Modelling of DVB-T2 system using Consistent Channel Frequency MATLAB

Buy This Project

Hello friends, today I am going to post a complete project designed on MATLAB named as Modelling of DVB-T2 system using Consistent Channel Frequency in MATLAB. This project is designed by our team and it involved a lot of effort to bring it into existence that's why its not free but as usual I have discussed all the details below related to it, which will help you understanding it and if you want to buy it then you can click on the Buy button shown above.

This project aims to implement a DVB-T2 (Digital Video Broadcasting for terrestrial television) system using consistent channel frequency responses. Tthe code is designed to use the same output from a channel model for different transmitter configurations so that consistency of performance results can be obtained. After that the overall project will be modified to repeat an experiment “n” times collecting data so that “x%” confidence intervals can be calculated. Historically, DVB is a project worked by more than 250 companies around Europe at first and now worldwide. DVB-T2 is the world’s most advanced digital terrestrial television (DTT) system, offering more robustness, flexibility and at least 50% more efficiency than any other DTT system. It supports SD, HD, mobile TV, or any combination thereof. The GUI for DVB-T2 parameters selection in MATLAB is shown on the left.

MATLAB Simulations & Results

DVB-T2 is the second generation standard technology used for digital terrestrial TV broadcasting. As it’s a new technology so it has many fields to explore and research, and the best way of researching on any new technology is via simulations. Simulations provide an easy and efficient way to evaluate the performance of any system. For simulation purposes, MATLAB software was chosen in this thesis because of its wide range of tools and ability to show graphical results in a very appropriate form. . Further, this DVB-T2 simulation model could be extended easily to simulate DVB-H, which shares many features with DVB-T2 (only the physical layer that needs modification). The most important feature, I discussed in my simulations are:

  • Comparison of Bit Error Rate (BER) and Signal to noise Ratio (SNR).

DVB-T2 scheme can handle wide range of sub carriers from a range of 1k to 32k; these sub carriers can be fixed or mobile. In this thesis, experiments are performed on mobile transmission of signals to 4000 sub carriers. Below are discussed three different mobile scenarios, for different speeds of mobiles user, which are:

  • Pedestrian moving at the speed of 0.1 km/h.
  • Bus moving at the speed of 5km/h.
  • Car moving at the speed of 10 km/h.

In all the scenarios, the factors mentioned below are kept constant so that a real comparison can be obtained and it could be checked that whether the speed affects the signal or not. These constant factors are:

  • Transmission Mode is SISO.
  • Transmitting Antenna Cross Correlation is 0.5.
  • Receiving Antenna Cross Correlation is 0.5.
  • Environment considered for these simulations is rural.
  • Radio Channel type is Rayleigh with k factor of 1000.
  • Carrier frequency is 91.429MHz.

1.1  Initial MATLAB model

During this thesis, help was taken from a MATLAB model of DVB-T2 transmission system designed by a student at Brunel University. First this initial model was studied and then enhanced it to a higher level. The first model designed by the student at Brunel University, performed the iterations on the DVB-T2 system and gives the results for just one cycle. Explanation of this initial model is discussed in detail below.

1.1.1    Explanation of Initial Model

After the user input all the values in the GUI, this model first calculates the below three values depending on the number of subcarriers attached to the DVB-T2 system.

  • Useful OFDM period.
  • Maximum number of sub carriers.
  • IFFT / FFT length.

After getting this information, the model performs the QAM modulation over the signal so that it could be sent from the transmitter to the receiver. Next, depending on the value of Pilot Pattern given by the user, it calculates the scattered Pilot Amplitudes for the system. After that, it calculates the distortion in transmission depending on area in which the signal is propagating.

In order to calculate the distortion, FFT technique is performed on the signals to get their frequency response. As the signal has already sent from the transmitter after QAM modulation so demodulation on the receiver side is necessary. The model performs the same and demodulates the signal and finally it calculates the value of Signal to noise ratio (SNR) and Bit Error Rate (BER). At the end, it simply plots the graphs of SNR and BER for the visual representation.

1.1.2    Results of Initial Model

Different experiments were performed on the initial model and checked its results. The results are given below for three different experiments, which are:

  • Speed of mobile = 0.1 km/s , Total Iterations = 5000
  • Speed of mobile = 1.0 km/s , Total Iterations = 5000
  • Speed of mobile = 10 km/s , Total Iterations = 5000

Results of these experiments are shown in figure 6.1, 6.2 and 6.3 respectively. Table 6.1, 6.2 and 6.3 gives the values of BER and average BER for all the values of SNR. If these three graphs are closely examined then it can be shown that the band limited impulse response increases as the speed increase and so as the BER and SNR.

