HYBRID PRECODING ALGORITHM FOR MASSIVE

MIMO SYSTEM WITH IMPERFECT CHANNEL

KNOWLEDGE
Bart Simpson and Homer Simpson Monkey King, Bajie Zhu and Seng Tang
Department of Electronics and Communication Engineering Department of Intelligent Robotics
University of Veltech University of Huaguoshan
Avadi , Chennai Huaguoshan, Jileshijie Province, China
{bart.simpson & homer.simpson}@uspringfield.edu [email protected]

Abstract - Millimeter wave MIMO systems provide orders spectral efficiency and approach capacity the fundamentals
of of pre-coding are the same regardless of carrier frequency,
Magnitude in data rates due to larger bandwidth than whereas there are some non-trivial practical constraints for
conventional MIMO systems. On the other hand, they signal processing in mm Wave MIMO system.
experience severe free-space path loss as carrier frequency ten-
fold increases. The resulting small wavelength helps to A. Pre-coding:
overcome path loss with beam forming gain using massive
antennas, known as pre-coding. In mm Wave MIMO system,
1) Pre-coding is a generalization of beam forming to
hybrid beam forming combines analog beam forming with
digital pre-coding to reduce the energy consumption and
support multi-stream or multi-layer transmission in
complexity. Hybrid beam forming uses a smaller number of RF multi-antenna wireless communications. In
chains, which have large power consumption, than the number conventional single-stream beam forming, the same
of antennas. In order to design analog and digital pre-coder, a signal is emitted from each of the transmit antennas
large number of calculations such as the SVD of channel is with appropriate weighting (phase and gain) such
required. In this paper, we propose low complexity hybrid pre- that the signal power is maximized at the receiver
coding algorithm using beam steering, which utilizes array output. When the receiver has multiple antennas,
response vectors of the channel. We first propose the method of single-stream beam forming cannot simultaneously
selection the analog beam forming matrix without any SVD or maximize the signal level at all of the receive
inversion of matrices. Then, we find the digital pre-coding antennas. In order to maximize the throughput in
matrix which performs a maximum of the total sum rate with
multiple receive antenna systems, multi-stream
fixed analog beam forming matrix. Since our proposed
transmission is generally required.
algorithm applies a set of array response vectors that are used
to form the channel, there is no need for complicated operations
as SVD like traditional pre-coding algorithms 2) Pre-coding for Point-Point MIMO system: In point-
to-point (MIMO) systems, a transmitter equipped
Index Terms – multiple input multiple output (MIMO), radio with multiple antennas communicates with a
frequency (RF) chains, singular valued decomposition (SVD), receiver that has multiple antennas. Most classic
millimeter wave (mm WAVE) pre-coding results assume narrowband, slowly
fading channels, meaning that the channel for a
I. INTRODUCTION
certain period of time can be described by a single
The fifth generation wireless communication system is channel matrix which does not change faster. In
required to achieve super high spectral efficiency with practice, such channels can be achieved, for
widespread use of smart phones, association with Internet of example, through OFDM. The pre-coding strategy
Things or vehicle communication. According to market that maximizes the throughput, called channel
demands for gigabit-per-second data rates in indoor wireless capacity, depends on the channel state information
systems, millimeter wave system is proposed for cellular available in the system.
networks. The integration of mm Wave and massive
multiple-input multiple-output technique can achieve much 3) Zero Forcing Pre-coding: Zero-forcing (or null-
higher system throughput due to larger bandwidth than steering) pre-coding is a method of spatial signal
conventional MIMO systems. Mm wave signals experience processing by which a multiple antenna transmitter
severe free-space path loss because of the increase of carrier can null the multiuser interference in a multi-user
frequency. On the other hand, the decrease in wavelength of MIMO wireless communication system. When the
this band enables packing a large number of antenna channel state information is perfectly known at the
elements, which can provide the beam forming gain to transmitter, then the zero-forcing pre-coder is given
overcome path loss. Further, massive antennas allow to pre- by the Moore-Penrose pseudo-inverse of the channel
coding multiple data streams which can improve system matrix. If the transmitter knows the downlink
 channel state information (CSI) perfectly, ZF-pre- Examining the patents, journal articles, and books available
coding can achieve almost the system capacity when on
the number of users is large. On the other hand, with OFDM, it is clear that this technique will have an impact on
limited channel state information at the transmitter Future communications. Some examples of current
(CSIT) the performance of ZF-pre-coding decreases applications using OFDM include GSTN (general Switched
depending on the accuracy of CSIT. ZF-pre-coding Telephone Network), Cellular radio, DSL & ADSL modems,
requires the significant feedback overhead with DAB (digital audio broadcasting) radio, DVB-T (terrestrial
respect to signal-to-noise-ratio (SNR) so as to Digital video broad casting), HDTV broadcasting,
achieve the full multiplexing gain. Inaccurate CSIT Hyper-LAN(high performance local area network standard),
results in the significant throughput loss because of and the wireless networking standard.
residual multiuser interferences. Multiuser Baseband
interferences remain since they cannot be null with Data
Modulation OFDM
beams generated by imperfect CSIT. (QPSK,QA IFFT D/A Signal
M, etc.)
4) Hybrid Pre-coding: Hybrid pre-coding scheme, in
where signal is pre-processed in both analog and
digital domains, is recently proposed to reduce the Baseband
hardware cost. An orthogonal beam forming Data OFDM
codebook for two-dimensional channel environment Modulation Signal
is proposed to eliminate the efforts of searching for (QPSK, FFT A/D
angle of departure. QAM, etc.)
out

