IEEE Quick Preview
  • Abstract

SECTION I

INTRODUCTION

The unprecedented growth in demand for mobile data and the advent of the Internet of Things (IoT) [1]–[2] [3] [4] [5][6] have driven extensive researches and developments on the fifth generation (5G) wireless communication system [7]–[8] [9] [10] [11][12]. 5G wireless network is expected to provide far more gigantic capacity than fourth generation (4G) technology to meet the ever-increasing users’ needs in the future. Massive multiple-input multiple-output (MIMO) [13]–[14] [15][16], which enables very high energy efficiency and spectral efficiency [17], [18] by exploiting a large number of excess service antennas at the base stations (BSs) and the law of large numbers, is one of the most promising and disruptive wireless technologies that can bring 5G into reality [19]–[20] [21][22]. Moreover, massive MIMO helps alleviate the burden of user equipments (UEs) greatly. For instance, only the BSs need to learn the channel state information (CSI) in time-division duplexing (TDD) massive MIMO systems [16]. This distinctive feature of massive MIMO potentially contributes to the promotion of the forthcoming IoT, in which the communication nodes would probably be lightweight, i.e., with limited number of antennas, limited power, low software and hardware complexity and low storage capacity. It is also certainly desirable that the portable UEs in 5G wireless cellular networks can reduce the cost on software and hardware, and have lower power consumption and longer life time while maintaining an acceptable performance by employing massive MIMO at the BSs.

On the other hand, information security is clearly a critical aspect for IoT and 5G network. However, from the perspective of information security, it seems unfavorable that the software and hardware of UEs are overly lightweight. Computational complexity based traditional cryptographic technologies [23] may need large signal processing and computational resource to ensure proper function of encryption, decryption, key management and distribution. Recently, physical layer security [24], [25], which is theoretically based on information theory, offers an alternative and complementary approach to strengthen the secrecy performance of wireless communication from the physical layer. However, although physical layer security circumvents the computational complexity problem of traditional cryptographic technologies, most of the existing physical layer security techniques are still too complex for the lightweight UEs to use directly [26]. Mukherjee [26] investigated the physical layer security techniques in IoT scenario and presented some methods that are suitable for IoT applications. These methods have the advantage of achieving the secrecy goal while being suitable for lightweight UEs. Nonetheless, it dose not take massive MIMO into account. The advantage can be further enhanced by exploiting massive MIMO and physical layer security together, and the combination of these two technologies benefits both the communication performance and secrecy performance [27]–[28][29]. However, adequate attention on the combination of massive MIMO and physical layer security is yet to be received at the time of writing.

Currently, the research pertaining to massive MIMO physical layer security mainly focuses on the downlink transmission. By performing a large-system analysis, [30] and [31] investigate the secrecy performance in downlink multi-cell MIMO scenario with the knowledge of the eavesdroppers’ CSI. Zhu et al. [32], [33] comprehensively examined the secrecy performance of matched-filter, zero-forcing, regularized channel inversion, traditional artificial noise (AN) [34] and proposed AN precoders on the downlink transmission in a multi-cell massive MIMO system. The power scaling feature of massive MIMO is exploited in [35]–[36][37] to study the secrecy rates on both uplink and downlink transmissions in a massive MIMO network without the aid of AN or jamming. The corresponding results implicitly present the inherent security nature of massive MIMO, and also demonstrate that the secrecy performance cannot be guaranteed if the number of eavesdropper antennas exceeds a certain limit. Instead of Rayleigh fading, Rician fading channels are studied in [38], and the corresponding downlink AN aided jamming design is investigated in detail. Uplink pilot training phase jamming attack on the massive MIMO downlink transmission is considered in [39], and a Formula$\delta $-conjugate beamforming method is proposed to secure the communication with the help of pilot encryption. In [40], passive and active eavesdroppers are both taken into account, and a simple protocol is proposed to detect the active adverse attack. In [41], downlink secure transmit signal design with AN over correlated fading channels in multi-cell massive MIMO scenario is investigated in the presence of an active multi-antenna eavesdropper, where the number of eavesdropper antennas is limited comparing with that of the BS. Secure transmission in a decode and forward massive MIMO relaying system is studied in [42] with the existence of a single-antenna eavesdropper. Secrecy performance in term of co-located or distributed placement of multi-antenna eavesdroppers is discussed in an uplink massive MIMO system [43], where the number of eavesdropper antennas is optimistically assumed to be fewer than the number of BS antennas. Motivated by the possibility that the eavesdropper may equip with much more powerful large antenna arrays, an original symbol phase rotated (OSPR) secure transmission scheme is recently proposed in [44] to defend against massive MIMO eavesdropper on the downlink transmission. To the best of the authors’ knowledge, no work except [44] has ever considered the security problem in a massive MIMO scenario where the number of eavesdropper antennas is far more than the number of BS antennas, i.e., a much more powerful massive MIMO eavesdropper.

It is worth emphasizing that the communication security of uplink transmission is equally important although relatively fewer attention has been given. Especially in IoT and 5G network, more and more lightweight UEs will be used everywhere, and private information is accordingly gathered and transmitted to the BSs or sink nodes through uplink transmission. Adequate attention should be paid to securing the uplink transmission since lightweight UEs are likely with limited number of antennas, limited power, low signal processing and data computing capabilities, which may inherently weaken their security level during the uplink communication stage even with massive MIMO employed at the BS. Moreover, it is also beneficial to take into account the worst case, which has received little attention, that the eavesdropper adopts much more number of antennas to intercept the information sent by the lightweight UEs, which will severely deteriorate the uplink secrecy performance. Overall, simple but effective massive MIMO physical layer security approaches for lightweight UEs on the uplink call for more research, since they are significantly beneficial in protecting the private information in future IoT and 5G systems.

