A Scheduling Scheme for Improving the Performance and Security of MU-MIMO Systems

Abstract

For the receiver of multiple-input multiple-output (MIMO) systems, linear detectors are an interesting option due to their good performance and low complexity. Nevertheless, MIMO systems lose diversity in exchange for eliminating interference when linear detectors are used. Aiming to maintain the system diversity while mitigating interference between users, this work proposes a scheduling scheme for the uplink of multiuser MIMO (MU-MIMO) systems that employ A antennas and the zero-forcing (ZF) detector at the receiver in the base station (BS). The channel model includes Rician fading and additive white Gaussian noise (AWGN) in an imperfect channel estimation scenario. The proposed scheme selects U users from a group of Ut users to transmit simultaneously, so that the signal-to-noise ratio (SNR) is maximized. For this, an exact expression to evaluate the SNR of the users is obtained. With this result, the scheduling strategy is proposed. Results show that as Ut increases, the outage probability (OP), and the bit error rate (BER) decrease as the system diversity increases, even when the system is completely loaded, i.e., when U=A. Moreover, it is shown that the scheduling scheme counteracts the imperfect channel estimation effects as Ut increases. Finally, the proposed scheme is tested in presence of an external eavesdropper trying to decode the information sent by the users. The results show that the presented proposal allows for a reduction of the secrecy-outage-probability (SOP) as Ut increases.

Keywords: MU-MIMO; Rician fading; imperfect channel estimation; scheduling; zero-forcing.

Figure 1

MU-MIMO system with $A=3$ antennas in the BS, in which $U=3$ users are served simultaneously out of a total of $U_t=4$ users.

Figure 2

Channel assignment scenarios with the proposed scheduling scheme considering $U_t=4$ and $U=3$ for $A=3$ receiving antennas.

Figure 3

OP as a function of $E_b/N_0$, parameterized by $U_t$ and by e, considering BPSK modulation, $A=4$, $U=4$, $T=0$ dB, and a channel with Rician factor $K=1$.

Figure 4

OP as a function of $U_t$, parameterized by K and the threshold T, considering BPSK modulation, $A=2$, $U=2$, and $e=0.1$.

Figure 5

BER as a function of $E_b/N_0$, parameterized by $U_t$, and by the Rician factor K, considering BPSK modulation, $A=4$, $U=4$, and $e=0.1$.

Figure 6

BER as a function of e, parameterized by $U_t$, considering BPSK modulation, $A=6$, $U=6$, $E_b/N_0=10$ dB and a channel with Rician factor $K=2$.

Figure 7

BER as a function of $E_b/N_0$, parameterized by e, and the scheduling algorithm considering BPSK modulation, $A=4$, $U=4$, $U_t=5$, and a channel with Rician factor $K=2$.

Figure 8

BER as a function of $E_b/N_0$, parameterized by e, and the scheduling algorithm considering 16-QAM, $A=4$, $U=4$, $U_t=5$, and a channel with Rician factor $K=2$.

Figure 9

SOP as a function of c, parameterized by $U_t$, considering $M=4$, $A=2$, $A_E=2$, $U=2$, $E_b/N_0=15$ dB, $e=0.1$, and $R=1$ bps in a Rician fading channel with $K=2$.

Figure 10

SOP as a function of c, parameterized by $U_t$, considering $M=4$, $A=3$, $A_E=2$, $U=2$, $E_b/N_0=15$ dB, $e=0.1$, and $R=1$ bps in a Rician fading channel with $K=2$.

Figure 11

SOP as a function of $E_b/N_0$, parameterized by $A_E$ and $U_t$, considering $M=4$, $A=3$, $U=3$, $e=0.1$, $c=0.05$, and $R=1$ bps in a Rician fading channel with $K=2$.