B. Millimeter Wave Technology: millimetre wave massive

multiuser multiple-input multiple-output (MU-MIMO)
system usually utilizes hybrid pre-coding to reduce complex Fig.1: Block Diagram
hardware and high energy consumption. However, most
hybrid pre-coding scheme consider the fully connected Signals are orthogonal if they are mutually independent of
architecture, which may result in a significant loss of energy each other. Orthogonally is a property that allows multiple
efficiency. In this paper, we propose a hybrid pre-coding information signals to be transmitted perfectly over a
scheme based on the overlapped sub-array architecture common channel and detected, without interference. Loss of
(OSA), where the analogue part of the MU-MIMO system is Orthogonally results in blurring between these information
realized by a OSA network. Based on the OSA architecture, a signals and degradation in communications. Many common
modified generalized low rank approximation of matrices multiplexing schemes are inherently orthogonal. Time
(GLRAM) based hybrid pre-coding scheme is designed to Division Multiplexing (TDM) allows transmission of
maximize the achievable sum-rate, where the analogue pre- multiple information signals over a single channel by
coding algorithm apply the phase alignment technology and assigning unique time slots to each separate information
GLRAM approach to obtain RF pre-coder/combiner with signal. During each time slot only the signal from a single
high array gain provided by massive MIMO system. Then, source is transmitted preventing any interference between the
the block diagonalizable (BD) technology is performed to multiple information sources. Because of this TDM is
obtain the digital pre-coding matrix for the multiplexing orthogonal in nature. In the frequency domain most FDM
gain. Simulation results show that the modified GLRAM systems are orthogonal as each of the separate transmission
based hybrid pre-coding algorithm for OSA architecture can signals are well spaced out in frequency preventing
reduce the number of iterations. Compared with the existing interference.
hybrid pre-coding algorithm, it achieves higher spectral
efficiency and lower complexity. B. Convolution Coding
Convolution code is a type of error-correcting code that
II. CONCEPTS generates parity symbols via the sliding application of a boo-
A. OFDM lean polynomial function to a data stream. The sliding
With the rapid growth of digital communication in application represents the 'convolution' of the encoder over
recent years, The need for high-speed data transmission has the data, which gives rise to the term 'convolution coding'.
increased. New multicarrier modulation techniques such as The sliding nature of the convolution codes facilitates trellis
OFDM are currently being implemented to keep up with the decoding using a time-invariant trellis. Time invariant trellis
demand for more communication capacity. The processing decoding allows convolution codes to be maximum-
power of modern digital signal processors has increased to a likelihood soft-decision decoded with reasonable complexity.
point where OFDM has become feasible and economical. The code rate of a convolution code is commonly modified
via symbol puncturing. Convolution codes are often
described as continuous. However, it may also be said that Quadrature Amplitude Modulation, QAM utilize both
convolution codes have arbitrary block length, rather than amplitude and phase components to provide a form of
being continuous, since most real-world convolution modulation that is able to provide high levels of spectrum
encoding is performed on blocks of data. Convolution usage efficiency.
encoded block codes typically employ termination. The QAM, quadrature amplitude modulation has been used for
arbitrary block length of convolution codes can also be some analogue transmissions including AM stereo
contrasted to classic block codes, which generally have fixed transmissions, but it is for data applications where it has
block lengths that are determined by algebraic properties. In come into its own. It is able to provide a highly effective
telecommunication, a convolution code is a type of error- form of modulation for data and as such it is used in
correcting code that generates parity symbols via the sliding everything from cellular phones to Wi-Fi and almost every
application of a boo-lean polynomial function to a data other form of high speed data communications system QAM
stream. The sliding application represents the 'convolution' of is a signal in which two carriers shifted in phase by 90
the encoder over the data, which gives rise to the term degrees ( sine and cosine) are modulated and combined. As a
'convolution coding'. The sliding nature of the convolution result of their 90° phase difference they are in quadrature and
codes facilitates trellis decoding using a time-invariant trellis. this gives rise to the name. Often one signal is called the In-
Time invariant trellis decoding allows convolution codes to phase or “I” signal, and the other is the quadrature or “Q”
be maximum-likelihood soft-decision decoded with signal. The resultant overall signal consisting of the
reasonable complexity. combination of both I and Q carriers contains of both
amplitude and phase variations. In view of the fact that both
