Category Archives: LTE

A 4G technology that is now being replaced by 5G worldwide.

Massive MIMO Fundamentals and Code

Background

Just like different frequency bands and time slots can be used to multiplex users, spatial domain can also be exploited to achieve the same result. It is well known that if there are 4 transmit antennas and 4 receive antennas then four simultaneous data streams can be transmitted over the air. This can be scaled up to 8 x 8 or in the extreme case to 128 x 128. When the number of transmit or receive antennas is greater than 100 we typically call it a Massive MIMO scenario and we need specialized signal processing techniques to handle this case. Computationally complex techniques like Maximum Likelihood (ML) become quite difficult to implement in real-time and we have to resort to some simplified linear array processing techniques. This is the topic of this blog post.

Description of the Scenario

To understand the scenario that we will discuss here, please look at our previous post. Also note that when we talk about nT x nR case we do not necessarily mean nT transmit antennas and nR receive antennas, it could also mean nT users with 1 antenna each and co-located nR receive antennas, such as at a base station. We typically assume that the number of receive antennas is greater than the number of users. When the number of users is greater than the number of receive antennas we call it overloaded case and this is not discussed here. Here the number of users is fixed at 16 (randomly distributed between 0 and 360 degrees within the cell) and the number of receive antennas is varied from 20 to 100.

Plane Wave Impinging Upon a ULA
Plane Wave Impinging Upon a ULA from a Particular User/Direction
Beamforming Using a Uniform Linear Array
Beamforming Using a Uniform Linear Array
Linear Signal Processing Techniques for Massive MIMO

The four signal processing techniques that are applied at the receive array are:

  1. Matched Filtering (MF)
  2. Moore Penrose Pseudo-Inverse without controlling the threshold (PINV)
  3. Moore Penrose Pseudo-Inverse with a specified threshold (PINV with tol)
  4. Minimum Mean Squared Error (MMSE)
Simulation Results

There are some other techniques that we experimented with but are omitted here for the sake of brevity. The MATLAB code and simulation results showing bit error rate as a function of receive array size (nR) are given below. It is seen that simple Matched Filter works quite well when the receive array size is small but with increasing nR the performance improvement is not that great. Least Squares (LS) technique using Moore-Penrose Pseudo Inverse shows improved performance with increasing nR and this can be further improved by controlling the threshold (tol). We found that a threshold of 0.1 gave significantly improved results as compared to no threshold case. Lastly we implemented MMSE and found that it gave us the best results. It must be noted that we also implemented ML for a limited size of receive array and found that its BER performance was far superior than any other technique.

BER as a Function of Receive Array Size
BER of a MU-MIMO Receive Array at the Base Station (LOS)
MATLAB Code for Massive MIMO Scenario
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% MASSIVE MIMO BEAMFORMING
% COPYRIGHT RAYMAPS (C) 2018
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

clear all
close all

% SETTING THE PARAMETERS FOR THE SIMULATION
f=1e9;        %Carrier frequency
c=3e8;        %Speed of light
l=c/f;        %Wavelength
d=l/2;        %Rx array spacing
N=20;         %Receive array size
M=16;         %Transmit array size (users)
theta=2*pi*(rand(1,M));         %Angular separation of users
EbNo=10;                        %Energy per bit to noise PSD
sigma=1/sqrt(2*EbNo);           %Standard deviation of noise

n=1:N;                          %Rx array number
n=transpose(n);                 %Row vector to column vector

% RECEIVE SIGNAL MODEL (LINEAR)
s=2*(round(rand(M,1))-0.5);           %BPSK signal of length M
H=exp(-i*(n-1)*2*pi*d*cos(theta)/l);  %Channel matrix of size NxM
wn=sigma*(randn(N,1)+i*randn(N,1));   %AWGN noise of length N
x=H*s+wn;                             %Receive vector of length N

% LINEAR ARRAY PROCESSING - METHODS
% 1-MATCHED FILTER
y=H'*x;

% 2-PINV without tol
% y=pinv(H)*x;

