Mimo 3

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TCSII.2018.2822845, IEEE
Transactions on Circuits and Systems II: Express Briefs
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMSŁII: EXPRESS BRIEFS 1

Transmit Designs for Spectral Coexistence of


MIMO Radar and MIMO Communication System
Junhui Qian, Student Member, IEEE, Zishu He, Member, IEEE, Nuo Huang, and Bo Li, Student Member, IEEE,

Abstract—This paper considers the transmit designs in spectral systems for spectrum sharing originally is studied by Li and
coexistence of multiple-input multiple-output (MIMO) commu- Petropulu. The authors conduct a joint design of the MIMO-
nication and MIMO radar systems. A joint optimization of MC radar transmit precoder, radar sub-sampling scheme, and
the communication transmit covariance matrix and the radar
transmit waveform is designed to maximize the communication the communication transmit covariance matrix to maximize
rate, with constraints concerning the similarity with a standard the radar SINR subject to the constraints of communication
radar waveform and the transmit energies of both the radar and rate and communication transmit power.
the communication system. The original design is constructed by In this paper, we consider the joint design of the transmit
solving two convex optimization problems. Then an algorithm covariance matrix for MIMO wireless communication system
of alternating iterative procedure is developed, whose solutions
are provided in closed form. Simulation results are presented to and the transmit waveform for MIMO radar system. We adopt
validate the effectiveness of the proposed algorithm. a geometric-based channel model for the radar-communication
channel as well, extending to the case of co-existing systems
Index Terms—Multiple-input multiple-output (MIMO) com-
munication, MIMO radar, spectrum sharing, alternating opti- the one outlined in [11] with reference to MIMO communi-
mization. cations. We formulate the design problem as a constrained
maximization of the communication rate subject to a number
of constraint guaranteeing both the radar performance, through
I. Introduction a similarity constraint with some standard waveform and

I N current technological development, the explosive devel-


opment of services exploiting the terrestrial radio channel
has brought up the problem of designing composite architec-
energy constraints of the coexistence system transmitters. To
solve the non-convex problem, we develop a novel cyclic
algorithm with low complexity. Moreover, the solutions are
tures, consisting of systems with different functions which provided in closed form. A thorough performance assessment
jointly use the major physical resource. Moreover, keeping is finally undertaken, so as to illustrate the merits of the
under surveillance through radar system poses a problem of proposed algorithm.
radar/communication co-existence, i.e., of efficient exploita-
tion of the assigned spectrum and cross interference minimiza- II. System model
tion [1]–[8]. Spectrum sharing is an emerging area of research Consider a joint communication-radar coexistence system
whose aim is to enable radar and communication system to where a MIMO communication system and a colocated MIMO
share the spectrum efficiently by minimizing mutual interfer- radar system share the same carrier frequency. Fig. 1 outlines a
ence [1]. For the radar system, the key interference mitigation model of the considered scenario. The MIMO communication
techniques for directional communication are beamforming system is equipped with Mt transmit antennas and Mr receive
and nulling [2]. In [3], [4], without considering the commu- antennas. The MIMO radar system is equipped with equal
nication system design, the authors investigate the synthesis number of transmit antennas and receive antennas, denoted by
of optimized radar waveforms ensuring spectral compatibility N. In the considered narrow-band, flat-fading environment, it
with the overlayed licensed electromagnetic radiators. Howev- is assumed that the communication system and radar system
er, in the coexisting scenario, the radar signal usually contains are synchronized in terms of sampling times and all channels
interference from communication system, and it is necessary to remain the same over L symbol intervals. The signal at the
conduct research on the communication system design. To the communication receiver for sampling time index l ∈ N+L can
best of our knowledge, the MIMO system boosts the flexibility be written as
in the design [9], [10]. In [5]–[8], the coexisting of MIMO-
yC (l) = Hx (l) + z (l) + n (l) , (1)
matrix completion (MC) radar and MIMO communication
where H ∈ C Mr ×Mt denotes the communication channel matrix,
Junhui Qian and Zishu He are with Electrical Engineering Department,
University of Electronic Science and Technology of China (UESTC), Xi Yuan
x(l) ∈ C Mt ×1 denotes
 the lth communication transmit symbol
Road no. 2006, Chengdu, Sichuan Province 611731, Peoples Republic of vector, n (l) ∼ CN 0, σC2 I Mr is the additive white Gaussian
China (e-mail: junhuiq123@163.com; zshe@uestc.edu.cn). noise vector, z (l) represents the interference from the radar
Nuo Huang is with National Mobile Communications Research Laboratory,
Southeast University, Nanjing 210096, Peoples Republic of China (e-mail:
system given by
huangnuo@seu.edu.cn).
B. Li is with Department of Electrical and Computer Engineering, Rutgers, z (l) = G1 s (l) e jαil , (2)
The State University of New Jersey, Piscataway NJ 08854 USA (e-mail:
paul.bo.li@rutgers.edu). where G1 ∈ C Mr ×N denotes the interference channel from the
radar transmitter to the communication receiver, s (l) ∈ CN×1

