Вы находитесь на странице: 1из 8

Physical Communication 32 (2019) 192–199

Contents lists available at ScienceDirect

Physical Communication
journal homepage: www.elsevier.com/locate/phycom

Full length article

High performance OFDM with index modulation



Yusuf Acar a,b , , Todor Cooklev a
a
Wireless Technology Center, Purdue University-Fort Wayne, USA
b
Department of Electrical and Electronics Engineering, Istanbul Kultur University, 34156, Istanbul, Turkey

article info a b s t r a c t

Article history: Orthogonal frequency division multiplexing (OFDM) with index modulation (OFDM-IM), which has been
Received 15 May 2018 recently proposed as an alternative waveform to classical OFDM, generally achieves lower bit error rate
Received in revised form 10 October 2018 compared to classical OFDM because it is more robust to bit errors by encoding the information in the
Accepted 2 December 2018
location of the active subcarriers instead of only modulating the subcarriers. In this paper, we propose a
Available online 6 December 2018
higher-performance OFDM-IM scheme. The proposed system is more reliable than OFDM-IM, particularly
Keywords: for channels that have deep fading. The proposed system uses generalized prefix (GP) as a guard interval
Index modulation and this GP provides additional degrees of freedom, which can be optimized. Comprehensive computer
OFDM simulation results are included to demonstrate the significant performance improvement for different
Generalized prefix channel models, such as deep Rayleigh fading and realistic channels with high mobility.
Deep fading channels
© 2018 Published by Elsevier B.V.

1. Introduction plain OFDM-IM. The performance of OFDM-IM is analyzed in [9]


considering important system metrics such as Euclidean distance
There has been an increasing interest in effective techniques for and PAPR. Dual-mode OFDM-IM (DM-OFDM-IM) has been pro-
broadband communications systems to meet the requirements of posed to deal with empty subcarriers problem in [10] for increase
reliability, capacity, high data rate and mobility in practical wire- the spectral efficiency. Moreover, the OFDM-IM structure has been
less environments. Orthogonal frequency division multiplexing applied to visible light communications (VLC) [11,12], and MIMO
(OFDM) is being used by all broadband communications systems systems [13–16]. For further details on this topic, the interested
such as IEEE 802.11 (Wi-Fi), long term evolution (LTE), etc. [1– readers are referred to [17]. As a result, future wireless systems
3]. Because, OFDM has become the most popular modulation will require higher-performance modulation techniques. It is also
technique, it continues to be extensively studied. OFDM with likely that multicarrier modulation will be the foundation of these
index-modulation (OFDM-IM) has been recently proposed by Basar techniques. Therefore, OFDM-IM is a candidate for future wireless
et al. [4]. The essential idea of OFDM-IM is to use the indices systems.
of the subcarriers to carry information. Some carriers are set OFDM-IM relies on the robustness of encoding information in
to zero (i.e. inactive); the indices of the active subcarriers carry the subcarrier locations. If this is not the case, then OFDM-IM
information. Furthermore, M-ary PSK or M-ary QAM is used on loses its advantages to OFDM. If the channel has a spectral null
the active subcarriers. Encoding information in the location of the for an active subcarrier f , then the determination of the set of
active subcarriers is robust, and OFDM-IM achieves better average active subcarriers and the symbol transmitted on that subcarrier
bit error probability (ABEP) compared with classical OFDM while both have high probabilities of errors, regardless of the signal-
it exhibits comparable decoding complexity. to-noise ratio (SNR). An error in determining the set of active
At present OFDM-IM is being actively investigated. In [5], OFDM subcarriers will impact all subcarriers. In other words, for high SNR
with generalized index modulation is proposed by applying a more values spectral nulls introduce an error floor or a loss in frequency
flexible subcarrier activation procedure. To obtain additional di- diversity. As a result, the advantages of OFDM-IM may not always
versity gains, the authors of [6] proposed a linear constellation pre- be obtainable in practice since Rayleigh channels often have deep
coded OFDM-IM scheme. To improve its error performance, block fades and spectral nulls.
interleaving is considered for OFDM-IM in [7]. In [8], coordinate The proposed prefix structure is an appropriate solution to
interleaving is introduced to further improve the diversity order of address this problem. OFDM-IM requires a guard interval between
the multi-carrier symbols. All OFDM-IM systems that have been
∗ Corresponding author at: Department of Electrical and Electronics Engineer- implemented use a cyclic prefix (CP) as a guard interval. If the pre-
ing, Istanbul Kultur University, 34156, Istanbul, Turkey. fix is longer than the channel delay spread the frequency selective
E-mail addresses: acary@pfw.edu, y.acar@iku.edu.tr (Y. Acar), fading channel is converted into multiple parallel and independent
cooklevt@pfw.edu (T. Cooklev). subchannels, each characterized by flat fading. In the literature,

https://doi.org/10.1016/j.phycom.2018.12.001
1874-4907/© 2018 Published by Elsevier B.V.
Y. Acar and T. Cooklev / Physical Communication 32 (2019) 192–199 193