% 3-PINV with tol
% y=pinv(H,0.1)*x;

% 4-Minimum Mean Square Error (MMMSE)
% y=(H'*H+(2*sigma^2)*eye([M,M]))^(-1)*(H')*x;

% DEMODULATION AND BER CALCULATION
s_est=sign(real(y));          %Demodulation
ber=sum(s!=s_est)/length(s);  %BER calculation

%Note: Please select the array processing technique
%you want to implement (1-MF, 2-LS1, 3-LS2, 4-MMSE)

Note:

  1. In the code above, N=nR and M=nT
  2. What is labelled here as Matched Filter (MF) is strictly speaking Maximal Ratio Combining (MRC)
  3. The case we discuss above is categorized as Multiuser MIMO (MU-MIMO) for the uplink
  4. MU-MIMO for the downlink is not that straight forward and will be the subject of some future post
  5. We have considered a deterministic channel model as opposed to a probabilistic channel model
  6. Probabilistic channel model can be easily implemented by assuming that channel coefficients are independent and identically distributed (IID) complex Gaussian random variables with mean zero and variance of 0.5 per dimension
  7. The initial results we have obtained using this probabilistic channel model are much better than the results shown above, but the question remains which is the more accurate representation of a real channel
Update: Simulation Using a Probabilistic Channel

Since most of the literature in Massive MIMO uses a probabilistic channel instead of a deterministic channel, we decided to investigate this further. To implement such a channel model we simply need to change one line of the MATLAB code shown above. Instead of defining H as:

H=exp(-i*(n-1)*2*pi*d*cos(theta)/l);

We define H as:

H=(1/sqrt(2))*randn(N,M)+i*(1/sqrt(2))*randn(N,M);

The results are shown below. It is seen that the BER performance is orders of magnitude better. We would next investigate the performance degradation if the channel coefficients are not independent and identically distributed (IID) but have some correlation. This is closely tied to inter-element separation of the antenna array.

BER as a Function of Receive Array Size
BER of a MU-MIMO Receive Array at the Base Station (NLOS)
Concluding Remarks

The fundamental question that needs to be asked is why the performance in the NLOS scenario (probabilistic) is better than LOS scenario (deterministic). This has to do with the Signal to Noise Ratio [1]. In the above we have assumed the Signal to Noise Ratio (SNR) for the two scenarios to be the same. But realistically speaking this is never the case. Although the NLOS case assumes a rich scattering environment providing a high multiplexing gain (dependent on the rank of the channel matrix H) its SNR would always be lower due to reflection, diffraction and scattering loss. So a fair comparison between the LOS and NLOS case is only possible if we downward adjust the SNR for the NLOS case. Simulation results have shown that the SNR for the NLOS case needs to be downgraded by about 25 dB to have similar BER performance as the LOS case. Lastly it must be noted that the BER performance of the NLOS case would deteriorate once the channel coefficients are not IID and there is some correlation between them.

[1] Zimu Cheng, Binghao Chen, and Zhangdui Zhong “A Tradeoff between Rich Multipath and High Receive Power in MIMO Capacity”, International Journal of Antennas and Propagation, Volume 2013.

Fundamentals of Linear Array Processing – Receive Beamforming

In the previous two posts we discussed the fundamentals of array processing particularly the concept of beamforming (please check out array processing Part-1 and Part-2). Now we build upon these concepts to introduce some linear estimation techniques that are used in array processing. These are particularly suited to a situation where multiple users are spatially distributed in a cell and they need to be separated based upon their angles of arrival. But first let us introduce the linear model; I am sure you have seen this before.

x=Hs+w

Here, s is the vector of symbols transmitted by M users, H is the N x M channel matrix, w is the noise vector of length N and x is the observation vector of length N. The channel matrix formed by the channel coefficients is deterministic (as opposed to probabilistic) in nature as it is purely dependent upon the phase shifts that the channel introduces due to varying path lengths between the transmit and receive antennas. The impact of a channel coefficient can be thought of as a rotation of the complex signal without altering its amplitude.

