Compressive Sensing Based Adaptive Channel Estimation for TDS-OFDM System using Sparsity Level of Channel

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 04 Issue: 08 | Aug -2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 5.181 | ISO 9001:2008 Certified Journal | Page 1880
Compressive Sensing based adaptive channel estimation for TDS-OFDM
system using sparsity level of channel
T.S.BRINDA1, K.DIVYA2
1,2 Assistant Professor, Dept Of Ece,Prince Dr.K.Vasudevan College Of Enginerring And Technology,
Tamil Nadu, India
----------------------------------------------------------------------------***--------------------------------------------------------------------------
Abstract - In this paper, Time Domain Synchronous
Orthogonal Frequency Division Multiplexing (TDS-OFDM)
transmission system is proposed as an alternative
approach to the traditional CP-OFDM technology as it
provides significant improvement in spectral efficiency
since no additional pilots are required and it achieves
faster synchronization. This paper introduces the
compressive sensing (CS) method which depends on
sparsity level of the channel. By using CS based TDS-
OFDM it can support higher order modulation and avoid
performance loss over fast fading channels. Initially, the
sparsity estimation is utilized to detect actual sparsity level
of wireless channel, which is detected by using the
restricted isometric principle. If the channel is sparse
enough then the priori aided subspace pursuit algorithm is
used which should meet the CS model else the improved
iterative method is used. Finally the accurate channel is
estimated. Simulation results determine that the proposed
system achieves better MSE performance and robustness
compared with traditional CS based methods.
Key Words: Time domain synchronous OFDM (TDS-
OFDM), Compressive sensing(CS),sparsity, subspace
pursuit (SP)
1. INTRODUCTION
Orthogonal frequency division multiplexing (OFDM) is
mainly categorized into three types of transmission
schemes[1] : cyclic prefix OFDM (CP-OFDM), zero
padding OFDM (ZP-OFDM), and time domain
synchronous OFDM (TDS-OFDM). In traditional OFDM, a
cyclic prefix (CP) is inserted as the guard interval
between nearby OFDM symbols. CPOFDM and ZP-OFDM
require many numbers of pilots, but this can be saved in
TDS-OFDM. Thus, TDS-OFDM has higher spectral efficiency
than CP-OFDM and ZP-OFDM. Also this TDS-OFDM has fast
and reliable synchronization compared to the others. This
can also used in digital television terrestrial broadcasting
(DTTB) standard called the digital terrestrial multimedia
broadcasting (DTMB). Digital terrestrial television
broadcasting (DTTB) currently attracts a significant
amount of attention in television services, as it provides for
fasterand more reliable transmission than conventional
analog television services. Digital Television Terrestrial
Multimedia Broadcasting (DTMB) is the TV standard for
mobile and fixed terminals used in china.
Most wireless communications and wireless broadcasting
implementations use CPOFDM, which can be considered as
the “classic” version of OFDM with the lowest complexity.
In recent years TDS-OFDM has been introduced, especially
for TV broadcasting. DTMB can implement in both single
and multicarrier modulation with a distinct structure
called time domain synchronous OFDM. This standard is
based on time-domain synchronous OFDM (TDS-
OFDM),where the level of the M-QAM modulation can
either be 4, 16, 32, or 64.DTMB claim this standard that it
has faster and more accurate channel tracking than the
DVB-T standard does due to its use of TDSOFDM.
Compressive sensing (CS) is based on the sparsity level of
the channel in which priori information is exceedingly
important which is directly related to the estimation
accuracy. The major aim is to reconstruct the signal. In this
paper we find out the sparsity level of the channel which is
not always satisfied with CS models [3]. To combine the
theoretical CS model and the complex practical channel, we
obtain the sparsity level of the channel into account and
propose the performance analysis of sparsity level of
channel based CS methods [2]. Firstly, a sparsity detection
method is proposed to detect the real sparsity level of the
channel. Then, we check whether the sparsity level meets
the theoretical model, if it meets the CS method we
propose the priori aided subspace pursuit (PA-SP)
algorithm to reconstruct the sparse channel. Besides, by
using the PA-SP algorithm for channel estimation it could
acquire lower complexity and higher accuracy than other
methods. Otherwise, when the channel is not sparse, we
adopt improved iterative algorithm in which the sparsity
level is used to identify the significant taps in the
reconstructed channel. Finally, the proposed channel
estimation scheme has improved accuracy and robustness
compared with traditional CS based methods.
2. TDS-OFDM
OFDM is a very popular and very important modulation
technology that is widely used in wireless communication
and broadcasting systems. A cyclic prefix is inserted at the
beginning of each OFDM symbol before transmission and
removed before demodulation. The length of the cyclic
prefix is greater than or equal to that of the channel
impulse response to eliminate the inter-symbol
interference. The three type of OFDM are shown in Fig .1