The reason for such behavior is that because as the speed of the mobile increase, signal distortion also increases and it becomes difficult for the receiver to catch the signal, that’s the main reason that user travelling in high speed vehicle faces more distortion as compared to a pedestrian.

Results of First Experiment (Speed = 0.1km/s, Iterations = 5000)

 
SNR BER & Average BER
SNR: 0 BER:0.103833
SNR: 0 NoAvrg_BER:0.160358
SNR: 5 BER:0.014366
SNR: 5 NoAvrg_BER:0.033706
SNR: 10 BER:0.000206
SNR: 10 NoAvrg_BER:0.001528
SNR: 15 BER:0.000002
SNR: 15 NoAvrg_BER:0.000107
SNR: 20 BER:0.001543
SNR: 20 NoAvrg_BER:0.002319
SNR: 25 BER:0.000076
SNR: 25 NoAvrg_BER:0.000184
SNR: 30 BER:0.000000
SNR: 30 NoAvrg_BER:0.000164

BER & Average BER Vs. SNR for experiment 1

Results of Second Experiment (Speed = 1km/s, Iterations = 5000)

SNR BER & Average BER
SNR: 0 BER:0.140855
SNR: 0 NoAvrg_BER:0.195596
SNR:5 BER:0.046527
SNR:5 NoAvrg_BER:0.071364
SNR:10 BER:0.011363
SNR:10 NoAvrg_BER:0.019860
SNR:15 BER:0.003815
SNR:15 NoAvrg_BER:0.006448
SNR:20 BER:0.000604
SNR:20 NoAvrg_BER:0.001222
SNR:25 BER:0.000214
SNR:25 NoAvrg_BER:0.000404
SNR:30 BER:0.000233
SNR:30 NoAvrg_BER:0.000503
BER & Average BER vs. SNR for experiment 2

Results of Third Experiment (Speed = 10km/s, Iterations = 5000)

SNR BER & Average BER
SNR: 0 BER:0.128177
SNR: 0 NoAvrg_BER:0.182924
SNR:5 BER:0.056198
SNR:5 NoAvrg_BER:0.084254
SNR:10 BER:0.023229
SNR:10 NoAvrg_BER:0.035131
SNR:15 BER:0.006793
SNR:15 NoAvrg_BER:0.010362
SNR:20 BER:0.001748
SNR:20 NoAvrg_BER:0.002801
SNR:25 BER:0.000425
SNR:25 NoAvrg_BER:0.000691
SNR:30 BER:0.000354
SNR:30 NoAvrg_BER:0.000515

BER & Average BER vs. SNR for experiment 3

Although the results given by these simulations were quite accurate but they were not accurate enough to be trusted, as they were performing the process just for one period and getting the results on the basis of that.

Final MATLAB Model after Modifications

The initial MATLAB model is modified in this thesis, in order to use the same output from the channel model with different transmitter configurations to obtain more consistent results that can be compared with each other. Then theDVB-T2model will be modified so that it can be simulated using Matlab n times collecting data so that an x% confidence interval can be measured.

The results obtained after modifications were very consistent as they were performing the whole scenario for N times (defined by the user), this attribute lacks in the initial model as it was performing the complete task just for one cycle of time and any kind of distortion could fluctuate the results. While in modified model, the same process was performed by N times defined by the user and the results obtained are actually the average of all the cycles and hence providing a very consistent output, which couldn’t be distorted by any external factors.

Moreover, this new model further enhanced the initial model to calculate the Mean BER as it will give the overall performance of BER and average BER. Furthermore, calculates the standard BER on the basis of which global BER is also calculated.

As the simulation of DVB-T2 requires a lot of input parameters from the user, that’s why a GUI is also designed in MATLAB, which makes the working of this project user friendly. User can easily change the parameters of the system using that GUI. On startup, the GUI looks like as shown in figure 4.2:

GUI for DVB-T2 parameters selection

As mentioned above, taking all the other parameters constant, three experiments are performed for the mobile user moving at different speeds with different Iterations and no. of repeats, which are:

  • Pedestrian moving at the speed of 0.1 km/h, with no of iterations = 500 and N=2.
  • Bus moving at the speed of 5km/h, with no of iterations = 200 and N=2.
  • Car moving at the speed of 10 km/h, with no of iterations = 1000 and N=2.