Table 1 active. As a result, the total transmitted bits of each subblock can
OFDM-IM parameters and symbols. be calculated as
Parameters Definition
m Number of information bits in each OFDM block
⌊log2 (C (n, k))⌋ + log2 (M)k = m1 + m2 , (1)
g Number of groups each containing p bits
where M corresponds to modulation size, C (n, k) denotes the bi-
p Number of bits per group (i.e., p = m/g)
N Number of OFDM subcarriers nomial coefficient and ⌊.⌋ is the floor function.
NC Length of cyclic prefix Which subcarrier indices are active is shown in Table 2. For
F Discrete Fourier transform (DFT) matrix example, when the incoming bit sequence is [0, 0] the first and the
K Number of non-silent subcarriers (i.e. K = kg) second subcarriers contain modulated symbols. The other subcar-
OFDM subblock length (i.e. n = N /g)
n
riers are not used. At each sub-block β , indices are selected by the
k Number of active subcarriers in each subblock
p1 Total number of bits that are mapped onto the active index selector. The selected indices are given as follows:
indices in each subblock
Iβ = iβ,1 . . . iβ,k ,
{ }
p2 Total number of modulation bits in each subblock (i.e.
(2)
p2 = klog2 M)
where iβ,γ ∈ [1, . . . , n] for β = 1, . . . , g and γ = 1, . . . , k.
m1 The total number of data bits conveyed by indices in the
OFDM block
The modulated symbols at the output of the M-ary QAM mod-
m2 The total number of data bits conveyed by the modulations ulator are given by the vector
Iβ Selected indices for subblock β
sβ = sβ (1) . . . sβ (k) ,
[ ]
sβ Modulated symbol for subblock β (3)
where sβ (γ ) ∈ S , β = 1, . . . , g , γ = 1, . . . , k. By using Iβ and sβ
Table 2 for all β , the OFDM block generator creates all of the sub-blocks
A look-up table of OFDM-IM scheme for n = 4, k = 2, p1 = 2. and then creates the N × 1 OFDM-IM symbols as follows:
Bits Subblock indices Subblock symbols
x = [x(1) x(2) . . . x(N)]T , (4)
[0, 0] {1,2} [s1 , s2 , 0, 0]T
[0, 1] {2,3} [0, s1 , s2 , 0]T where x(n) ∈ {0, S } , n = 1, . . . , N. As shown in Fig. 3 there
[1, 0] {3,4} [0, 0, s1 , s2 ]T are some zero subcarriers the position of which also carries infor-
[1, 1] {1,4} [s1 , 0, 0, s2 ]T
mation unlike the conventional OFDM. Then, N-point inverse fast
Fourier transform (IFFT) of the vector x is obtained as X = FH x,
where FH is the N-point inverse FFT matrix.
there is no work on other prefix constructions for OFDM-IM sys-
tems. 2.2. Proposed method
This paper proposes a generalized prefix (GP) as a guard inter-
In the proposed system, the guard interval consists of a prefix
val [18]. Similar to the CP, this prefix construction decomposes the
constructed by multiplying the last C samples with a complex
channel into N parallel and independent subchannels. However,
number ψ as shown in Fig. 2. This is equivalent to:
the difference is that the fades and nulls of these subchannels can
be less deep. The GP can be optimized as an additional link adap- X = [diag {D}]−1 FH x, (5)
tation technique. CP is not the optimal guard interval; the reason
where diag {D} is a diagonal matrix with ψ along the main diago-
n
that known OFDM-IM systems employ a CP is convenience and
nal:
computational simplicity for detection and synchronization. The
1
⎡ ⎤
proposed GP maintains these advantages and, at the same time,
increases the robustness of the subcarrier location information. ⎢ ψ ⎥
While the proposed technique always performs better than the diag {D} = ⎢
⎢ ψ2 ⎥.

