower-case variables refer to vectors.• Upper-case variables represent real-valued numbers.• Lower-case non-bold variables refer to entries related to certain vectors or matrices.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.