1549-7747 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TCSII.2018.2822845, IEEE
Transactions on Circuits and Systems II: Express Briefs
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMSŁII: EXPRESS BRIEFS 2

III. Proposed method


h iT h iT
Define x = xT (1), · · · , xT (L) , s = sT (1), · · · , sT (L) ,
h iT h iT
z = zT (1), · · · , zT (L) , and n = nT (1), · · · , nT (L) . As
aforementioned, we can obtain a compact form of the whole
 
observation vector as
yC = H̄x + z + n, (5)
where H̄ = (H ⊗ IL ). The covariance of interference plus noise
in the communication system can be expressed as
n o
RCin = E (z + n) (z + n)H
P   H (6)
Fig. 1: A MIMO communication system sharing spectrum with = σ2p G̃ (p) ssH G̃ (p) + σC2 I Mr L ,
P
a colocated MIMO radar system. p=1

where G̃ (p) = Ḡ (p) ⊗ IL with Ḡ (p) = a∗r ϕ p vtH θ p . It


   

should be noted that the instaneous information rate may


represents the radar transmit waveform vector, and e jαil is be difficult to determine if the radar interference plus noise
the random phase offset between the radar transmitter and is not necessarily Gaussian. Instead, a lower bound for the
the communication receiver [6]. For the radar-communication rate can be achieved when the communication codeword x is
channel, we adopt a geometric-based channel model, extend- distributed as CN (0, R x ) [14]. According to [13], the lower
ing to the case of co-existing systems the one outlined in bound for the rate per channel use and per Degree Of Freedom
[1] with reference to MIMO communications1 . Specifically, (DOF) can be defined as
the channel matrix can be written in terms of the physical ∆  
propagation path as C (R x , s) = Mt M
1
log2 det I Mr L + RCin
−1
H̄R x H̄H
rL (7)
P
bits/DOF/channel use.
β(p) a∗r ϕ p vtH θ p ,
X    
G1 = (3) In the following, we aim to maximize the lower bound
p=1 for the communication transmit rate in (7). Concerning the
where β(p) denotes the fading coefficient of the pth propaga- constraints, we begin with those forced on the radar transmit
tion path. We assume that β(p), p = 1, 2, · · · , P, are zero-mean waveform [15]. In particular, the energy of the transmit
waveforms is limited, i.e., ksk2 = ER . Besides, we expect
h i
uncorrelated random variables with variance σ2p = E |β(p)|2 .
ϕ p and θ p are the angle of arrival and the angle of departure, re- the designed transmit waveform to lie in the neighborhood
spectively. ar (·) and vt (·) respectively denote the corresponding of a reference waveform s0 (ks0 k2 = ER ) with desirable
Mr × 1 communication receive steering vector and N × 1 radar properties (e.g., in terms of range resolutions, side-lobe level,
transmit steering vector, which have the following general and envelope constancy), yielding the similarity constraint
form ks − s0 k2 ≤ ε, where ε ∈ [0, 2ER ] is a real parameter ruling the
h iT extent of the similarity. In regard to the communication system,
ar (ϕ) = √1M 1 e j2πd1r sinϕ/λ · · · e j2π(Mr −1)d1r sinϕ/λ , we force an
hr iT (4) n energyo constraint on the transmit waveforms,
vt (θ) = √1N 1 e j2πd2t sinθ/λ · · · e j2π(N−1)d2t sinθ/λ , namely E Tr xxH = Tr (R x ) ≤ EC . Mathematically, the
optimization problem can be formulated
where λ is the carrier wavelength, d1r and d2t denote the  
interelement spacing in the communication receive and radar max log2 det I Mr L + RCin
−1
H̄R x H̄H ,
R x ,s
transmit arrays, respectively. Note that in the radar receiver, s.t. Tr (R x ) ≤ EC , (8)
the communication signal can be considered as radar signal- ks − s0 k2 ≤ ε,
independent interference due to the communication signal ksk2 = ER , R x  0.
model, which can be suppressed through adaptive beamform-
ing techniques. By iteratively suppressing the radar signal- Problem (8) is non-convex and it is difficult to obtain the
independent interference from the communication system with optimal solution [16]. To solve this optimization problem, we
the radar receive filter [12], our recent work in [13] aimed adopt the alternating method in the following two steps.
to maximize the radar output signal-to-interference-plus-noise In the first step, we optimize R x while fixing s:
ratio (SINR) under the constraint of the communication rate. 
max log2 det I Mr L + RCin−1