Fig-1. (a) CP-OFDM (b) ZP-OFDM (c) TDS-OFDM
The TDS-OFDM technique can attain fast signal capture
and robust synchronization.The special feature of this
technique is that a PN sequence with good
autocorrelation
property inserted between consecutive information
symbols. A TDS-OFDM system model that consist of PN
sequence and an OFDM data block that is shown in Fig.1c
Fig-2. TDS-OFDM compressive sensing
Improvements at the demodulation stage were recently
obtained with an OFDM variant, which is based on a
newly proposed time-division synchronization OFDM
(TDSOFDM)scheme, in which the PN sequence is used.
The PN sequence can furthermore be used as a training
sequence (TS) for synchronization and channel
estimation. The major benefit of this scheme is removal of
frequency domains pilots. Therefore TDSOFDM has been
choosing as the key technology of the international digital
television terrestrial broadcasting (DTTB) standard. TDS-
OFDM contain interference from earlier OFDM block,
which exists in IBI-free region. The new compressive
sensing (CS) theory, which is fundamentally different
from the classical Shannon-Nyquist sampling theorem,
has the ability to solve the problems of TDS-OFDM. After
the modulation and interleaving data converted to time
domain, then the PN sequence is added. At the receiver
section channel estimation is done and all the reverse
process that are done in the transmitter are done, that is
removal of PN sequence, FFT, and de-interleaving, is also
done here.
3. COMPRESSIVE SENSING
Compressive sensing (CS) has been introduced to
efficiently reconstruct sparse signals. The methodology of
CS has been applied in coding, information theory,
statistical signal processing. Based on this methodology,
CS can be employed to take advantage of the inherent
sparsity of wireless channels. This is significant because
fewer pilot tones to estimate the channel impulse
response results in more subcarriers for data
transmission CS based TDS- OFDM By using CS based
method, it can support higher order modulation. In this
paper, the compressive sensing (CS) theory is adopted to
solve those problems of TDS-OFDM, but it heavily relies
on the sparsity level of the channel. The impulse response
can be modeled as,
= ∑ [n- ], 0 K-1 (1)
is the nth entry of the CIR vector.Let A be the path delay
set and B be the path gain set and they can be modeled as,
A= {τ0, τ1,...., τK-1}, (2)
B= {β0, β1,...., βK-1} (3)
The PN sequence length is designed to K-1 combat the
worst case of fading channels in TDS-OFDM system. This
indicates that even though the received PN sequence is
contaminated by the previous OFDM data block, there
exists an IBI-free region r = [r0,r1,...,rG-1]- of small size
G = M - L + 1 at the receiver end of PN sequence[3]:
r = Φh+n (4)
where n=[w0 , w1,...., wG-1]
T
denotes the additive white
Gaussian noise (AWGN), and
(5)
where Φ denotes the Toeplitz matrix of size G*Ldetermined
by the PN sequences.
The restricted isometry principle less (RIPless) condition
for stable recovery [4], should satisfy the sparsity level of
channel (K)
K≤G/4∆ (6)
where ∆ indicates the parameter C0(log(L/G)+ 1) with C0
as an constant value, which is equal to 1. If we want to