(6)
known OFDM-IM, when the transmission is over Rayleigh channels
⎢ .. ⎥
⎣ . ⎦
the performance improvement is even more considerable.
ψ N −1
The rest of our paper is organized as follows. In Section 2, we
describe the proposed system. The optimization procedure is dis- Then a guard interval is added of length C which can be represented
cussed in Section 3. In Section 4, the detection procedure is given. in matrix form as
Extensive computer simulation results are provided in Section 5. xψ = Gψ X, (7)
Finally, conclusions are given in Section 6. 1
where
φ.IC
[ ]
2. High-performance OFDM-IM ψ O
G = C ×(N −C ) . (8)
IN
2.1. OFDM-IM
The complex numbers φ and ψ are related, φ = ψ N , and their
absolute values are one. Hence, we can maintain the same PAPR.
In this paper, we consider an OFDM-IM system, illustrated in After the addition of a GP, parallel-to-serial and digital-to-
Fig. 1, operating over a frequency-selective Rayleigh fading chan- analog conversions, this signal is sent through a L-tap Rayleigh
nel. The parameters of the system are summarized in Table 1. The fading channel given by
subcarriers are split into g groups or sub-blocks of n subcarriers
each. In each group only k out of the n available subcarriers are hT = [hT (1) . . . hT (L)]T , (9)
where the coefficients are circularly symmetric complex Gaussian
1 Notation: Small and bold letters like ‘a’ denote vectors. Capital and bold letters random variables, hT (l) ∈ CN (0, 1L ), l = 1, . . . , L.
like ‘A’ denote matrices. The notations (.)T , (.)H , ∥.∥ and (.)−1 denote transpose, At the receiver, after removing the prefix (by applying M),
Hermitian (transpose, Euclidean norm and inverse of a vector or a matrix, respec- matrix multiplication with diag {D}, and finally the FFT, one obtains
tively. CN 0, σ 2 is circularly symmetrical Gaussian distribution with variance σ 2 .
)
that:
S denotes the complex signal constellation of size M. IC denotes the C × C identity
matrix. OC ×M denotes the C × M zero matrix. y = F[diag {D}]MHT xψ + w, (10)
194 Y. Acar and T. Cooklev / Physical Communication 32 (2019) 192–199

Fig. 1. The system model of the OFDM-IM.

Fig. 2. The system model of the proposed OFDM-IM system.

where M = [ON ×C IN ON ×(L−1) ], w = F[diag {D}]Mn is an N × 1 and


frequency-domain noise vector with variance N0,F which is related N −1
ψ
to the noise variance in the time domain as N0,F = (K /N)N0,T and ψ n hT [n] e−j2π fn .

hF ( α ) = (13)
HT is an (N + C + L − 1) × (N + C ) channel matrix expressed as: n=0

In other words, the channel (9) has been transformed into an


hT (1) 0 ··· 0 equivalent channel with transfer function
⎡ ⎤
⎢hT (2) hT (1) 0 ··· 0 ⎥ N −1
⎢ . .. H ψ (z) =

⎢ . ψ n hT (n)z −n ,

(14)
⎢ . .


⎢ h (L) n=0
··· hT (1) 0 ··· 0

⎢ T
⎥.

⎢ ..
HT = ⎢ .. ⎥ where ψ = ejα is a new variable; ψ = 1 corresponds to CP.
⎢ . . ⎥
3. Calculation of ψ for OFDM-IM system
⎢ 0 ··· 0 hT (L) ··· hT (1) ⎥


⎢ . .. ⎥
⎣ .. .

Fig. 3 shows an illustrative example where two subcarriers (16
0 ··· 0 hT (L) and 17) out of 32 are affected by a spectral null. In the case of
ψ conventional OFDM systems only 2 subcarriers out of 32 (6.25%)
Because the product h̄F = F[diag {D}]MHT Gψ [diag {D}]−1 FH is will be impacted. For OFDM-IM assuming group length n = 4 and
a diagonal matrix, the received signal can be written as follows: active subcarriers k = 2, 2 symbols out of 16 will be impacted, or
ψ 12.5%. Furthermore, if the group length is increased more symbols
y = h̄F .x + w, (11)
will be affected by the spectral null of channel. For example, if
Therefore, the channel frequency response is n = 16 and k = 2, all symbols will be affected. As seen from
the figure, the proposed technique eliminates the impact of the
ψ ψ ψ ψ
h̄F = diag[hF (0) hF (1) . . . hF (N − 1)] (12) spectral null.
Y. Acar and T. Cooklev / Physical Communication 32 (2019) 192–199 195

Fig. 3. Comparison of OFDM-IM systems.