In this paper, we consider the secure communication on the uplink transmission, during which multiple autonomous lightweight single-antenna UEs transmit confidential messages to a massive MIMO BS, in the presence of a much more powerful massive MIMO eavesdropper. It is highly plausible for the adversary to employ many more antennas than legitimate UEs and BSs as indicated in [45], particularly in 5G era in which the massive MIMO technology will be mature and popular. It breaks the optimistic assumption which has been made in most literatures that the eavesdropper always has not enough antennas to intercept. However, it is actually not difficult for the eavesdropper to increase its aperture by equipping more antennas in practical scenario [45]. To the best of our knowledge, no work has specifically involved such uplink scenario before. The basic idea of the OSPR secure transmission scheme proposed in [44] for downlink transmission is borrowed here to defend against the massive MIMO eavesdropper on the uplink transmission. We show that the OSPR scheme can be well extended to the uplink case after proper modifications. Hereafter, it is referred to as uplink OSPR (UOSPR) scheme in this work. We present that in addition to protecting the private information on the uplink, the UOSPR secure transmission scheme is well suitable for lightweight UEs. This is due to its low complexity requirement, which does not need powerful signal and data processing capabilities or extra jamming cost for the UEs to achieve their security goals. Our work show that by the employment of UOSPR scheme, the lightweight single-antenna UEs with limited power and processing capabilities can achieve a good secrecy performance, even while facing the eavesdropper equipped with an infinite number of antennas.

The remainder of this paper is organized as follows. In Section II, we introduce the uplink system model of the considered single-cell multi-user massive MIMO network with the existence of a powerful passive massive MIMO eavesdropper. In Section III, we present the UOSPR secure transmission scheme for the legitimate lightweight UEs and the corresponding secrecy analysis in detail. Simulation results are provided in Section IV, and then Section V concludes the paper.

Notation: Vectors and matrices are denoted by lowercase boldface letters and uppercase boldface letters, respectively. Formula${\mathbf {X}}^{T}$, Formula${\mathbf {X}}^{H}$ and Formula${\mathbf {X}}^{-1}$ stand for the transpose, conjugate transpose and inverse of matrix Formula${\mathbf {X}}$, respectively. Formula${\mathbf {I}}_{N}$ denotes the Formula$N\times N$ identity matrix. Formula${\left |{x}\right |}$, Formula${\left \|{ {\mathbf {x}} }\right \|}$ and Formula${\left \|{ {\mathbf {X}} }\right \|}$ denote the Euclidean norm of a scalar, a vector and a matrix, respectively. Formula${\mathrm {tr}}\left ({\mathbf {X}}\right )$ denotes the trace operation of matrix Formula${\mathbf {X}}$. Formula${\mathop {\mathrm {E}}}\left ({x}\right )$ and Formula${\mathop {\mathrm {Var}}}\left ({x}\right )$ are used to denote the expectation and variance of a random variable Formula$x$, respectively. Formula${\mathbb {C}^{m \times n}}$ denotes the field of all Formula$m \times n$ complex matrices. Formula$\mathbf {x}\sim \mathcal {CN}({\mathbf {0}}_{N},\boldsymbol {\Sigma })$ stands for a circularly symmetric complex Gaussian random vector Formula${\mathbf {x}} \in {\mathbb {C}^{N \times 1}}$ with zero mean and covariance matrix Formula$\boldsymbol {\Sigma }$.

SECTION II

UPLINK SYSTEM MODEL

In this section, we introduce the uplink channel model of the considered TDD-based single-cell multi-user massive MIMO system. Formula$K$ autonomous legitimate lightweight single-antenna UEs simultaneously send private information to the massive MIMO BS. We make the worst case assumption that the passive eavesdropper uses unlimited number of antennas to intercept the confidential messages on the uplink transmission. The massive MIMO BS has Formula$N_{b}$ antennas, and Formula$N_{e}$ denotes the number of antennas at the massive MIMO eavesdropper. It is clear that Formula${N_{e}}\gg {N_{b}}\gg {K}>{0}$, where Formula$N_{e}\rightarrow \infty $, Formula$N_{b}$ and Formula$K$ are finite. The corresponding system model is illustrated in Fig. 1.

Figure 1
Fig. 1. The considered uplink eavesdropping scenario.

A. Uplink Channel

The Formula$k$th autonomous lightweight single-antenna UE sends confidential information to the massive MIMO BS through the following uplink channel FormulaTeX Source\begin{equation} {{\mathbf {h}}_{k}} = \left [{ {\begin{array}{*{20}{c}} {{h_{k,1}}}\\ {{h_{k,2}}}\\ \vdots \\ {{h_{k,{N_{b}}}}} \end{array}} }\right ]\!, \end{equation} where Formula${h_{k,i}}$, Formula$k = 1,2, \ldots ,K$ and Formula$i = 1,2, \ldots ,N_{b}$ denotes the CSI from UE Formula$k$ to the Formula$i$th BS antenna.

In the meanwhile, the massive MIMO eavesdropper can intercept the private information transmitted by UE Formula$k$ through the following channel during uplink transmission FormulaTeX Source\begin{equation} {{\mathbf {g}}_{k}} = \left [{ {\begin{array}{*{20}{c}} {{g_{k,1}}}\\ {{g_{k,2}}}\\ \vdots \\ {{g_{k,{N_{e}}}}} \end{array}} }\right ]\!, \end{equation} where Formula${g_{k,j}}$, Formula$k = 1,2, \ldots ,K$ and Formula$j = 1,2, \ldots ,N_{e}$ denotes the CSI from UE Formula$k$ to the Formula$j$th eavesdropper antenna.