amplitude and phase variations are present it may also be
C. IFFT and FFT considered as a mixture of amplitude and phase modulation.
The concepts used in the simple analog OFDM
implementation can be extended to the digital domain by E. Pulse Shaping
using a combination of Fast Fourier Transform (FFT) and Transmitting a signal at high modulation rate through a
Inverse Fast Fourier Transform (IFFT) digital signal band-limited channel can create inter symbol interference. As
processing. These transforms are important from the OFDM the modulation rate increases, the signal's bandwidth
perspective because they can be viewed as mapping digitally increases. When the signal's bandwidth becomes larger than
modulated input data (data symbols) onto orthogonal the channel bandwidth, the channel starts to introduce
subcarriers. In principle, the IFFT takes frequency-domain distortion to the signal. This distortion usually manifests
input data (complex numbers representing the modulated itself as inter symbol interference. The signal's spectrum is
subcarriers) and converts it to the time-domain output data determined by the modulation scheme and data rate used by
(analog OFDM symbol waveform). the transmitter, but can be modified with a pulse shaping
The Fourier transform is used to convert the signals from filter. Usually the transmitted symbols are represented as a
time domain to frequency domain and the inverse Fourier time sequence of delta pulses. This theoretical signal is then
transform is used to convert the signal back from the filtered with the pulse shaping filter, producing the
frequency domain to the time domain. The Fourier transform transmitted signal. In many base band communication
is a powerful tool to analysis the signals and construct them systems the pulse shaping filter is implicitly a boxcar filter.
to and from their frequency components. If the signal is Its Fourier transform is of the form sin(x)/x, and has
discrete in time that is sampled, one uses the discrete Fourier significant signal power at frequencies higher than symbol
transform to convert them to the discrete frequency form rate. This is not a big problem when optical fiber or even
DFT, and vice verse, the inverse discrete transform IDFT is twisted pair cable is used as the communication channel.
used to back convert the discrete frequency form into the However, in RF communications this would waste
discrete time form. To reduce the mathematical operations bandwidth, and only tightly specified frequency bands are
used in the calculation of DFT and IDFT one uses the fast used for single transmissions.
Fourier transform algorithm FFT and IFFT which
corresponds to DFT and IDFT, respectively.
In transmitters using OFDM as a multicarrier modulation
technology, the OFDM symbol is constructed in the III. PROCEDURE
frequency domain by mapping the input bits on the I- and Q- A channel model is a mathematical representation of the
components of the QAM symbols and then ordering them in effects of a communication channel through which wireless
a sequence with specific length according to the number of signals are propagated. The channel model can represent the
subcarriers in the OFDM symbol. That is by the mapping and power loss incurred by the signal as it travels through the
ordering process one constructs the frequency components of wireless medium. In a more general case, the channel model
the OFDM symbol. To transmit them, the signal must be is the impulse response of the channel medium in the time
represented in time domain. This is accomplished by the domain or its Fourier transform in the frequency domain. The
inverse fast Fourier transform IFFT. channel impulse response of a wireless communication
D. QAM Modulation system typically varies randomly over time. By using
channel models with your wireless system design in mat-lab words, solving very ill-conditioned linear systems by
and simulation, you can optimize link performance, perform classical methods
system architecture tradeoffs, and provide a realistic is not usual. The Krylov subspace methods (without pre-
assessment of the overall system performance. conditioners) and also the iterative methods of the AOR
OFDM divides the used signal bandwidth B into narrow-band family, both require a remarkable time to find
sub channels whose bandwidth does not exceed the a reliable solution. Note that the Accelerated Over
coherency bandwidth of the mobile radio channel under Relaxation family or the AOR
consideration. Under this restriction, the transmitted signal is
merely distorted by a non-frequency-selective attenuation
and phase shift which corresponds mathematically to
V. RESULTS
multiplication by a complex number. If this complex
transmission coefficient is calculated for every OFDM sub
channel in the receiver – for example, based on known
reference or pilot signals – one-tap equalization can be
achieved by multiplying every sub channel with the inverse
of its transmission coefficient. However, the signal is still
affected by the unavoidable, additive overlay of thermal
noise and in band radio interference, making it difficult to
determine the transmission coefficients exactly and in fact
providing only a certain degree of accuracy. This is one of
the main reasons why the process of determining these
transmission coefficients is called channel estimation.