3.1. Optimization

The impact of optimizing α on the BER performance is ]illus-



trated in Fig. 4. Assuming the two path channel h = 1 1 / 2
[
which has a spectral null. For this channel model Fig. 4 shows the
BER of OFDM-IM systems as a function of α . The optimum value
of α depends on n, k, and the SNR. The value α = 0 corresponds
to the cyclic prefix case, and it is seen that α = 0 is far from
optimum. Since the BER strongly depends on α , optimizing α is
important. For example, as seen in Fig. 4, the use of CP achieves
BER= 6 × 10−2 , the optimum value of α leads to BER= 1.5 × 10−5 ,
an improvement ratio of about 4 × 102 for n = 4 and k = 1!
Therefore, the optimization of α improves the frequency diversity
of the system.
The optimum frequency shift α ∗ can be determined using an
optimization criterion given by
α ∗ = arg min ] Pe (ψ )|ψ=ejα , (15)
α∈ 0, 2Nπ
[

where Pe (ψ ) is the average bit error probability of the proposed Fig. 4. The impact of α on the BER performance of proposed GP OFDM-IM systems
OFDM-IM which can be expressed as follows [11]-(20): with BPSK.

1 ∑∑
Pe (ψ ) ≈ Pψ (Y → Ŷ)e(Y, Ŷ) (16)
pnY procedures as in [18] cannot be used directly. Furthermore, the
Y Ŷ
optimization procedure not only requires channel state informa-
where p is total number of bits per subblock (i.e., p = m/g), e(Y, Ŷ) tion (CSI), but also knowledge of n and k. Therefore, the optimiza-
represents the number of bit errors for the corresponding pairwise tion must be done at the transmitter. This is possible since many
error event and nY is the number of the possible[ realizations of modern wireless systems have evolved to the point where CSI is
Y [4]. The search space is limited to the interval 0, 2Nπ because
]
provided to the transmitter.
the frequency response (13) is periodic with respect to α and the
period is 2Nπ .
There are many optimization algorithm to find the optimum 4. Detection
frequency shift α ∗ . One of the efficient algorithm which is used in
this woks is the golden section search method. In this method [p, q] The detection problem for OFDM-IM is more challenging than
is a given interval and the points a and b can be calculated as: the detection problem for OFDM. It is clear that, the receiver not
only needs to detect the M-ary symbols, but also the indices of
a = q − (q − p)κ, b = p + (q − p)κ, (17)
√ active subcarriers for the OFDM-IM scheme. There are two main
where κ = 5−1
is the golden section ratio. Then, a new search
2
types of detectors, based on log-likelihood ratio (LLR) and maxi-
interval is obtained according to whether Pe (eja ) ≤ Pe (ejb ) or not. mum likelihood (ML) [4].
Note that the optimum value of α depends on n, k, and the In order to make a joint decision on the M-ary constellation
SNR (i.e. α ∗ = α (n, k, L)). Therefore, one-dimensional optimization symbols and the active indices it is possible to use a ML detector
196 Y. Acar and T. Cooklev / Physical Communication 32 (2019) 192–199

Fig. 7. BERs as a function of L for the standard and proposed OFDM-IM systems
Fig. 5. BERs as a function of k for the standard and proposed OFDM-IM systems with n = 4 and k = 2.
with n = 32.

with different scenarios. The SNR is defined as Es /N0 where Es


is energy per symbol and N0 is the noise power. The number of
subcarriers N is set to 128. At the receiver, a LLR detector is used
which is computationally simpler than ML.