This means that the channel acts like a single tap filter and the process of convolution is reduced to simple multiplication (a reasonable assumption if the symbol length is much larger than the channel delay spread). The channel model does not accommodate for path loss and fading that are also inherent characteristics of the channel. But the techniques are general enough for these effects to be factored in later. Furthermore, it is assumed that the channel H is known at the receiver. This is a realistic assumption if the channel is slowly varying and can be estimated by sending pilot signals.

Beamforming Using a Uniform Linear Array
Beamforming Using a Uniform Linear Array

So going back to the linear model we see that we know x and H while s and w are unknown. Here w cannot be estimated since it’s random in nature (remember what the term AWGN stands for?) and we ignore it for the moment. The structure of s is known. For example if we are using BPSK modulation then the m symbols of the signal vector s can either be +1 or -1. So we can start the process of symbol detection by substituting all possible combinations of s1, s2…sm and determine the combination that minimizes

||x-Hs||

This is called the Maximum Likelihood (ML) solution as it determines the combination that was most likely to have been transmitted based upon the observation.

Although ML is conceptually very appealing and yields good results it becomes prohibitively complex as the constellation size or number of transmit antennas increases. For example for 2-Transmit case and BPSK modulation there are 2^(1 bit x 2 antennas)=2^2=4 combinations, which seems quite simplistic. But if 16-QAM modulation is used and there are 4-Transmit antennas the number of combinations increases to 2^(4 bits x 4 antennas)=2^16=65536. So we conclude that ML is not the solution we are looking for if computational complexity is an issue (which might become less of an issue as the processing power of devices increases).

Next we turn our attention to a technique popularly known as Zero Forcing or ZF (the origins of the name I still do not know). According to this technique the channel has a multiplicative effect on the signal. So to remove this effect we simply divide the signal by the channel or in the language of matrices we perform matrix inversion. Mathematically we have:

x=Hs+w

H-1x=H-1Hs+H-1w

H-1x=s+H-1w=s̄

So we see that we get back the signal s but we also get a noise component enhanced by inverse of the channel matrix. This is the well-known problem of ZF called Noise Enhancement. Then there are other problems such as non-existence of the inverse when the channel H is not a square matrix (which only happens when the number of transmit and receive antennas is the same). The inverse of H also cannot be calculated if H is not full rank or determinant of H is zero.  So we now introduce another technique called Least Squares (LS). According to this the signal vector can be estimated as

s̄=(HHH)-1HHx

This is also sometimes referred to as the Minimum Variance Unbiased Estimator, as described by Steven M. Kay in his classical book on Estimation Theory [Fundamentals of Statistical Signal Processing Vol-1]. This can be easily implemented in MATLAB using Moore Penrose Pseudo Inverse or pinv(H). This is much more stable than going for the direct inversion methods.

We next plot the Bit Error Rate (BER) using the code below. The number of receive antennas is varied from two to ten while the number of transmit antennas is fixed at four. The transmit antennas are assumed to be positioned at 30, 40, 50 and 60 degrees from the axis of the receive array. The receive antennas are separated by λ/2 meters. The frequency of operation is 1GHz but it is quite irrelevant to the scenario considered as everything is measured in multiples of wavelengths. The Eb/No ratio (roughly the signal to noise ratio) is varied from 5dB to 20dB in steps of 5dB.

Bit Error Rate for Changing Rx Array Length
Bit Error Rate for Changing Rx Array Length

As expected the BER for the two methods, other than ML, is more or less the same and decreases rapidly once the number of receive antennas becomes greater than number of transmit antennas (or number of signals).  The case where the number of receive antennas is less than number of signals (equal powered and with a small angular separation) is dealt with by Overloaded Array Processing (OLAP) techniques and have been discussed in detail by James Hicks [Doctoral Dissertation] a student of Dr. Reed at Virginia Tech.

Strangely enough it is seen that the overloaded case is not the worst part of the BER curve. The worst BER is observed when the number the number of transmit and receive antennas is the same (four in this case). In other words the BER gradually increases as the rank of the channel matrix increases and then decreases once it reaches its maximum value. This is quite interesting and obviously has to do with Noise Enhancement that we discussed earlier. This will be further investigated in future posts.