FIG.3:

Fig.2: Block Diagram

IV. HOTELLING BODEWIG ALGORITHM

The Hotelling Bodewig algorithm is simple to describe
and to analyze, and is numerically stable for nonsingular
input matrices. This was the idea of developing iterative
methods of this type for the solution of linear systems. In this FIG.4: 8*128
article, we have developed an iterative method in inverse-
finding of matrices. it used for minimum residual iterations It
is widely known that the solutions of linear systems of
equations are sensitive to the round-off error. For some linear CONCLUSION
systems a small change in one of the values of the coefficient
matrix or the right-hand side vector causes a large change in We have proposed Hybrid pre-coding algorithm for
the solution vector. When the solution is highly sensitive to Massive MIMO system with imperfect channel
the values of the coefficient matrix Aor the right-hand side knowledge .In this project we use a special algorithm called
constant vector b, the equations are said to be ill-conditioned. Hotelling Bodewig Algorithm. it is a matrix inversion
Therefore, we cannot easily rely on the solutions coming out algorithm which is more easy to implement and more
of an ill-conditioned system. Ill-conditioned systems pose efficient compare to other algorithms. We proposed work is
particular problems when the coefficients are estimated from low complexity Pre-coding for Multiple user data transfer.
experimental results. For a system with condition number on
its coefficient matrix, one can expect a loss of roughly
decimal places in the accuracy of the solution. In other
 REFERENCES
[1]X. Wang, M. Jia, Q. Guo, I. W.-H. Ho, and J. Wu, ``Joint power, original
bandwidth, and detected hole bandwidth allocation for multi-homing
heterogeneous networks based on cognitive radio,'' IEEE Trans. Veh-.
Technol.,vol. 68, no. 3, pp. 2777§2790, Mar. 2019.

[2]S. Mumtaz, J. Rodriguez, and L. Dai, Mm Wave Massive MIMO: A

Paradigm for 5G. New York, NY, USA: Academic, 2016.

[3]A. Alkhateeb, G. Leus, and R. W. Heath, ``Limited feedback hybrid pre-

coding for multi-user millimeter wave systems,'' IEEE Trans. Wireless
Communication., vol. 14, no. 11, pp. 6481§6494, Nov. 2015.

[4]S. Han, C.-I. I, Z. Xu, and C. Rowell, ``Large-scale antenna systems with
hybrid analog and digital beam forming for millimeter wave 5G,'' IEEE
Communication. Mag., vol. 53, no. 1, pp. 186§194, Jan. 2015.

[5]X. Gao, L. Dai, S. Han, I. Chih-Lin, and R. W. Heath, ``Energy-efficient

hybrid analog and digital pre-coding for Mm Wave MIMO systems with
large antenna arrays,'' IEEE J. Sel. Areas Communication., vol. 34, no. 4,
pp. 998§1009, Apr. 2016.

[6] L. Dai, B. Wang, M. Peng, and S. Chen, ``Hybrid precoding-based

Millimeter wave massive MIMO NOMA with simultaneous wireless
information and power transfer,'' IEEE J. Select. Areas Commun., vol. 37,
no. 1, pp. 131§141, Jan. 2019.

[7]A. Alkhateeb and R.W. Heath, Jr., ``Frequency selective hybrid pre-coding
for limited feedback millimeter wave systems,'' IEEE Trans. Communication.,
vol. 64, no. 5, pp. 1801§1818, May 2016.

[8] O. El Ayach, S. Rajagopal, S. Abu-Surra, Z. Pi, and R. W. Heath, Jr.,

``Spatially sparse precoding in millimeter wave MIMO systems,'' IEEE
Trans. Wireless Commun., vol. 13, no. 3, pp. 1499§1513, Mar. 2014.