5.1. Scenario-1

In this scenario, we consider the BERs as a function of subblock


parameters such as n and k. Fig. 5 shows the effect of increasing
k on the BER performance of the proposed scheme. As k increases
the OFDM-IM system approaches the classical OFDM system. It is
shown that, the proposed method performs better than the CP case
for all values of k. For example, the conventional OFDM-IM with
16-QAM and k = 5 at SNR = 40 archives BER = 2 × 10−5 while the
proposed method leads to BER = 10−6 . The improvement ratio is
about 20.
The effect of increasing n on the BER performance of the pro-
Fig. 6. BERs as a function of n for the standard and proposed OFDM-IM systems. posed scheme is illustrated in Fig. 6. The performance of OFDM-
IM system gets better while n increases. Hence, as n increases
the improvement brought by the proposed OFDM-IM system also
minimizing the following metric for each subblock increases. For instance, the conventional OFDM-IM with 16-QAM,
k n = 32 and k = 3 at SNR = 35 archives BER = 5 × 10−5 , while the
proposed method leads to BER = 7 × 10−6 . The results illustrate
( )
⏐y (iβ,γ ) − hβ (iβ,γ )sβ (γ )⏐2
∑ ⏐ β
Îβ , ŝβ

= arg min F F (18)
Iβ ,sβ that taking these parameters into account during the optimization
γ =1
procedure is important.
β β
where yF (ξ ) = yF (n(β − 1) +ξ ) and hF (ξ ) = hF (n(β − 1) +ξ ) are the
vectors of the received signal and fading coefficients, respectively. 5.2. Scenario-2
It is seen that, in (18) all possible sub-block realizations are con-
sidered by searching for all possible subcarrier index combinations
In this scenario, we consider the BERs as a function of the
and the signal constellation points. However, the complexity of ML
number of channel coefficients L. Generally, if the channel length is
decoding is not suitable for high values of n and k.
increased there are more deep fade effects so the proposed method
We use the LLR detector which is computationally simpler. It
evaluates the logarithm of the ratio of the posteriori probabilities also performs better. In Fig. 7, the conventional OFDM-IM with 8-
of received signal for each subcarrier: QAM, L = 64, C = 68, n = 4 and k = 2 archives BER = 10−4 at
∑M SNR = 35 dB while the proposed method has BER = 10−6 . In other
χ=1 P(x(f ) = sχ |yF (f ) ) words, the proposed method is 100 times better (!) than standard
λ(f ) = ln , (19)
OFDM-IM.
P(x(f ) = 0 |yF (f ) )
Fig. 8 shows the effect of increasing the number of channel
where sχ ∈ S. It is seen that, the higher the value of λ(f ) the more
coefficients on the BER performance of the OFDM-IM schemes with
likely the subcarrier transmitting x(f ) is active.
n = 4 and k = 2. Again, as seen from Fig. 8, the performance
5. Simulation results advantage of the proposed system increases. For example, at BER=
10−6 the proposed method has about 17 dB (!) better performance
To demonstrate the improved performance of the proposed than the currently known best OFDM-IM scheme with L = 30 and
techniques the BER performance of OFDM-IM systems is evaluated C = 36.
Y. Acar and T. Cooklev / Physical Communication 32 (2019) 192–199 197

Fig. 8. The effect of varying L values on BER performance of proposed GP OFDM-IM Fig. 10. BER performance comparison of CP OFDM-IM and proposed GP OFDM-IM
schemes for 16-QAM with n = 4 and k = 2. with L = 10, C = 16, n = 16 and k = 10.

Fig. 9. BER performance comparison of CP OFDM-IM and proposed GP OFDM-IM Fig. 11. BER performance comparison of CP OFDM-IM and proposed GP OFDM-IM
with L = 10, C = 16, n = 4 and k = 2. under imperfect CSI with L = 10, C = 16, n = 16 and k = 10.