H̄R x H̄H ,
In contrast, this paper focuses on the transmit design for the Rx (9)
coexistence system. The goal is to maximize the transmission s.t. Tr (R x ) ≤ EC , R x  0.
rate of the communication system, subject to the transmit It is easy to see that problem (9) is convex and admits a
constraints of the coexistence system. solution based on the water-filling procedure as [17]
R?x = UΣUH , (10)
1 The
radar receiver design is beyond the scope of this paper, and the channel
matrix G2 in Fig. 1 is ignored, which does not affect our proposed design. where U is the right singular matrix of H̃ = RCin
−1/2
H̄, Σ =

1549-7747 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TCSII.2018.2822845, IEEE
Transactions on Circuits and Systems II: Express Briefs
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMSŁII: EXPRESS BRIEFS 3

+ ∆
Diag(γ1 , · · · , γr ) with γi = η − 1/σ2i , (·)+ = max (·, 0), r and
  
ship Tr (DS) = Tr DssH = sH Ds and removing the constant
{σi }ri=1 denote the rank and positive singular values of H̃. η is terms, problem (16) can be equivalently transformed into the
NP
min following quadratic programming:
chosen such that γi = EC , where Nmin , min (Mt L, Mr L). It
i=1 min sH Ds,
can be shown that the corresponding optimized communication s
rate is s.t. ks − s0 k2 ≤ ε, (17)
Nmin
γi λ2i  ksk2 = ER .
X  
Cmax (EC ) = log2 1 + 2 ,

(11)
i=1
σC To determine the solution to problem (17), we first consider
the above problem without the similarity constraint, i.e.,
where λi is the ith eigenvalue of H̃. Note that: (1) Cmax (EC )
is a monotonically increasing function with respect to EC ; (2) min sH Ds, s.t. ksk2 = ER . (18)
s
Cmax (EC ) will achieve the largest achievable communication
rate if there is no interference from radar transmitter to the It is readily known that the corresponding solution sobt of
communication receiver, i.e., the radar transmits in the null problem (18) is the eigenvector of D corresponding to its
space of the interference matrix channel G1 to the communi- smallest eigenvalue, scaled so that ksobt k2 = ER . If the
cation receiver. obtained solution satisfies the similarity constraint, then it is
?
In the second step, s is optimized with fixed R x . The our sought solution  s to problem (17). Otherwise, we have
optimization problem can be cast as ε < 2ER − 2R sobt s0 ≤ 2ER .
H