adopt the CS algorithm, the channel of sparsity level
should satisfy the RIPless condition.
4.METHOD FOR ESTIMATING THE CHANNEL
From the Fig. 3, sparsity detection is generally used to
roughly estimate the channel and detect the noise.
Initially we make use of a sparsity detection to identify
the actual channel sparsity level. Then we check whether
the sparsity level meets the CS model or not, i.e., K ≤
G/(4∆). When the wireless channel is sparse enough, the
proposed PA-SP algorithm is used. Otherwise, when the
channel is not sparse, we adopt improved iterative
algorithm. Finally, the accurate channel is estimated.
Fig-3. Structure for adaptive channel
estimation
4.1 PA-SP Based Channel Estimation
In contrast with the traditional SP algorithm [6], we can
find that they follow similar procedure, but they are
distinct in the features described below:
1) Initialization: We estimate the initial subspace as the
sparsity detection result ̂ instead of the largest K
magnitude entries in the vector Φ*z.Thus the initialization
not only estimates the real subspace well, but also doesn’t
require any computation.
2)Halting condition: The delay set ̂ is used to quit the
iteration instead of residual zr, which keep unchanged
when the algorithm acquires the highest correlation
subspace.
3) Number of iteration: Compared to previous iteration
number log(K), the proposed scheme only requires K0
iterations, so the complexity is reduced.
The channel estimation for the proposed PA-SP algorithm
has lower complexity and high accuracy, since the channel
sparsity level K and the initial subspace D are known
in the prior.
4.2 Improved Iterative Channel Estimation
The improved iterative algorithm is utilized when the
sparsity level does not satisfy the CS model. The main
concept of this method is that we choose the
strongest K values in the channel estimation result
regarded as real taps and weaken the rest as noise to
enhance the estimation accuracy.At lth iteration, we can
obtain the channel estimate result as
̂(l)=h+n(l) (7)
where h is the actual channel and n(l) is the noise in lth
iteration. The accuracy of this method is better when
compared to the traditional counterparts.
5. RESULTS AND DISCUSSION
The results obtained by using the proposed algorithm
is for reconstructing the signal using compressive
sensing method.
A.PA-SP ALGORITHM
Fig-4.PA-SP Algorithm
The PA-SP algorithm is used when the channel
sparsity is met with the restricted isometry property
condition. Here the RIP condition is met. From the
fig.4, if RIP exceeds the sparsity level then the improved
iterative algorithm is used.If RIP condition is met, the
channel strength is low so the PA-SP algorithm is
implemented.
B.MSE PERFORMANCE COMPARISON
The MSE performance is analyzed for the CP-OFDM and
the TDS-OFDM system. When RIP condition is met with CS
method, then the TDS-OFDM is 1dB better than the CP-
OFDM. When the RIP condition is not met with the CS

method, then the TDS-OFDM is 2dB better than CP-
OFDM.MSE performance is analyzed after the RIP
condition is met.
Fig-5.MSE performance Comparison
C.PN SEQUENCE WITH OVERLAPPING DATA
Fig- 6. PN sequence with overlapping data
D.IMPROVED SIGNAL
The improved iterative algorithm is used when the
sparsity level does not meet the CS method. This
method reconstructs the PN sequence through
overlapping and non-overlapping data and performs
channel estimation in iterative manner. Based on the PN
sequence and RIP level, the improved signal is estimated.
Fig-7. Improved signal
6. CONCLUSION
In this paper, we have aimed to reconstruct the weaker
signal by using CS method, which rely on channel sparsity.
Besides, by using the PA-SP algorithm for channel
estimation it could acquire lower complexity and higher
accuracy than other methods. The results of simulation
show that the proposed scheme achieves improved MSE
performance than the traditional CS based methods. For
future work, we plan to adapt the optimal maximum
likelihood algorithm and show the capacity maximization
by using water filling algorithm
7. REFERENCES
[1] L. Dai, Z. Wang, and Z. Yang, “Compressive sensing
based time domain synchronous OFDM transmission for
vehicular communications,” IEEE J. Sel. Areas Commun.,
vol. 31, no. 9, pp. 460–469, Sep. 2013.
[2] W. Ding, F. Yang, C. Pan, L. Dai, and J. Song,
“Compressive sensing based channel estimation for OFDM
systems under long delay channels”, IEEE Trans.
Broadcast., vol. 60, no. 2, pp. 313–321, Jun. 2014.
[3] R. G. Baraniuk, “Compressive sensing [lecture notes],”
IEEE Signal Process. Mag., vol. 24, no. 4, pp. 118–121, Jul.
2007.
[4] E. J. Candes and Y. Plan, “A probabilistic and RIPless
theory of compressed sensing,” IEEE Trans. Inf. Theory,
vol. 57, no. 11, pp. 7235–7254, Nov. 2011.
[5] W. Ding, F. Yang, C. Pan, L. Dai, and J. Song,
“Compressive sensing based channel estimation for OFDM
systems under long delay channels,” IEEE Trans.
Broadcast., vol. 60, no. 2, pp. 313–321, Jun. 2014.
[6] W. Dai and O. Milenkovic, “Subspace pursuit for
compressive sensing signal reconstruction,” IEEE Trans.
Inf. Theory, vol. 55, no. 5, pp. 2230–2249, May 2009.

Compressive Sensing Based Adaptive Channel Estimation for TDS-OFDM System using Sparsity Level of Channel

More Related Content

What's hot (20)

Similar to Compressive Sensing Based Adaptive Channel Estimation for TDS-OFDM System using Sparsity Level of Channel (20)

More from IRJET Journal (20)

Recently uploaded (20)

Compressive Sensing Based Adaptive Channel Estimation for TDS-OFDM System using Sparsity Level of Channel