Discrete Particle Swarm Optimization Based
Antenna Selection for MIMO LoRa IoT Systems
1
st
Jiajun Xu
College of Electronics and Information Engineering
Shenzhen University
Shenzhen, China
xujiajun2017@email.szu.edu.cn
2
nd
Peichang Zhang
College of Electronics and Information Engineering
Shenzhen University
Shenzhen, China
pzhang@szu.edu.cn
3
rd
Shida Zhong
College of Electronics and Information Engineering
Shenzhen University
Shenzhen, China
shida.zhong@szu.edu.cn
4
th
Lei Huang
College of Electronics and Information Engineering
Shenzhen University
Shenzhen, China
lhuang@szu.edu.cn
Abstract—In recent years, as one of the most efficient low
power wide area network (LPWAN) techniques, long range
(LoRa) has been widely employed in the internet of things
(IoT) applications due to its wide coverage and high interfer-
ence resistance. In this paper, a multiple-input multiple-output
(MIMO) LoRa system is proposed for further enhancing the
performance of LoRa based IoT systems, where correlated
channel environment is considered between the LoRa gateway
and the node. More explicitly, the multi-antenna gateway employs
LoRa modulation and the signals are precoded by space-time
block coding for long-range and reliable data transmission in
LPWAN. Moreover, since power consumption is a key point to
consider in LPWAN, we further propose a discrete particle swarm
optimization based antenna selection (DPSOBAS) algorithm for
the MIMO LoRa system to reduce the hardware costs and
power consumption, while retaining the MIMO advantages.
The corresponding simulation results show that the proposed
DPSOBAS is capable of substantially improving the performance
of the proposed MIMO LoRa IoT system in correlated channel
environment, and approaching the performance of ergodic search
based optimal AS scheme with significantly lower computational
complexity.
Index Terms—Long Range (LoRa), multiple-input-multiple-
output (MIMO), antenna selection (AS), particle swarm opti-
mization (PSO)
I. INTRODUCTION
In recent years, low power wide area network (LPWAN) has
drawn substantial attention due to its advantages of low power
consumption and wide coverage [1]. Long Range (LoRa),
which is one of the popular LPWAN technologies, has been
exhibiting tremendous commercial growth to establish itself
among the front runners within LPWAN area [2].
As a physical layer technology, LoRa modulates transmitted
signals in sub-GHz ISM band by chirp spreading spectrum
(CSS) [3] [4] for the sake of offering wide connectivity,
This work is supported in part by the Natural Science Foundation of China
under Grants 61601304, U1713217, U1501253, 61801297, and 61801302, in
part by the Foundation of Shenzhen under Grant JCYJ20170302142545828,
and in part by the Foundation of Shenzhen University under Grant 2016057.
consuming ultra low power and reducing hardware costs.
[5]–[7] have studied LoRa modulation in terms of its bit
error ratio (BER) performance analysis [5], quality of service
(QoS) [6] and capacity planning [7]. To further improve
the performance of LoRa internet of things (IoT) system,
[1] has concluded that LPWAN techniques have considerable
scalability, which may employ multi-antenna communication
to parallelize transmission to and from the connected devices.
Therefore, it has been recognized that introducing multiple-
input multiple-output (MIMO) to LPWAN systems may help
to achieve high channel capacity, high data rate and strong
robustness [8].
Conventionally, a MIMO system employs the same number
of antennas and radio frequency (RF) chain pairs, implying
that enlarging the scale of MIMO system increases both
the number of antennas and RF chains, which may lead
to high power consumption and exorbitant hardware costs.
However, one of the most significant advantages of LoRa
LPWAN systems is to realise ultra low power consumption at
relatively low costs. In this case, directly introduce the MIMO
concept to LoRa LPWAN systems may not only increase
the power consumption, but also raise the systems’ hardware
costs, which may severely limit the applications of LoRa
in IoT systems. As a result, for the sake of lowering the
power consumption and reducing hardware costs in MIMO
LoRa systems, antenna selection (AS) technique is tailored
to reduce the number of RF chains required and maintain
MIMO communication performance. [9]–[12] have proposed
two categories of AS schemes, namely the norm-based AS
(NBAS) [9] [10] and capacity-based AS (CBAS) [11] [12].
The idea of NBAS is to select the subset of full MIMO
channel associated with the highest channel gain, while CBAS
is to find the channel subset resulting in the highest channel
capacity. In independent MIMO channel environment, NBAS
usually achieves comparable performance with CBAS, but
with lower computational complexity. However, in correlated