  Consider the Lagrange function of problem (17)


max log2 det I Mr L + RCin
−1
H̄R x H̄H ,  
s f1 (s, λ̃, µ̃) = sH Ds+λ̃(ksk2 − ER )+ µ̃(2ER −2R sH s0 −ε), (19)
s.t. ks − s0 k2 ≤ ε, (12)
ksk2 = ER . where λ̃ and µ̃ are the real-valued Lagrange multipliers with
µ̃ ≥ 0 and λ̃ satisfying
Problem (12) is non-convex since we want to maximize a
convex function and ksk2 = ER defines a non-convex set. D + λ̃I  0, (20)
Define S = ssH and S0 = s0 s0H [18]. Problem (12) can be
further transformed to the following matrix formulation so that the cost function can be minimized with respect to s.
  (20) means that λ̃ should be greater than the opposite of the
max log2 det I Mr L + RCin
−1
H̄R x H̄H , smallest eigenvalue of D. After some algebraic manipulations,
S0
2
(13) (19) can be rewritten as
s.t. Tr (SS0 ) ≥ ER − 2ε ,

 
Tr (S) = ER , rank (S) = 1. f1 s, λ̃, µ̃ = [s − µ̃(D + λ̃I)−1 s0 ]H (D + λ̃I)[s − µ̃(D + λ̃I)−1 s0 ]
−µ̃2 s0H (D + λ̃I)−1 s0 − λ̃ER + µ̃ (2ER − ε) .
Observe that problem (13) is still non-convex. Interestingly, (21)
the first order Taylor approximation can be used as a lower The optimization problem (21) resembles the design in [20].
bound of the objective function [19]. Let S = s sH , where s Then, the unconstrained minimization of f1 s, λ̃, µ̃ with re-

is updated as the solution of previous iteration. Then, the first
spect to s for fixed λ̃ and µ̃ leads to
order Taylor expansion of C(R x , S) around S is given by
 −1
     s = µ̃ D + λ̃I s0 (22)
C(R x , S) ≈ C R x , S̄ − Tr D S − S̄ (14)
with ε
with ER −
∆  ∂C(Rx ,S) T µ̃ = 2
−1
, (23)
D=− ∂S s0H (D + λ̃I) s0
S=S̄ !−1
P  P
and λ̃ being the solution to the following equation
H  P  H
= G̃ (p) S̄ G̃ (p) + σC2 I Mr L
P
G̃ (p) 
p=1 p=1
P
!−1    ∆ s0H (D + λ̃I)−2 s0 ER
f2 λ̃ = h i2 =  2 .
H
(24)

G̃ (p) S̄ G̃ (p) + σC I Mr L + H̄R x H̄  G̃ (p) .

P 2 H

ER − 2ε
−1
p=1 s0H (D + λ̃I) s0
(15)
It is clear that the matrix D is positive semi-definite [19]. Then Substituting (23) into (22) yields
problem (13) can be recast as  −1
 ε  D + λ̃I s0
s = ER − . (25)
    
max C R x , S̄ − Tr D S − S̄ , 2 sH D + λ̃I−1 s
S0 0
2 0
s.t. Tr (SS ) ≥ E − ε , (16)

0 R 2 From (25), we can obtain a closed-form expression for s with
Tr (S) = ER , rank (S) = 1. 
given λ̃. It is shown that f2 λ̃ is a monotonically decreasing
Problem (16) can be solved by applying semidefinite relax- function of λ̃. Thus, the value of λ̃ can be obtained efficiently
ation (SDR), which requires high computational complexity. via, e.g., a Newton’s method.
To obtain a low-cost solution of problem (16), an efficient Next, we derive lower and upper bound on λ̃. The matrix
algorithm can be implemented based on Lagrangian multiplier D can be eigen-decomposed as D = ΦΛΦH , where the
method, yielding a closed-form solution. Using the relation- columns of Φ contain the eigenvectors of D, and the diagonal