5.3. Scenario-3 of the vector h is ĥ = h + e, where the vector of channel estimation


errors e is independent of h [19]. The covariance matrix of the
In this scenario, the BERs of the known and proposed OFDM- vector e is E {eeH } = σe2 In . In Figs. 11–12, we present computer
IM systems are compared in Figs. 9 and 10 as functions of the simulation results for imperfect CSI with different power of the
SNR. In all cases, the performance improvements are significant. estimation error. In Fig. 11, we assume that the power of the
For example, as seen from Fig. 9, at BER = 10−6 our proposed estimation error decreases with increasing SNR i.e., N0,F /σe2 = 1.
method achieves approximately 10 dB, 11.5 dB, 12 dB and 12.5 It is shown that the proposed method is (about 4 dB) better that CP
dB better performance than the currently known best OFDM-IM OFDM-IM for all cases at BER= 10−4 . In Fig. 12, we assume that
with k = 2 and n = 4 for BPSK, QPSK, 8-QAM and 16-QAM, the power of the estimation error is constant which leads to an
respectively. As seen from Fig. 10 for the higher rate k = 10 and error floor at high SNR values. However, this figure indicates that
n = 16, the proposed method provides significant improvements the error floor for the proposed scheme occurs at much lower bit
than standard OFDM-IM schemes. The figures also illustrate the error rate values.
performance comparison when the interleaved subcarrier index
(OFDM-ISIM) technique is used [8]. It is shown that the proposed 5.5. Scenario-5
method with OFDM-ISIM achieves considerable performance im-
provement compared to GP-OFDM-IM and CP-OFDM-IM with low- In this scenario, we compare the performance under realistic
order modulation schemes such as BPSK and QPSK. channel conditions such as very high mobility. One of the critical
issue of new generation wireless communication systems is the
5.4. Scenario-4 high mobility case. In Fig. 12, we compare the BER performance
of proposed method for a terminal moving at speeds of 200 km/h.
In this scenario we consider the impact of imperfect CSI. It is The channel between the mobile antenna and the receiver antenna
known that, in general, system performance degrades significantly is fast fading based on the Typical Urban (TU) model with 12 taps.
due to the channel estimation error. We assume that the estimate The channel parameters are simulated using the model described
198 Y. Acar and T. Cooklev / Physical Communication 32 (2019) 192–199

Fig. 12. BER performance comparison of CP OFDM-IM and proposed GP OFDM-IM Fig. 14. BER performance comparison of CP OFDM-IM and proposed GP OFDM-IM
under imperfect CSI with QPSK, L = 10, C = 16, n = 4 and k = 2. under channel coding with L = 10, C = 16, n = 4 and k = 2.

6. Conclusions

In this paper, we propose a new prefix structure for OFDM-IM


system. This method provides an additional degree of freedom,
which can be optimized. The prefix of the OFDM-IM symbol is
optimized taking into account several parameters. For practical
Rayleigh channels extensive computer simulations show that the
performance improvement is significant. Therefore, the proposed
technique is a candidate for next-generation practical OFDM-IM
wireless systems.

Acknowledgments

This work has been supported from the Scientific and Tech-
nological Research Council of Turkey (TUBITAK) under the BIDEB-
2219 Postdoctoral Research Program.

References

Fig. 13. BER performance comparison of CP OFDM-IM and proposed GP OFDM-IM [1] B.P. Crow, I. Widjaja, J.G. Kim, P.T. Sakai, Ieee 802.11 wireless local area
under fast fading channel with moving at a speed of 200 km/h with BPSK.
networks, IEEE Commun. Mag. 35 (9) (1997) 116–126.
[2] E. Dahlman, S. Parkvall, J. Skold, 4G: LTE/LTE-advanced for Mobile Broadband,
Academic press, 2013.
in [20]. As seen from Fig. 13, compared to standard OFDM-IM, the [3] S. Sesia, M. Baker, I. Toufik, LTE-the UMTS Long term Evolution: From Theory
proposed method has almost the same performance. to Practice, John Wiley & Sons, 2011.
[4] E. Basar, Ü. Aygölü, E. Panayırcı, H.V. Poor, Orthogonal frequency division
multiplexing with index modulation, IEEE Trans. Signal Process. 61 (22)
5.6. Scenario-6 (2013) 5536–5549.
[5] R. Fan, Y.J. Yu, Y.L. Guan, Orthogonal frequency division multiplexing
In this scenario, we consider our proposed method with channel with generalized index modulation, in: Global Communications Conference
(GLOBECOM), IEEE, 2014, pp. 3880–3885.
coding. In simulations, generator polynomials, constraint length
[6] M. Wen, B. Ye, E. Basar, Q. Li, F. Ji, Enhanced orthogonal frequency division
and coding rate chosen as (171, 133), 7 and 1/2, respectively. At the
multiplexing with index modulation, IEEE Trans. Wireless Commun. 16 (7)
receiver, hard Viterbi decoding algorithm is employed to decode (2017) 4786–4801.
binary data. We consider a coded OFDM system with N = 256 [7] Y. Xiao, S. Wang, L. Dan, X. Lei, P. Yang, W. Xiang, OFDM with interleaved
tones to compare with the uncoded case under the same bit rate. subcarrier-index modulation, IEEE Commun. Lett. 18 (8) (2014) 1447–1450.
Simulation results in Fig. 14 show that the BER performance of the [8] E. Basar, OFDM with index modulation using coordinate interleaving, IEEE
proposed method slightly better than standard OFDM-IM under Wireless Commun. Lett. 4 (4) (2015) 381–384.
channel coding. In particular, coded GP OFDM-IM is shown that [9] N. Ishikawa, S. Sugiura, L. Hanzo, Subcarrier-index modulation aided OFDM-
will it work? IEEE Access 4 (2016) 2580–2593.
about 3 dB gain is achieved at BER value of 10−6 , as compared with
[10] T. Mao, Z. Wang, Q. Wang, S. Chen, L. Hanzo, Dual-mode index modulation
coded GP OFDM-IM for 8−QAM modulation scheme. Moreover, aided OFDM, IEEE Access 5 (2017) 50–60.
GP OFDM-IM with uncoded case has better performance than [11] E. Basar, E. Panayırcı, Optical OFDM with index modulation for visible light
coded CP OFDM-IM case while has two times higher bit spectral communications, in: Optical Wireless Communications (IWOW), 2015 4th
efficiency. International Workshop on, IEEE, 2015, pp. 11–15.
Y. Acar and T. Cooklev / Physical Communication 32 (2019) 192–199 199