For further information on the above methods visit this interesting article.

Update-1

So we struggled for a while to find out why the BER is worst at full rank and thought that there is something wrong in our model but ultimately we found that this has to do with how the pseudoinverse works and the way the tolerance limit (tol in MATLAB) for the singular values is set. We have found quite interesting results while experimenting with various inversion methods and the results are pending publication. Will keep you updated about the progress.

Update-2

We experimented with the MATLAB function pinv by changing the tolerance parameter. Previously we had used the default tolerance that is built into the function pinv. The default tolerance (tol) is defined as:

tol = max (size (H)) * sigma_max (H) * eps

where sigma_max (H) is the maximal singular value of channel matrix H

and eps is the machine precision.

More precisely, eps is the relative spacing between any two adjacent numbers in the machine’s floating point system. This number is obviously system dependent. On machines that support IEEE floating point arithmetic, eps is approximately 2.2204e-16 for double precision and 1.1921e-07 for single precision.

So back to the subject we experimented with two values of tol; 1.0 and 0.1 while changing the signal to noise ratio. The number of transmit antennas (users) is fixed at 4 while number of receive antennas is varied from 2 to 8. For tol value of 1.0 it is seen that changing the value of EbNo does not change the results much up to 6 receive antennas but after that the BER results rapidly diverge. For tol value of 0.1 the results are quite unexpected. The BER drops with increasing number of antennas up to N=5 but then there is an unexpected increase in the BER for N=6. This needs to be further investigated.

BER for tolerance of 1.0
BER for tolerance of 0.1

MATLAB CODE USED TO GENERATE ABOVE PLOT

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% MULTIUSER DETECTION USING 
% A UNIFORM LINEAR ARRRAY
% AKA RECEIVE BEAMFORMING
% COPYRIGHT RAYMAPS (C) 2018
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

clear all
close all

% SETTING THE PARAMETERS FOR THE SIMULATION
f=1e9;        %Carrier frequency
c=3e8;        %Speed of light
l=c/f;        %Wavelength
d=l/2;        %Rx array spacing
N=10;         %Receive array length

theta=([30 40 50 60])*pi/180;   %Angular placement of Tx array (users)
EbNo=10;                        %Energy per bit to noise PSD
sigma=1/sqrt(2*EbNo);           %Standard deviation of noise

n=1:N;                          %Rx array vector
n=transpose(n);                 %Converting row to column
M=length(theta);                %Tx array length 

% RECEIVE SIGNAL MODEL (LINEAR)
s=2*(round(rand(M,1))-0.5);           %BPSK signal of length M
H=exp(-i*(n-1)*2*pi*d*cos(theta)/l);  %Channel matrix of size NxM
wn=sigma*(randn(N,1)+i*randn(N,1));   %AWGN noise of length N
x=H*s+wn;                             %Receive vector of length N

% PINV without tol
% y=pinv(H)*x;

% PINV with tol
y=pinv(H, 0.1)*x;

% DEMODULATION AND BER CALCULATION
s_est=sign(real(y));          %Demodulation
ber=sum(s!=s_est)/length(s);  %BER calculation

Basics of Beamforming in Wireless Communications

In the previous post we had discussed the fundamentals of a Uniform Linear Array (ULA). We had seen that as the number of array elements increases the Gain or Directivity of the array increases. We also discussed the Half Power Beam Width (HPBW) that can be approximated as 0.89×2/N radians. This is quite an accurate estimate provided that the number of array elements ‘N’ is sufficiently large.

Plane Wave Impinging Upon a ULA
Plane Wave Impinging Upon a ULA

But the max Gain is always in a direction perpendicular to the array. What if we want the array to have a high Gain in another direction such as 45 degrees. How can we achieve this? This has application in Radars where you want to search for a target by scanning over 360 degrees or in mobile communications where you want to send a signal to a particular user without causing interference to other users. One simple way is to physically rotate the antenna but that is not always a feasible solution.