1549-7747 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TCSII.2018.2822845, IEEE
Transactions on Circuits and Systems II: Express Briefs
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMSŁII: EXPRESS BRIEFS 4

elements of the diagonal matrix Λ, γ̃1 ≥ γ̃2 ≥ · · · ≥ γ̃N are In the following, we analyze the computational complex-
the corresponding eigenvalues in the descending order. Let ity of Algorithm 1. For the R x design, the optimal so-
v = ΦH s0 , and vn denote the nth element of v. (24) can be lution requires to calculate the closed-form
  expression in
rewritten as (10), which has the complexity of O Mt3 L3 . For the s de-
N
P |vn |2 sign, the major computational cost comes from the matrix
2
n=1 (γ̃n +λ̃) eigen-decomposition and inversion, yielding a complexity of
 
f2 λ̃ = " #2 . (26)
PN
|vn |2
O(N 3 L3 ). Overall, the
 computational complexity of the itera-
n=1
(γ̃n +λ̃) tive procedure is O kN 3 L3 + kMt3 L3 .
From the equations (24) and (26), we have
IV. Simulation Results
N
P |vn |2 ks0 k2 In this section, we present numerical results to verify the
2
(γ̃1 + λ̃)
2
ER n=1 (γ̃n +λ̃)
2
(γ̃N +λ̃) performance of the proposed method. Consider a MIMO
=" #2 ≤ = . (27)
ε 2 ks0 k4
ER (γ̃N + λ̃)
2 communication system consisting of Mt = 5 transmit elements
 
N
ER − P |vn |2
(γ̃1 +λ̃)
2
2
n=1
(γ̃n +λ̃) and Mr = 5 receive elements spaced half a wavelength apart.
The colocated MIMO radar is equipped with a uniform linear
According to (20) and (27), we can obtain the following array of N=6 elements spaced half a wavelength. For the
lower bound and upper bound on λ̃ similarity constrained design, the orthogonal linear frequen-
γ̃1 − ER
ER − 2ε γ̃N cy modulation (LFM) waveform is chosen as the reference
−γ̃N ≤ λ̃ ≤ ER
. (28) waveform s0 given in [15]. Specifically, the (n, l)th entry of
ER − 2ε −1 waveform matrix S0 is
√ 2 exp{ j2πn(l−1)/N} exp{ jπ(l−1)2 / N }
S0 (n, l) =
2 
Moreover, since ER vH N s0 = R s H
s0 < (ER − ε/2)2 with √
LN
(29)
vN denoting the last eigenvector in Φ, there exists a unique
where n = 1, 2, · · · , N, and l = 1, 2, · · · , L. Obviously, we
solution λ̃ ≥ −γ̃N to equation (24). Then λ̃ can be obtained
have s0 = vec(S0 ). It is assumed that σC2 = 0.001, and
by solving equation (26), based on the knowledge that the
EC = ER = 1. The entries of H are independent and
solution is unique and it has an upper bound and a lower
identically distributed Gaussian variables with zero mean and
bound. Consequently, problem (12) can be solved and admits
unit variance. For the channel matrix G1 , without loss of
a closed-form solution.
generality, we assume P = 21 paths in the links between the
It is easy to show that the objective function in (8), i.e., radar and the communication system, and the angle parameters
the transmission rate of the communication system is non-
θ p = ϕ p are uniformly generated in the range . of [−30 , −10 ].
◦ ◦
decreasing during the alternating iterations of R x and s, and
The interference-to-noise-ratio (INR = σ p σC ) equals to 15
2 2
is upper bounded. According to the convergence theorem in
dB. As to the stop criteria of the proposed algorithm, we set
[21], [22], the alternating method is guaranteed to converge.
δ = 10−3 .
The overall procedure of the alternating method is summarized
in Algorithm 1. The iteration procedure will continue until the
2.05
termination condition is reached.
2
Communication Rate

Algorithm 1 Iterative Procedure for Jointly Optimizing s and


1.95 ε=1.8
Rx
ε=0.5
Input: H, G1 , ER , EC , s0 , δ. 1.9 ε=0.1
Initialize: Set the iteration index k = 0, s(k) = s0 ; ε=0.01
1.85
1) Compute R(0) x according to (10);
2) Repeat 1.8