It is assumed that all the CSI entries given above are independent and identically distributed (i.i.d.) Formula$\mathcal {CN}(0,1)$ random variables, which implies that all the antenna spacing is no less than half a wavelength. That is the standard independent Rayleigh fading channel model. The CSI entries remain constant within a certain coherence interval length Formula$T$, and vary block by block, i.e., block fading. The downlink channel and uplink channel are reciprocal due to the employment of TDD mode, which is widely known as the concept ‘channel reciprocity’.

B. Uplink Transmission and Eavesdropping

Single-antenna UE Formula$k$ radiates the original information bearing symbol Formula$s_{k}$ omnidirectionally, independently and simultaneously on the uplink, where Formula$k = 1,2, \ldots ,K$ and Formula$s_{k}$ submits to the power constraint Formula${\mathop {\mathrm { E}}}\left ({ {{{\left \|{ {s_{k}} }\right \|}^{2}}} }\right ) \le P_{k}$. Then the received uplink signal of UE Formula$k$ at the massive MIMO BS can be written as FormulaTeX Source\begin{equation} {{\mathbf {r}}_{b}} = {{\mathbf {h}}_{k}}{s_{k}} + \sum \limits _{i = 1,i \ne k}^{K} {{{\mathbf {h}}_{i}}{s_{i}}} + {{\mathbf {n}}_{b}}, \end{equation} where Formula${\mathbf {r}}_{b}\in \mathbb {C}^{N_{b} \times 1}$ is the BS receiving signal vector, and the elements of the receiver noise Formula${{\mathbf {n}}_{b}}\in \mathbb {C}^{N_{b} \times 1}$ are assumed to be i.i.d. Formula$\mathcal {CN}(0,\sigma _{b}^{2})$ random variables. The first term on the right hand side of (3) represents the signal for UE Formula$k$, and the second term represents the interference which comes from other UEs’ signals.

The corresponding intercepted signal of UE Formula$k$ at the massive MIMO eavesdropper is given by FormulaTeX Source\begin{equation} {{\mathbf {r}}_{e}} = {{\mathbf {g}}_{k}}{s_{k}} + \sum \limits _{i = 1,i \ne k}^{K} {{{\mathbf {g}}_{i}}{s_{i}}} + {{\mathbf {n}}_{e}}, \end{equation} where Formula${\mathbf {r}}_{e}\in \mathbb {C}^{N_{e} \times 1}$ is the eavesdropping signal vector, and the elements of the receiver noise Formula${{\mathbf {n}}_{e}}\in \mathbb {C}^{N_{e} \times 1}$ are assumed to be i.i.d. Formula$\mathcal {CN}(0,\sigma _{e}^{2})$ random variables. The first term on the right hand side of (4) represents the signal intercepted of UE Formula$k$, and the second term represents the interference from other UEs. We consider the worst case that the massive MIMO eavesdropper is able to cancel all the interference and noise, and can extract the radiated signal of UE Formula$k$ from the first term perfectly.

SECTION III

SECURE THE UPLINK TRANSMISSION WITH UOSPR SCHEME

As described in Section II, we consider the uplink communication security problem of defending against a passive eavesdropper equipped with infinite number of antennas. The UOSPR scheme is adopted to help achieve a good secrecy performance for the lightweight single-antenna UEs, which are probably with limited power and weak signal and data processing capabilities. The basic idea of the employed UOSPR approach is to replace the original information bearing symbols transmitted by UE Formula$k$ with randomly rotated symbols. After the massive MIMO BS receives the uplink signals, the UOSPR scheme will assure that the BS is able to recover the original symbols by acquiring the accurate phase rotation, while the massive MIMO eavesdropper can learn little about the rotated phase even with unlimited number of antennas. Therefore, the powerful massive MIMO eavesdropper will be badly confused by the randomly rotated symbols, and the original information is kept secret successfully. The complexity associated with implementing the UOSPR secure transmission scheme is low, which makes it suitable for lightweight UEs on the uplink transmission and a potential security scheme candidate for future IoT applications and 5G networks. In the following, the three stages and the uplink coherence interval structure of the UOSPR secure transmission scheme are discussed in detail, along with the corresponding secrecy analysis.

A. Uplink Training and Channel Estimation Stage

In the uplink training and BS channel estimation phase, the Formula$K$ lightweight single-antenna UEs each transmits a length Formula$\tau $ uplink pilot sequence to assist the massive MIMO BS in acquiring the corresponding CSI Formula${\mathbf {h}}_{k}$, where Formula$K\leq \tau <T$ and Formula$k = 1,2, \ldots ,K$. The acquired CSI elements are stored at the BS for further use. The pilot sequences sent by different UEs are mutually orthogonal. We assume that the massive MIMO BS are able to perfectly estimate the uplink CSI based on the received pilot sequences, and then obtain the corresponding downlink CSI by exploiting TDD channel reciprocity.