Going back to the basics remember that the Electric field pattern depends upon the constructive and destructive interference of incoming waves. If we have a vector (usually called the steering vector) that aligns the rays coming in from a particular direction we would get a high Gain in that direction. Similarly we can steer a null in a particular direction if we want to reject a particular signal. This we will discuss in a future post.

MATLAB CODE

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% BEAMFORMING USING A
% UNIFORM LINEAR ARRRAY
% COPYRIGHT RAYMAPS (C) 2018
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

clear all
close all

f=1e9;
c=3e8;
l=c/f;
d=l/2;
no_elements=10;
phi=pi/6;

theta=0:pi/180:2*pi;
n=1:no_elements;
n=transpose(n);

X=exp(-i*(n-1)*2*pi*d*cos(theta)/l);
w=exp( i*(n-1)*2*pi*d*cos(phi)/l);
w=transpose(w);
r=w*X;

polar(theta,abs(r),'b')
title ('Gain of a Uniform Linear Array')

The figure below shows the Electric field pattern of a 10 element array steered towards 0, 30, 60 and 90 degrees respectively. We see that selectivity of the array is higher on the Broadside than on the Endfire. In my opinion this has to do with how the cosine function behaves from 0 to 90 degrees. The rate of change of cosine function is much faster around 90 degrees than at 0 degrees or 180 degrees. The slowly changing cosine in the latter case causes a wide response on the Endfire.

Beamforming Using a Ten Element Array

We did calculate the HPBW for a range of steering angles and found that it varied widely from as small as 10.17 degrees to as large as 48.62 degrees. This shows that simple Beamforming using a steering vector has its limitations. The detailed results along with a graph are shown below. It is seen that as the steering angle increases from about 20 degrees there is a sudden decrease in HPBW. For one degree increase of steering angle (phi) from 24 to 25 degrees there is decrease of approx 9 degrees in HPBW. We will investigate this further in future posts.

Case 1: phi = 0, HPBW = 48.62  deg
Case 2: phi = 30, HPBW = 21.69 deg
Case 3: phi = 60, HPBW = 11.75 deg
Case 4: phi = 90, HPBW = 10.17 deg

Half Power Beamwidth of a ULA
Half Power Beamwidth of a ULA as a function of Steering Angle (Phi)

For further visualization of the variation in antenna pattern as a function of the steering angle please have a look at this Interactive Graph. The parameters that can be varied include the angle of the beam, number of antenna elements and separation of the antenna elements. This is taken from an excellent online resource by the name of Geogebra. For further information on how you can use this tool for your own mathematical problems please do visit their website.

MY FIRST GEOGEBRA VISUALIZATION

Open Signal Coverage Maps for Pakistan

Open Signal is a mobile application that collects the data about your wireless network (2G/3G/4G) and generates coverage maps and host of other reports. The data is collected in the background while the user is busy in his daily routines. But data can also be collected on the request of the user. This is much better than drive testing since the data is collected in real life scenarios and on thousands of different devices that are in use.

The app works while the user is indoor or outdoor, at rest or in motion, on land or on water, at sea level or on a mountain, in dry weather or in rain. Basically anywhere and anytime there are wireless signals available. There are currently 20 million users of the app (both Android and iOS combined) and this number is increasing. In Pakistan all major networks are supported including Jazz, Telenor, Zong and Ufone (both 2G/3G and 4G networks are supported).

JAZZ Islamabad Coverage Map

Telenor Islamabad Coverage Map

Zong Islamabad Coverage Map

Ufone Islamabad Coverage Map

BER for BPSK-OFDM in Frequency Selective Channel

OFDM Tx-Rx Block Diagram

As the data rates supported by wireless networks continue to rise the bandwidth requirements also continue to increase (although spectral efficiency has also improved). Remember GSM technology which supported 125 channels of 200KHz each, which was further divided among eight users using TDMA. Move on to LTE where the channel bandwidth could be as high as 20MHz (1.4MHz, 3MHz, 5MHz, 10MHz, 15MHz and 20MHz are standardized).