3) Set k ← k + 1; 1.75
4) Solve problem (18) with R x replaced by R(k−1) x to 0 5 10
Number of Iterations
15

obtain sobt ;
Fig. 2: The communication rate versus iteration number for
5) If ksobt − s0 k2 ≤ ε then
radar similarity parameter ε = 0.01, 0.1, 0.5, 1.8 (s0 in [15] as
6) s(k) = sobt ;
the initial point).
7) Else
8) Solve equation (26) for λ̃;
9) Calculate s(k) according to (25); First, we analyze the performance of the devised algorithm
10) End if for a MIMO communication system with different similarity
Calculate R(k) parameter values. Fig. 2 shows the communication rate versus
11) x according to (10) with s replaced by
(k)
s ;  the iteration number. The results highlight that the communi-
  (k+1)  cation rate is increasing with the iteration number, and the
12) Until C R(k) , (k)
, < δ;
(k+1)

x s − C R x s
proposed method converges very fast. Additionally, higher
Output: R?x = R(k+1)
x , s? = s(k+1) . communication rate can be achieved for larger ε. The main
reason is that as ε increases, the feasibility set gets larger. This

1549-7747 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TCSII.2018.2822845, IEEE
Transactions on Circuits and Systems II: Express Briefs
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMSŁII: EXPRESS BRIEFS 5

trend highlights that the radar waveform design successfully References


allocates the radar transmit energy in directions that can [1] E. Yousif, F. Khan, T. Ratnarajah, and M. Sellathurai, “On the spectral
effectively reduce the interference energy at communication coexistence of colocated MIMO radars and wireless communications
receiver. systems,” in Proc. IEEE 17th Workshop Signal Process. Adv. Wireless
Commun, (SPAWC). IEEE, 2016, pp. 1–5.
Next, we evaluate the effect of the interference channel G1 [2] H. Deng and B. Himed, “Interference mitigation processing for
with different σ2p . For comparison, we also implement the spectrum-sharing between radar and wireless communications systems,”
method which only optimizes communication system R x with IEEE Trans. Aerosp. Electron. Syst., vol. 49, no. 49, pp. 1911–1919,
2013.
fixed s, and the radar transmit antennas use the reference code [3] A. Aubry, A. De Maio, M. Piezzo, and A. Farina, “Radar waveform
s0 (labeled as “Selfish Comm.”). Other parameters remain the design in a spectrally crowded environment via nonconvex quadratic
same as in the first scenario. Fig. 3 shows the communication optimization,” IEEE Trans. Aerosp. Electron. Syst., vol. 50, no. 2, pp.
1138–1152, 2014.
rate versus INR from radar. As expected, the communica- [4] A. Aubry, A. De Maio, Y. Huang, M. Piezzo, and A. Farina, “A new
tion rate decreases with the increasing INR. Specially, the radar waveform design algorithm with improved feasibility for spectral
communication rate of the “Selfish Comm.” method degrades coexistence,” IEEE Trans. Aerosp. Electron. Syst., vol. 51, no. 2, pp.
1029–1038, 2015.
severely in high INR region because the method which only [5] B. Li and A. Petropulu, “Spectrum sharing between matrix completion
optimizes communication system R x without s design has based mimo radars and a mimo communication system,” in Proc. IEEE
no ability to reduce the radar transmit energy towards the Int. Conf. Acoust., Speech, Signal Process. (ICASSP), 2015, pp. 2444–
2448.
communication receiver. In contrast, the curve of the joint [6] B. Li, A. P. Petropulu, and W. Trappe, “Optimum co-design for spectrum
design method for ε = 1.8 stays almost steady, indicating sharing between matrix completion based mimo radars and a mimo
the effectiveness of the proposed method. Additionally, we communication system,” IEEE Trans. Signal Process., vol. 64, no. 17,
pp. 1–1, 2016.
can observe that the performance of the joint design method [7] B. Li, H. Kumar, and A. P. Petropulu, “A joint design approach for
gets worse as ε decreases, which results from the similarity spectrum sharing between radar and communication systems,” in Proc.
parameter influence. The results suggest a suitable tradeoff IEEE Int. Conf. Acoust., Speech, Signal Process. (ICASSP), 2016, pp.
3306–3310.
between optimizing the communication rate and controlling [8] B. Li and A. Petropulu, “Joint transmit designs for co-existence of mimo
other desired radar waveform properties. wireless communications and sparse sensing radars in clutter,” IEEE
Trans. Aerosp. Electron. Syst., vol. PP, no. 99, pp. 1–1, 2017.
[9] N. Huang, X. Wang, and M. Chen, “Transceiver Design for MIMO VLC
2.1 Systems with Integer-Forcing Receivers,” IEEE J. Sel. Areas Commun.,
vol. PP, no. 99, pp. 1–1, 2017.
[10] J. Qian, Z. He, W. Zhang, Y. Huang, N. Fu, and J. Chambers, “Robust
Communication Rate