Secrecy Analysis: We consider the worst case scenario that the massive MIMO eavesdropper has full knowledge of the pilot sequences, and it can perfectly estimate the corresponding CSI Formula${\mathbf {g}}_{k}$, Formula$k = 1,2, \ldots ,K$, between itself and the Formula$K$ UEs. Notwithstanding, it is unable to acquire any useful information about the main CSI Formula${\mathbf {h}}_{k}$ between the massive BS and the legitimate UEs at this stage. This is because all the channels are independent as stated in Section II, and no information of the main CSI Formula${\mathbf {h}}_{k}$ is involved during the uplink pilot training process from the massive MIMO eavesdropper’s point of view. Hence the only information that the massive MIMO eavesdropper can obtain at this stage is the CSI between itself and the Formula$K$ lightweight UEs.

B. Random Antenna Selection and Phase Rotation Indication Stage

After channel estimation, the massive MIMO BS randomly chooses one antenna from among the large number of antennas it has. We assume that the Formula$i$th antenna is uniformly and randomly selected from the Formula$N_{b}$ antennas. It is noteworthy that the random antenna selection process is all done locally and quietly at the BS, no other nodes in the network can predict which antenna will be selected.

Then the massive MIMO BS transmits a common reference signal to the Formula$K$ single-antenna UEs through the uniformly and randomly chosen Formula$i$th antenna. The Formula$K$ UEs have perfect knowledge of the reference signal. The length of the reference signal is Formula$\lambda $, which satisfies the constraint Formula$1\leq \lambda < T-\tau $. The reference signal is used as a phase rotation indicator, which informs the Formula$K$ UEs the symbol phase they should change before they send messages to the massive MIMO BS on the uplink transmission. Note that the length of this reference signal is not long, since it is only used for assisting the lightweight UEs in acquiring the corresponding rotation phase. Thus the cost of this step is small.

By receiving the short reference signal radiated from the randomly chosen Formula$i$th antenna at the BS, the Formula$k$th UE is now able to estimate the corresponding CSI Formula${{h_{k,i}}}$ between itself and the Formula$i$th antenna at the BS, where Formula$k = 1,2, \ldots ,K$ and Formula$i = 1,2, \ldots ,N_{b}$. We assume that the Formula$K$ lightweight UEs can perfectly estimate the CSI Formula${{h_{k,i}}}$, and then obtain the phase information of the complex value Formula${{h_{k,i}}}$, i.e., Formula${\angle {h_{k,i}}}$, which will be used at the next stage.

Secrecy Analysis: We assume that the massive MIMO eavesdropper has perfect knowledge of the radiated reference signal, and it is able to perfectly estimate the CSI between itself and the Formula$i$th antenna which is randomly selected at the BS based on the intercepted reference signal. However, the massive MIMO eavesdropper is still unable to learn which antenna is randomly selected, since the random antenna selection is done locally at the BS. More importantly, the number of BS antennas Formula$N_{b}$ is large enough to prevent the eavesdropper from exhaustively testing out any useful information about the overall channel matrix between itself and the BS through the reference signal during one coherence interval, which in turn further ensures that the randomly selected antenna is kept strictly confidential. Moreover, even with unlimited antennas, the massive MIMO eavesdropper can not learn any knowledge about the CSI Formula${{h_{k,i}}}$, Formula$k = 1,2, \ldots ,K$, at this stage due to the independence of different wireless channels. Therefore, the random rotation phase Formula${\angle {h_{k,i}}}$ is unknown to the massive MIMO eavesdropper and is kept strictly confidential as a secret key for the Formula$K$ UEs only. The random antenna selection and phase rotation indication at this stage are somehow similar with the key generation and distribution operations in traditional cryptographic methods, respectively.

C. Uospr Processing and Uplink Transmission Stage

Now the Formula$K$ UEs have secretly acquired the corresponding random rotation phase information after the operations presented above. The acquired Formula${\angle {h_{k,i}}}$, Formula$k = 1,2, \ldots ,K$, at each UE is used to rotate the original information bearing symbol, as given by FormulaTeX Source\begin{equation} {s_{uospr,k}} = {s_{k}} \cdot {e^{j\angle {h_{k,i}}}}, \end{equation} where Formula${s_{uospr,k}}$ represents the random phase rotated symbol, Formula$k = 1,2, \ldots ,K$ and Formula$i = 1,2, \ldots ,N_{b}$. Then the randomly rotated symbol Formula${s_{uospr,k}}$ will be sent by UE Formula$k$ instead of the original symbol Formula${s_{k}}$ on the uplink transmission. Therefore, the original information bearing symbol Formula${s_{k}}$ can be hidden from the overall uplink transmission as long as Formula${\angle {h_{k,i}}}$ is successfully kept secret from the massive MIMO eavesdropper.

We assume that the massive MIMO BS employs maximum ratio combining (MRC) receiver to detect the corresponding uplink signal from UE Formula$k$, as given by FormulaTeX Source\begin{align} {y_{k}}=&\frac {{{{\left ({ {{{\mathbf {h}}_{k}}} }\right )}^{H}}}}{N_{b}}{{\mathbf {r}}_{uospr,b}}= \frac {{{{\left ({ {{{\mathbf {h}}_{k}}} }\right )}^{H}}{{\mathbf {h}}_{k}}}}{N_{b}}{s_{uospr,k}}\notag \\&+ \frac {{{{\left ({ {{{\mathbf {h}}_{k}}} }\right )}^{H}}}}{N_{b}}\sum \limits _{i = 1,i \ne k}^{K} {{{\mathbf {h}}_{i}}{s_{uospr,i}}} + \frac {{{{\left ({ {{{\mathbf {h}}_{k}}} }\right )}^{H}}}}{N_{b}}{{\mathbf {n}}_{b}}, \end{align} where Formula${{\mathbf {r}}_{uospr,b}}$ is obtained by replacing Formula${s_{k}}$ with Formula${s_{uospr,k}}$ and Formula${s_{i}}$ with Formula${s_{uospr,i}}$ in (3), and it denotes the de facto received signal at the massive MIMO BS; Formula${y_{k}}$ represents the output of the MRC receiver for UE Formula$k$’s uplink information. Then the massive MIMO BS can acquire Formula${\hat s_{uospr,k}}$, which denotes the symbol estimation result of the randomly rotated symbol Formula${s_{uospr,k}}$, based on the receiver output Formula${y_{k}}$.

