Research On Channel Estimation And Resource Allocation For Distributed Massive MIMO

Distributed massive multiple-input multiple-output(MIMO)is one of the most promising physical layer transmission technologies in the 5th generation mobile communication systems(5G),which combines mas-sive MIMO and distributed antenna structure to significantly improve spectral efficiency and energy effi-ciency.However,distributed massive MIMO also faces many challenges.On the one hand,the wireless transmission performance of distributed massive MIMO depends on the the accuracy of channel state in-formation(CSI)acquisition.By exploiting the reciprocity between uplink and downlink channels during time division duplexing(TDD)mode,distributed massive MIMO with fully digital structure can estimate CSI through uplink pilot estimation,and the pilot overhead is proportional to the total number of antennas at users.However,due to the high cost and power consumption,fully digital structure is not applicable to distributed massive MIMO systems.Hybrid structure provides a promising solution,but the corresponding subspace sampling limitation poses huge challenge to channel estimation with low pilot overhead.On the other hand,the wireless transmission performance of distributed massive MIMO also relies on reasonable resource allocation,especially when considering physical layer security requirements.Due to the broadcast property of wireless channels,the private data of users may be wiretapped by some malicious eavesdrop-pers during uplink and downlink transmission,leading to privacy leakage.In traditional colocated massive MIMO systems,designing artificial noise(AN)shemes and optimizing power allocation can largely suppress the wiretapping rate,but it also causes interference to both uplink transmission and downlink reception.In distributed massive MIMO systems,directly applying such AN shemes will cause more severe interference.Hence,this dissertation focuses on the research on channel estimation and resource allocation for distributed massive MIMO.The main innovations and contributions of this dissertation are listed as follows:Firstly,we propose tensor decomposition based channel estimation method for broadband distributed hybrid massive MIMO systems which employ orthogonal frequency division multiplexing(OFDM)modu-lation.Considering transmission delay,angle of arrival(AOA),angle of departure(AOD),doppler shift and path gains,we build broadband sparse multipath parametric channel model and received pilot signal model.Based on the received pilot signal,we then propose a fourth-order structured CANDECOMP/PARAFAC de-composition(SCPD)based channel estimation method.Specifically,we first use the pilot orthogonality to eliminate inter-user interference,and transform the received pilot signal into fourth-order observation ten-sor.Then we find the noiseless observation tensor admits CP decomposition and two factor matrices have Vandermonde structures.Next we derive a sufficient uniqueness condition for the decomposition of noise-less observation tensor,which guides us to design a fourth-order SCPD based channel estimation method for noiseless case.Finally,we consider the effect of received noise,and propose a fourth-order SCPD based chan-nel estimation method for noisy case.Simulation results show that the proposed algorithm achieves better performance than existing tensor decomposition based channel estimation algorithms.Secondly,we propose two privacy-preserving channel estimation methods for distributed hybrid mas-sive MIMO systems with low pilot overhead.We build uplink transmission model to obtain received signal matrix with missing entries,and analyse that traditional channel estimation scheme may leak the location privacy of users.To reduce the pilot overhead,we exploit the low-rank property of noiseless received signal matrix to propose a matrix completion based channel estimation method.We show that the key component of the privacy-preserving channel estimator is privacy-preserving matrix completion,and propose a global-local computation model for privacy-preserving matrix completion.Based on the model,we propose two privacy-preserving matrix completion algorithms and compare their computation complexity and communi-cation overhead.We show that both channel estimation algorithms are joint differentially private.We also rigorously analyze the estimation error bounds for the two algorithms,and conclude that the estimation error decreases with the increase of data payload size.Simulation results verify this conclusion and illustrate the tradeoff between privacy and channel estimation performance for the two algorithms.Thirdly,we propose an efficient algorithm to maximize the global average secrecy energy efficiency(GASEE)during downlink secure transmission for distributed massive MIMO systems.We derive closed-form expressions for the downlink achievable secrecy rate,and build a power consumption model that in-corporates transmit power,backhaul power,remote antenna unit(RAU)circuit and signal processing power.Then we consider transmit power constraints as well as SINR constraints for both users and the Eve,and jointly optimize power allocation,RAU clustering,RAU selection and AN selection to maximize the GASEE.To solve this problem,we design a double-loop procedure.In the outer loop,the denominator of objective is approximated as a linear function by reweightedl₁norm.In the inner loop,an efficient algorithm is proposed to find a near-optimal solution to the approximated problem by solving a sequence of sub-problems.Simula-tion results demonstrate that the proposed algorithm converges fast and achieves a higher GASEE than some heuristics.Finally,we propose two efficient algorithms to respectively maximize uplink total secrecy offloading data(TSOD)and secrecy energy efficiency(SEE)during uplink secure transmission for mobile edge comput-ing(MEC)-enabled distributed massive MIMO systems.To guarantee secure uplink computation offloading,we respectively propose single-antenna jamming scheme and multi-antenna jamming scheme,and prove that the multi-antenna jamming scheme can suppress the wiretapping rate while does not cause interference to uplink offloading by designing appropriate AN precoder.Based on the expression of uplink secrecy rate in multi-antenna jamming scheme and computation model,we formulate two optimization problems,which jointly optimize power resource,computation resource allocation and offloading ratio to respectively maxi-mize the TSOD and SEE under the requirements of offloading rate,wiretapping rate,delay,power consump-tion and computation capacity.To solve this two problems,we first apply block coordinate descent(BCD)method to decompose them into three subproblems:power resource allocation,computation resource allo-cation and offloading ratio optimization.Then we propose efficient algorithms to solve them respectively.Simulation results show that the proposed jamming scheme can achieve higher TSOD and SEE.