This advancement poses a unique challenge referred to as frequency selective fading. This means that different parts of the signal spectrum would see a different channel (different amplitude and different phase offset). Look at this in the time domain where the larger bandwidth means shorter symbol period causing intersymbol interference (as time delayed copies of the signal overlap on arrival at the receiver).

The solution to this problem is OFDM that divides the wideband signal into smaller components each having a bandwidth of a few KHz. Each of these components experiences a flat channel. To make the task of equalization simple a cyclic prefix (CP) is added in the time domain to make the effect of fading channel appear as circular convolution. Thus simplifying the frequency domain equalization to a simple division operation.

Shown below is the Python code that calculates the bit error rate (BER) of BPSK-OFDM which is the same as simple BPSK in a Rayleigh flat fading channel. However there is a caveat. We have inserted a CP which means we are transmitting more energy than simple BPSK. To be exact we are transmitting 1.25 (160/128) times more energy. This means that if this excess energy is accounted for the performance of BPSK-OFDM would be 1dB (10*log10(1.25)) worse than simple BPSK in Rayleigh flat fading channel.

Note:

  1. Although we have shown the channel as a multiplicative effect in the figure above, this is only true for a single tap channel. For a multi-tap channel (such as the one used in the code above) the effect of the channel is that of a filter which performs convolution operation on the transmitted signal.
  2. We have used a baseband model in our simulation and the accompanying figure. In reality the transmitted signal is upconverted before transmission by the antennas.
  3.  The above model can be easily modified for any modulation scheme such as QPSK or 16-QAM. The main difference would be that the signal would have a both a real part and an imaginary part, much of the simulation would remain the same. This would be the subject of a future post. For a MATLAB implementation of 64-QAM OFDM see the following post (64-QAM OFDM).
  4. Serial to parallel and parallel to serial conversion shown in the above figure was not required as the simulation was done symbol by symbol (one OFDM symbol in the time domain represented 128 BPSK symbols in the frequency domain).
  5. The channel model in the above simulation is quasi-static i.e. it remains constant for one OFDM symbol but then rapidly changes for the next, without any memory.

Rayleigh Fading Envelope Generation – Python

When wireless signals travel from a transmitter to a receiver they do so after reflection, refraction, diffraction and scattering from the environment. Very rarely is there a direct line of sight (LOS) between the transmitter and receiver. Thus multiple time delayed copies of the signal reach the receiver that combine constructively and destructively. In a sense the channel acts as an FIR (finite impulse response) filter. Furthermore since the transmitter or receiver may be in motion the amplitude and phase of these replicas varies with time.

There are several methods to model the amplitude and phase of each of these components. We look at one method called the “Smiths Fading Simulator” which is based on Clark and Gans model. The simulator can be constructed using the following steps.

1. Define N the number of Gaussian RVs to be generated, fm the Doppler frequency in Hz, fs the sampling frequency in Hz, df the frequency spacing which is calculated as df=(2*fm)/(N-1) and M total number of samples in frequency domain which is calculated as M=(fs/df).
2. Generate two sequences of N/2 complex Gaussian random variables. These correspond to the frequency bins up to fm. Take the complex conjugate of these sequences to generate the N/2 complex Gaussian random variables for the negative frequency bins up to -fm.
3. Multiply the above complex Gaussian sequences g1 and g2 with square root of the Doppler Spectrum S generated from -fm to fm. Calculate the spectrum at -fm and +fm by using linear extrapolation.
4. Extend the above generated spectra from -fs/2 to +fs/2 by stuffing zeros from -fs/2 to -fm and fm to fs/2. Take the IFFT of the resulting spectra X and Y resulting in time domain signals x and y.
5. Add the absolute values of the resulting signals x and y in quadrature. Take the absolute value of this complex signal. This is the desired Rayleigh distributed envelope with the required temporal correlation.

The Matlab code for generating Rayleigh random sequence with a Doppler frequency of fm Hz is given below.