Finally, the massive MIMO BS can recover the original information bearing symbol by taking the following inverse operation FormulaTeX Source\begin{equation} {\hat s_{k}} = {\hat s_{uospr,k}} \cdot {e^{j\left ({ { - \angle {h_{k,i}}} }\right )}}, \end{equation} where Formula$\angle {h_{k,i}}$ is obtained and stored at the BS channel estimation stage, and Formula${\hat s_{k}}$ is the final recovered result of the original information bearing symbol Formula$s_{k}$ that UE Formula$k$ intends to send on the uplink transmission.

Note that we use ‘UOSPR processing’ to denote and highlight the phase rotation operation at the lightweight UEs at this stage, since it is the critical part of the overall UOSPR secure transmission scheme.

Secrecy Analysis: At this stage, we consider the worst case scenario that the massive MIMO eavesdropper with infinite number of antennas is able to cancel all the interference and noise and perfectly intercept the de facto symbols transmitted by the Formula$K$ UEs. This means that the massive MIMO eavesdropper can perfectly obtain the de facto symbols Formula${s_{uospr,k}}$, Formula$k = 1,2, \ldots ,K$ transmitted by the Formula$K$ UEs on the uplink. Nonetheless, the eavesdropper is still unable to recover the original information bearing symbol Formula${s_{k}}$, since it can not obtain any useful information about the corresponding random rotation phase Formula${\angle {h_{k,i}}}$ which is used to ‘encrypt’ the original symbols, according to the secrecy analysis above. Therefore, the original messages from the lightweight UEs are successfully secured on the uplink transmission.

D. UOSPR Coherence Interval Structure and Complexity

These three stages of the UOSPR secure transmission scheme are specifically illustrated in Fig. 2, which actually presents the corresponding structure of the coherence interval. Moreover, as shown in Fig. 2, the stage B with red fill color is an optional choice to further enhance the secrecy performance if the coherence interval is long enough. In this case, the massive MIMO BS and the legitimate UEs can collaborate together to increase the frequency of random phase rotation during one coherence interval, which can more effectively confuse the massive MIMO eavesdropper. It is similar with increasing the secret key generation rate in traditional complexity based cryptographic methods. However, it is worth emphasizing that if the number of BS antennas is not large enough, the massive MIMO eavesdropper may benefit from the increase of the red ‘B’ operation, since it will also have more chance to learn the random selection process and exhaustively test out the original information. Therefore, we should make sure that there are enough number of antennas at the BS to perform the random antenna selection process, in which the level of ‘enough number’ is mainly determined by the frame parameters and wireless channel characteristics. This presents the key role of the massive MIMO BS in the UOSPR secure transmission scheme.

Figure 2
Fig. 2. The structure of the UOSPR coherence interval.

Furthermore, we can see that the process of the UOSPR scheme is not complex from the perspective of the lightweight single-antenna UEs. The lightweight UEs only need to perform two extra simple steps, i.e., short reference signal estimation and random phase rotation, comparing with general uplink protocols. Moreover, these two extra operations do not need the support of powerful signal processing and data computing capabilities. The corresponding complexity and cost are low while the uplink secrecy performance is well improved. This advantage makes the UOSPR scheme potentially suitable for network full of lightweight UEs on the uplink, e.g., IoT scenarios. Note that it is the massive MIMO BS who offers enough number of antennas to generate a good randomicity for encrypting the original information bearing symbols. This means that the large antenna array employed at the BS greatly alleviates the burden of the lightweight UEs on achieving favorable secrecy performance. It presents another pivotal role of the massive MIMO BS in the UOSPR scheme.

SECTION IV

SECRECY PERFORMANCE SIMULATION

The secrecy performance of the employed UOSPR secure transmission scheme is evaluated in this section. We assume that the Formula$K$ UEs modulate the radiated uplink symbols with practical quadrature phase-shift-keying (QPSK) constellation, which is given by FormulaTeX Source\begin{align} \left [{ {\begin{array}{*{20}{c}} {\dfrac {\sqrt {2}}{2}\left ({ { - 1 + j} }\right )}& {\dfrac {\sqrt {2}}{2}\left ({ {1 + j} }\right )}\\[0.6pc] {\dfrac {\sqrt {2}}{2}\left ({ { - 1 - j} }\right )}& {\dfrac {\sqrt {2}}{2}\left ({ {1 - j} }\right )} \end{array}} }\right ]\!, \end{align} where these four complex points belong to the corresponding quadrants and the constellation values are normalized. The number of BS antennas is Formula$N_{b} = 32, 64, 128, 256$, and the number of UEs is Formula$K =16$. We assume that Formula$N_{b} = 32$ is large enough to properly implement the random antenna selection process of the UOSPR scheme for the convenience of performance evaluation here. The transmit power Formula$P_{k}$ of UE Formula$k$ is normalized, then the transmit signal-to-noise ratio (SNR) of UE Formula$k$ is defined as Formula${\mathrm {SNR}}_{b} = \frac {1}{\sigma _{b}^{2}}$ at the BS, and similarly Formula${\mathrm {SNR}}_{e} = \frac {1}{\sigma _{e}^{2}}$ at the eavesdropper. The MRC receiver is employed at the massive MIMO BS and the eavesdropper to detect the receiving signals.