2
adaptive beamforming for multiple-input multiple-output radar with
spatial filtering techniques,” Signal Process., vol. 143, pp. 152–160,
1.9
2018.
[11] A. M. Sayeed and S. Member, “Deconstructing multi-antenna fading
Joint Design,ε=1.8
1.8 channels,” in IEEE Trans. Signal Process., 2002, pp. 2563–2579.
Joint Design,ε=0.5
[12] S. Wang, J. Feng, and K. T. Chi, “Kernel Affine Projection Sign
Joint Design,ε=0.1
1.7 Algorithms for Combating Impulse Interference,” IEEE Trans. Circuits
Joint Design,ε=0.01
Syst.II, Exp. Briefs, vol. 60, no. 11, pp. 811–815, 2013.
Selfish Comm.
[13] J. Qian, M. Lops, L. Zheng, and X. Wang, “Joint Design for Co-
1.6
−20 −10 0 10 20 30 existence of MIMO Radar and MIMO Communication System,” in Proc.
INR (dB) 51st Asilomar Conf. Signals, Syst. Comput. 2017 (Invited paper).
[14] S. N. Diggavi and T. M. Cover, “The worst additive noise under a
Fig. 3: The achieved communication rate under different INRs. covariance constraint,” IEEE Trans. Info. Theory, vol. 47, no. 7, pp.
3072–3081, 2001.
[15] G. Cui, X. Yu, V. Carotenuto, and L. Kong, “Space-time transmit code
V. Conclusion and receive filter design for colocated mimo radar,” IEEE Trans. Signal
Process., vol. 65, no. 5, pp. 1116–1129, 2017.
In this paper, we have proposed a novel approach with fast [16] S. Boyd and L. Vandenberghe, Convex optimization. Cambridge
convergence and closed-form transmit designs for spectral co- university press, 2004.
existence of MIMO radar and MIMO communication system. [17] D. N. C. Tse and P. Viswanath, Fundamentals of wireless communica-
tion. Cambridge, U.K.: Cambridge Univ. Press, 2005.
The radar transmit waveform, and the communication transmit [18] A. De Maio, S. De Nicola, Y. Huang, S. Zhang, and A. Farina, “Code
covariance matrix have been jointly designed to maximize design to optimize radar detection performance under accuracy and
the communication rate while meeting similarity and energy similarity constraints,” IEEE Trans. Signal Process., vol. 56, no. 11,
pp. 5618–5629, 2008.
constraints on the coexistence system. We have shown that the [19] Y. Sun, P. Babu, and D. P. Palomar, “Majorization-minimization algo-
proposed method successfully maximizes the communication rithms in signal processing, communications, and machine learning,”
rate and nullifies the radar energy towards the communication IEEE Trans. Signal Process., vol. 65, no. 3, pp. 794–816, 2017.
[20] J. Li, J. R. Guerci, and L. Xu, “Signal waveform’s optimal-under-
receiver. As possible future research tracks, it might be worth restriction design for active sensing,” IEEE Signal Process. Lett., vol. 13,
analyzing performance of radar system or investigating the no. 9, pp. 565–568, 2006.
robust design when the prior knowledge of the channel matrix [21] S. Wang, W. Wang, S. Duan, and L. Wang, “Kernel Recursive Least
Squares with Multiple Feedback and Its Convergence Analysis,” IEEE
and cross interference is mismatched. Trans. Circuits Syst.II, Exp. Briefs, vol. PP, no. 99, pp. 1–1, 2017.
[22] J. Yeh, Real analysis: theory of measure and integration. World
Acknowledgment Scientific, 2006.
This work was supported by the National Natural Science
Foundation of China under Grants 61671139 and 61401062,
and in part by China Scholarship Council (CSC).

1549-7747 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

You might also like