On the Mutual Information and Constellation Design
Criterion of Spatial Modulation MIMO Systems
†
Shuaishuai Guo,
†
Haixia Zhang
†
School of Information Science & Engineering,
Shandong University,
Jinan, Shandong, 250100, China
‡
Jia Zhang,
†
Dongfeng Yuan
‡
School of Information Science & Engineering,
Shandong Normal University
Jinan, Shandong, China, 250014
Abstract—The mutual information of spatial modulation (SM)
based multiple input and multiple output (MIMO) systems with
arbitrary number of transceiver antennas is investigated. An
expression of the system mutual information is derived. The
expression is not in a closed form and may lead to difficulties
when trying to optimize the mutual information. To solve this,
based on Jensen Inequality we propose an approximation for
mutual information, which is in a closed form. We show that the
approximation is accurate in both low and high SNR regimes.
In addition, based on the approximated version of the system
mutual information, two constellation design critera are proposed
to maximize the system mutual information with finite alphabet.
Numerical results show that both of the two criteria perform
well.
I. INTRODUCTION
Recently, there is a growing interest in applying a low-
complexity technique named spatial modulation (SM) into
multiple antennas transceiver designs. Rather than employ-
ing all the transmit antennas for transmission, SM achieves
spatial multiplexing by exploiting one single antenna at any
time, therefore causes no inter channel interference (ICI) and
synchronization problems. In SM, information bits are divided
into two separate parts: one part is used for antenna selection;
the other part is mapped into traditional signal constellation
point. As a result, the indices of transmit antennas are exploited
as an additional dimension for multiplexing.
As one of the promising modulation techniques, the con-
cept of SM has been developed into various forms, such as
space shift keying (SSK) [1], generalized spatial modulation
(GSM) [2], antenna subset modulation (ASM) [3], information
guided channel hopping (IGCH) [4], etc. A summary of these
schemes can be seen in the tutorial paper [5]. In this con-
tributed paper [5], Renzo et al offer a comprehensive overview
of the state of art in SM for generalized multiple-input
multiple-output (MIMO) technologies. In recent literature, SM
is extensively analyzed in the maximum transmission rate,
transmitter and receiver design, error performance, etc. For
instance, in [6], a framework for error performance analysis
of SM MIMO systems is proposed and the upper bound of the
average bit error probability (ABEP) of SM based multiple
input single output (SM MISO) systems is derived. In [7],
the optimal detector for SM MIMO systems is derived and
analyzed. All these works pay no attention on the mutual
information and capacity of SM MIMO systems. To the best
of our knowledge, only a few enlightening work on mutual
information and capacity analysis has been done in [4], [8].
Active
Antenna
Indice
Detector
Demapper
Fig. 1: The SM MIMO system block diagram.
[4] firstly proposed and analyzed IGGH in terms of outage
capacity and ergodic capacity by assuming Gaussian inputs.
Although Gaussian inputs are optimum, they can never be
realized in practical communication systems, leading to sub-
stantial performance degradation [9], [10], [11]. [8] considers
practical discrete inputs and derives an expression of the
mutual information. It is worth mentioning that both the work
in [4] and [8] is done based on the assumption of single receive
antenna, i.e. SM MISO systems. Therefore, mutual information
of SM based MIMO systems is still an open problem. In this
work, we develop an SM MIMO systems oriented information-
theoretic framework to compute the mutual information with
finite alphabet inputs. To maximize the obtained mutual in-
formation, we afterwards develop an adaptive constellation
design criterion (A-CDC) and an maximizing average pairwise
distance constellation design criterion (MAPD-CDC) for SM
MIMO systems.
In this paper, we use the following notations. Uppercase
letters denote matrices, lowercase boldface letters represent
vectors and lowercase non-boldface letters stand for scalars.
The field of complex numbers is denoted by C and the field
of nature numbers by N. E
x
{·} refers to the expected value
of random variable x. || · || denotes the Euclidean norm. The
superscripts (·)
T
and (·)
†
stand for transpose and conjugate
transpose operations, respectively.
II. SYSTEM MODEL AND PERFORMANCE ANALYSIS
We consider an SM based MIMO system equipped with
N
t
transmit antennas and N
r
receive antennas, as illustrated
in Fig. 1. The baseband input-output relationship can be given
by
y = Hx + v. (1)
where y ∈ C
N
r
×1
is the received signal, H ∈ C
N
r
×N
t
is the
complex channel matrix, x ∈ C
N
t
×1
is the base band input