Note that if the random rotation phase Formula$\angle {h_{k,i}} \leq \frac {\pi }{4}$ or Formula${\angle {h_{k,i}} \ge \frac {7\pi }{4}}$, Formula$k = 1,2, \ldots ,K$ and Formula$i = 1,2, \ldots ,N_{b}$, the original information bearing symbol Formula$s_{k}$ will not be rotated due to the employed QPSK modulation. Otherwise, Formula$s_{k}$ will be rotated to the corresponding quadrant according to (5), and replaced by the constellation point in that quadrant. All the channel estimations are perfect as assumed before in this work. The analytical result in term of the symbol error rate (SER) at the massive MIMO eavesdropper is similar with the downlink case in [44]. In the worst case scenario that the eavesdropper has infinite number of antennas and can intercept the de facto transmitted symbol Formula${s_{uospr,k}}$, Formula$k = 1,2, \ldots ,K$ perfectly by cancelling all the interference and noise on the uplink transmission, the SER of the eavesdropper is 0.75. Please refer to [44] for SER analysis in detail.

The symbols transmitted by the 1st UE on the uplink is chosen for simulation purpose. The SER simulation results as a function of the number of antennas at the massive MIMO BS and the massive MIMO eavesdropper when the UOSPR scheme is employed are presented in Fig. 3 and Fig. 4. 1000000 symbols are simulated for the case of 32, 64, 128 and 256 BS antennas. 5000000 symbols are simulated for the case of 1024, 2048, 4096 and 8192 eavesdropper antennas, which is far more than that at the BS. We assume that Formula${\mathrm {SNR}}_{b} = 5dB$ at the BS and Formula${\mathrm {SNR}}_{e} = 50dB$ at the eavesdropper, since the latter is much more powerful. The blue bars in Fig. 3 represent the SER results at the BS, and the red bar denotes the massive MIMO eavesdropper’s SER with 1024 antennas. It shows that the SER of the massive MIMO BS quickly decreases as the number of BS antennas grows. The reason is that the ability on cancelling the interference and noise is proportional to the number of BS antennas. However, we can see that the SER of the massive MIMO eavesdropper is very high and approximately 0.75, and the increase of eavesdropping antennas does not improve the SER performance at all as shown in Fig. 4. The data tip ‘X’ and ‘Y’ in Fig. 3 and Fig. 4 denote the result sequence and SER values, respectively. Note that in the simulation the number of eavesdropper antennas is set to several thousands in order to demonstrate that the eavesdropper can not recover the original symbols even with much more powerful massive MIMO. These simulations prove the corresponding analytical results. It demonstrates that even with many more antennas and much higher SNR than the BS, the eavesdropper is unable to break the UOSPR secure transmission scheme. The massive MIMO BS’s SER is much lower than that of the massive MIMO eavesdropper, which means the uplink information is successfully protected to a large extent.

Figure 3
Fig. 3. SER of the massive MIMO BS and eavesdropper on the uplink transmission.
Figure 4
Fig. 4. SER of the massive MIMO eavesdropper with different number of antennas.

Moreover, we present the signal constellation scatter results of the symbols before and after phase recover operation at the massive MIMO BS in Fig. 5 and Fig. 7 with original information bearing symbols Formula${\frac {\sqrt {2}}{2}\left ({ { - 1 + j} }\right )}$ in quadrant two, and Fig. 6 and Fig. 8 with original information bearing symbols Formula${\frac {\sqrt {2}}{2}\left ({ { 1 - j} }\right )}$ in quadrant four, respectively. The number of BS antennas for simulation is 64, and Formula${\mathrm {SNR}}_{b} = 5dB$. We can see that the initial output symbols of the MRC receiver at the BS are chaotically distributed in all the quadrants under the effects of random phase rotation, interference and noise as shown in Fig. 5 and Fig. 6. The chaotic symbols are then successfully rotated back to the second and fourth quadrant in Fig. 7 and Fig. 8 by taking the inverse operation according to (7), since the BS can acquire the phase rotation information.

Figure 5
Fig. 5. The constellation scatter plot of UE 1 at the BS before phase recover with original information bearing symbols in quadrant two.
Figure 6
Fig. 6. The constellation scatter plot of UE 1 at the BS before phase recover with original information bearing symbols in quadrant four.
Figure 7
Fig. 7. The constellation scatter plot of UE 1 at the BS after phase recover with original information bearing symbols in quadrant two.
Figure 8
Fig. 8. The constellation scatter plot of UE 1 at the BS after phase recover with original information bearing symbols in quadrant four.

Fig. 9 with original information bearing symbols in quadrant two and Fig. 10 with original information bearing symbols in quadrant four show the corresponding signal constellation scatter plots at the massive MIMO eavesdropper. The number of eavesdropper antennas for simulation is 1024, and Formula${\mathrm {SNR}}_{e} = 50dB$. It is obvious that the constellation results are much better than that at the BS. Nonetheless, it is unable to rotate the symbols back to the second and fourth quadrant which contain the original information bearing symbols, since the eavesdropper can not learn any useful information about the random phase rotation even with much more number of antennas. On the other hand, the massive MIMO BS is able to normally recover most of the original symbols although its constellation scatter plots are not so good as presented in Fig. 7 and Fig. 8. The scatter plots also indicate that the eavesdropper’s SER is approximately 0.75, since the points are almost uniformly distributed in the four quadrants, no matter which quadrant the original information bearing symbols are in. The eavesdropper can not distinguish the symbols due to the random phase rotation process. Furthermore, the massive MIMO eavesdropper may be deceived by the good constellation scatter plots in that it thinks it has received the correct confidential messages. The evaluation results above corroborate the secrecy analysis of the employed UOSPR scheme.