[12] T. Mao, R. Jiang, R. Bai, Optical dual-mode index modulation aided OFDM for Yusuf Acar received the B.S.E. degree (with honors),
M.S.E. degree and Ph.D. degrees in Electrical and Elec-
visible light communications, Opt. Commun. 391 (2017) 37–41.
tronics Engineering from Istanbul University, Istanbul,
[13] E. Basar, Multiple-input multiple-output OFDM with index modulation, IEEE
Turkey, in 2008, 2011 and 2015, respectively. From 2009
Signal Process. Lett. 22 (12) (2015) 2259–2263. to 2015 he was employed as a research and teaching
[14] E. Basar, On multiple-input multiple-output OFDM with index modulation for assistant at the faculty of the Department of Electrical
next generation wireless networks, IEEE Trans. Signal Process. 64 (15) (2016) and Electronics Engineering, Istanbul Kültür University.
3868–3878. In 2015, he joined the same faculty as an Assistant Profes-
[15] B. Zheng, M. Wen, E. Basar, F. Chen, Multiple-input multiple-output OFDM sor. He was a visiting scholar at the Purdue University Fort
with index modulation: Low-complexity detector design, IEEE Trans. Signal Wayne (PFW), USA, between the June 2012 and Septem-
Process. 65 (11) (2017) 2758–2772. ber 2012. He is now a researcher at the PFW, Fort Wayne,
USA, until September 2018. His general research interests cover communication
[16] L. Wang, Z. Chen, Z. Gong, M. Wu, Space-frequency coded index modulation
theory, estimation theory, statistical signal processing, and information theory. His
with linear-complexity maximum likelihood receiver in the MIMO-OFDM
current research activities are focused on software-defined platforms.
system, IEEE Signal Process. Lett. 23 (10) (2016) 1439–1443.
[17] E. Basar, M. Wen, R. Mesleh, M. Di Renzo, Y. Xiao, H. Haas, Index modula-
tion techniques for next-generation wireless networks, IEEE Access 5 (2017) Todor Cooklev is the Founding Director of the Wire-
16693–16746. less Technology Center at Indiana Purdue University-Fort
[18] T. Cooklev, H. Dogan, R.J. Cintra, H. Yildiz, A generalized prefix construction Wayne (PFW), Fort Wayne, Indiana, and the Harris Asso-
ciate Professor of Wireless Communication and Applied
for OFDM systems over quasi-static channels, IEEE Trans. Veh. Technol. 60 (8)
Research at the same institution. He has been involved
(2011) 3684–3693.
with the development of several wireless standards. He
[19] V. Tarokh, H. Jafarkhani, A.R. Calderbank, Space-time block codes from orthog- has received research grants and has served as a con-
onal designs, IEEE Trans. Inf. Theory 45 (5) (1999) 1456–1467. sultant for several major corporations and government
[20] Y. Acar, H. Doğan, E. Panayirci, Pilot symbol aided channel estimation for organizations. His research interests are in the areas of
spatial modulation-ofdm systems and its performance analysis with different digital signal processing, wireless standards, radio archi-
types of interpolations, Wirel. Pers. Commun. 94 (3) (2017) 1387–1404. tectures, and modern wireless systems for commercial
and military applications. He has published over 100 publications in these areas.

Вам также может понравиться