Figure 9
Fig. 9. The constellation scatter plot of UE 1 at the eavesdropper with original information bearing symbols in quadrant two.
Figure 10
Fig. 10. The constellation scatter plot of UE 1 at the eavesdropper with original information bearing symbols in quadrant four.

It is noteworthy that the secrecy performance simulation for the case of QPSK modulation above can be similarly applied to other practical finite constellation case.

SECTION V

CONCLUSION

More and more lightweight UEs appear in our life due to the coming of IoT and 5G era, and the security of uplink communication becomes more important. In this work, we present an alternative secure transmission scheme to protect the private information for lightweight UEs on the uplink. The UOSPR secure transmission scheme is employed to defend against the massive MIMO eavesdropper equipped with unlimited number of antennas on the uplink transmission for the lightweight single-antenna UEs in a single-cell scenario. We present the three stages and the corresponding coherence interval structure of the UOSPR scheme, and the secrecy analysis is provided in detail. The SER performance at the massive MIMO BS and the massive MIMO eavesdropper are investigated separately under certain assumptions. The simulation results corroborate the corresponding secrecy analysis of the UOSPR scheme. We show that the massive MIMO eavesdropper is unable to recover the major original information bearing symbols due to its high SER, while the massive BS with much fewer antennas can normally receive the information on the uplink. We present that the complexity of the UOSPR scheme is low from the perspective of the lightweight UEs, since only two extra simple operations are needed to implement the scheme comparing with traditional uplink protocols. This potentially makes it a candidate security approach for IoT like network which is full of lightweight UEs on the uplink transmission.

Note that optimistic assumptions are made in this paper, and more detailed theoretical analysis and the effects of imperfect CSI parameters are left for future work.

Footnotes

This work was supported in part by the National Natural Science Foundation of China under Grant 91338105 and Grant 61502518, the China Scholarship Council under Grant 201306110074, the Four-Year Doctoral Fellowship through The University of British Columbia, the Natural Sciences and Engineering Research Council of Canada, the Institute for Computing, Information, and Cognitive Systems/TELUS People and Planet Friendly Home Initiative at The University of British Columbia, TELUS, and other industry partners. The work of L. Shu was supported in part by the 2013 Special Fund of Guangdong Higher School Talent Recruitment, Educational Commission of Guangdong Province, China, under Project 2013KJCX0131, the Guangdong High-Tech Development Fund under Grant 2013B010401035, the 2013 top Level Talents Project in Sailing Plan of Guangdong Province, the National Natural Science Foundation of China under Grant 61401107, and the 2014 Guangdong Province Outstanding Young Professor Project. The work of J. J. P. C. Rodrigues has been partially supported by Instituto de Telecomunicações, Next Generation Networks and Applications Group (NetGNA), by National Funding from the FCT - Fundação para a Ciência e Tecnologia through the Pest-OE/EEI/LA0008/2013 Project, by Government of Russian Federation, Grant 074-U01, and by Finep, with resources from Funttel, Grant No. 01.14.0231.00, under the Radiocommunication Reference Center (CRR) project of the National Institute of Telecommunications (Inatel), Brazil.

Corresponding author: B. Chen

References

No Data Available

Authors

Bin Chen

Bin Chen

Bin Chen received the B.Sc. degree (summa cum laude) in communication engineering from the National University of Defense Technology (NUDT), Changsha, China, in 2010.

He is currently pursuing Ph.D. degree with the Department of Communication Engineering, College of Electronic Science and Engineering, NUDT. His research interests include massive MIMO, physical layer security, MIMO precoding, and detection.

Chunsheng Zhu

Chunsheng Zhu

Chunsheng Zhu received the B.E. degree in network engineering from the Dalian University of Technology, China, in 2010, the M.Sc. degree in computer science from St. Francis Xavier University, Canada, in 2012, and the Ph.D. degree in electrical and computer engineering from The University of British Columbia, Canada, in 2016.

He is currently a Post-Doctoral Research Fellow with the Department of Electrical and Computer Engineering, The University of British Columbia, Canada. He has authored over 60 papers published or accepted by refereed international journals (e.g., the IEEE TRANSACTIONS on INDUSTRIAL ELECTRONICS, the IEEE TRANSACTIONS on COMPUTERS, the IEEE TRANSACTIONS on INFORMATION FORENSICS and SECURITY, the IEEE TRANSACTIONS on EMERGING TOPICS in COMPUTING, the IEEE SYSTEMS JOURNAL, and the IEEE ACCESS) and conferences (e.g., the IEEE Globecom and the IEEE ICC). His current research interests mainly include wireless sensor networks, cloud computing, Internet of Things, social networks, and security.

Lei Shu

Lei Shu

Lei Shu (M’07–SM’16) received the B.Sc. degree in computer science from South Central University for Nationalities, China, in 2002, and the M.Sc. degree in computer engineering from Kyung Hee University, South Korea, in 2005, and the Ph.D. degree from the Digital Enterprise Research Institute, National University of Ireland, Galway, Ireland, in 2010.

He is currently a Full Professor with the College of Electronic Information and Computer, Guangdong University of Petrochemical Technology, China, also the Vice-Director of the Guangdong Petrochemical Equipment Fault Diagnosis Key Laboratory. He was a Specially Assigned Research Fellow with the Department of Multimedia Engineering, Graduate School of Information Science and Technology, Osaka University, Japan. He has authored over 200 papers in related conferences, journals, and books. He had been awarded the Globecom 2010 and ICC 2013 Best Paper Award. He is serving as an Editor-in-Chief for the EAI Endorsed Transactions on Industrial Networks and Intelligent Systems, and he has served as an Editor and a Guest Editor of more than 20 international journals. He has served as more than 50 various Co-Chair for international conferences/workshops, e.g., IWCMC, ICC, and ISCC; a TPC member of more than 150 conferences, e.g., ICC, Globecom, ICCCN, WCNC, and ISCC. He has served as a Reviewer of more than 50 journals. His research interests include wireless sensor network, sensor network middleware, multimedia communication, and security. He is a member of ACM.

Man Su

Man Su

Man Su received the B.Sc. degree in communication engineering from the National University of Defense Technology, China, in 2012.

She is currently pursuing the Ph.D. degree in mobile communications with the Institute for Communication Systems, 5G Innovation Centre, University of Surrey, U.K. Her research interests focus on MAC protocol, user scheduling, and radio resource management, especially for the densely deployed small cells.

Jibo Wei

Jibo Wei

Jibo Wei (M’03) received the B.S. and M.S. degrees from the National University of Defense Technology, Changsha, China, in 1989 and 1992, respectively, and the Ph.D. degree from Southeast University, Nanjing, China, in 1998, all in electronic engineering.

He is currently the Director and a Professor with the Department of Communication Engineering, NUDT. His research interests include wireless network protocols and signal processing in communications, particularly in the areas of multiple-input multiple-output, multicarrier transmission, cooperative communications, and cognitive networks.

Dr. Wei serves as an Editor for the Journal on Communications. He is a Senior Member of the China Institute of Communications and Electronics and a member of the IEEE Communication and Vehicular Technology Societies.

Victor C. M. Leung

Victor C. M. Leung

Victor C. M. Leung (S’75–M’89–SM’97 –F’03) received the B.A.Sc. (Hons.) and Ph.D. degrees in electrical engineering from the University of British Columbia (UBC), in 1977 and 1982, respectively. He is currently a Professor of Electrical and Computer Engineering and a Holder of the TELUS Mobility Research Chair with UBC. His research is in the areas of wireless networks and mobile systems. He has co-authored over 900 technical papers in archival journals and refereed conference proceedings, several of which had won best-paper awards. He is a Fellow of the Royal Society of Canada, the Canadian Academy of Engineering, and the Engineering Institute of Canada. He is serving on the Editorial Boards of the IEEE JSAC-SGCN, the IEEE WIRELESS COMMUNICATIONS LETTERS, the IEEE ACCESS, and several other journals. He has provided leadership to the technical program committees and organizing committees of numerous international conferences. He was a recipient of the 1977 APEBC Gold Medal, the NSERC Postgraduate Scholarships from 1977 to 1981, a 2012 UBC Killam Research Prize, and an IEEE Vancouver Section Centennial Award.

Joel J. P. C. Rodrigues

Joel J. P. C. Rodrigues

Joel J. P. C. Rodrigues (S’01–M’06–SM’06) is a professor and senior researcher at the National institute of Telecommunications (Inatel), Brazil and has been professor at the University of Beira Interior (UBI), Portugal, senior researcher at the Instituto de Telecomunicações, Portugal, and visiting professor at the University of Fortaleza (UNIFOR), Brazil. He received the Academic Title of Aggregated Professor from the UBI, the Habilitation in computer science and engineering from the University of Haute Alsace, France, a PhD degree in informatics engineering, an MSc degree from the UBI, and a five-year BSc degree (licentiate) in informatics engineering from the University of Coimbra, Portugal. His main research interests include e-health, sensor networks, IoT, vehicular communications, and mobile and ubiquitous computing. He is the leader of NetGNA Research Group (http://netgna.it.ubi.pt), the President of the scientific council at ParkUrbis – Covilhã Science and Technology Park, the Past-Chair of the IEEE ComSoc Technical Committee on eHealth, the Past-chair of the IEEE ComSoc Technical Committee on Communications Software, Steering Committee member of the IEEE Life Sciences Technical Community and Publications co-Chair, and Member Representative of the IEEE Communications Society on the IEEE Biometrics Council. He is the editor-in-chief of the International Journal on E-Health and Medical Communications, the editor-in-chief of the Recent Advances on Communications and Networking Technology, the editor-in-chief of the Journal of Multimedia Information Systems, and editorial board member of several high-reputed journals. He has been general chair and TPC Chair of many international conferences, including IEEE ICC, GLOBECOM, and HEALTHCOM. He is a member of many international TPCs and participated in several international conferences organization. He has authored or coauthored over 500 papers in refereed international journals and conferences, 3 books, and 2 patents. He had been awarded several Outstanding Leadership and Outstanding Service Awards by IEEE Communications Society and several best papers awards. Prof. Rodrigues is a licensed professional engineer (as senior member), member of the Internet Society, an IARIA fellow, and a senior member of ACM and IEEE.

Cited By

No Data Available

Keywords

Corrections

None

Multimedia

No Data Available
This paper appears in:
No Data Available
Issue Date:
No Data Available
On page(s):
No Data Available
ISSN:
None
INSPEC Accession Number:
None
Digital Object Identifier:
None
Date of Current Version:
No Data Available
Date of Original Publication:
No Data Available

Text Size


Comment Policy
comments powered by Disqus