1.2.1    Results of First Experiment of Modified Model:

Results of the first experiment are shown in the figure 6.5, 6.6 and 6.7 respectively. While the theoretical values of BER and average BER for the corresponding SNR are shown in table 6.4 and the Mean BER and std BER are shown in table 6.5.

  • Value of Global BER for the experiment comes out to be -2.4021.

Results for Experiment 1 (Speed=0.1km/s, Iterations=500, N=2)

Results for Experiment 1 (Speed=0.1km/s, Iterations=500, N=2)

Results for Experiment 1 (Speed=0.1km/s, Iterations=500, N=2)

For N=1 For N=2
SNR: 0 BER:0.131272 SNR: 0 BER:0.131542
SNR: 0 NoAvrg_BER:0.185916 SNR: 0 NoAvrg_BER:0.186218
SNR:1 BER:0.107086 SNR:1 BER:0.106672
SNR:1 NoAvrg_BER:0.157698 SNR:1 NoAvrg_BER:0.157319
SNR:2 BER:0.086805 SNR:2 BER:0.086459
SNR:2 NoAvrg_BER:0.129841 SNR:2 NoAvrg_BER:0.129562
SNR:3 BER:0.087178 SNR:3 BER:0.086924
SNR:3 NoAvrg_BER:0.128066 SNR:3 NoAvrg_BER:0.127755
SNR:4 BER:0.081465 SNR:4 BER:0.081709
SNR:4 NoAvrg_BER:0.116619 SNR:4 NoAvrg_BER:0.116581
SNR:5 BER:0.028071 SNR:5 BER:0.028074
SNR:5 NoAvrg_BER:0.051596 SNR:5 NoAvrg_BER:0.051751
SNR:6 BER:0.016450 SNR:6 BER:0.016439
SNR:6 NoAvrg_BER:0.030762 SNR:6 NoAvrg_BER:0.030725
SNR:7 BER:0.012705 SNR:7 BER:0.012607
SNR:7 NoAvrg_BER:0.022399 SNR:7 NoAvrg_BER:0.022108
SNR:8 BER:0.036446 SNR:8 BER:0.036612
SNR:8 NoAvrg_BER:0.052421 SNR:8 NoAvrg_BER:0.052642
SNR:9 BER:0.026200 SNR:9 BER:0.026378
SNR:9 NoAvrg_BER:0.039987 SNR:9 NoAvrg_BER:0.040434
SNR:10 BER:0.014162 SNR:10 BER:0.014155
SNR:10 NoAvrg_BER:0.023779 SNR:10 NoAvrg_BER:0.023805
SNR:11 BER:0.007526 SNR:11 BER:0.007539
SNR:11 NoAvrg_BER:0.013874 SNR:11 NoAvrg_BER:0.013838
SNR:12 BER:0.015524 SNR:12 BER:0.015382
SNR:12 NoAvrg_BER:0.023693 SNR:12 NoAvrg_BER:0.023602
SNR:13 BER:0.005303 SNR:13 BER:0.005448
SNR:13 NoAvrg_BER:0.008758 SNR:13 NoAvrg_BER:0.008764
SNR:14 BER:0.008712 SNR:14 BER:0.008823
SNR:14 NoAvrg_BER:0.014517 SNR:14 NoAvrg_BER:0.014421
SNR:15 BER:0.013224 SNR:15 BER:0.013144
SNR:15 NoAvrg_BER:0.019547 SNR:15 NoAvrg_BER:0.019305
SNR:16 BER:0.001919 SNR:16 BER:0.001890
SNR:16 NoAvrg_BER:0.003767 SNR:16 NoAvrg_BER:0.003703
SNR:17 BER:0.002873 SNR:17 BER:0.002907
SNR:17 NoAvrg_BER:0.004932 SNR:17 NoAvrg_BER:0.005001
SNR:18 BER:0.000610 SNR:18 BER:0.000641
SNR:18 NoAvrg_BER:0.001197 SNR:18 NoAvrg_BER:0.001243
SNR:19 BER:0.006294 SNR:19 BER:0.006231
SNR:19 NoAvrg_BER:0.009262 SNR:19 NoAvrg_BER:0.009209
SNR:20 BER:0.001799 SNR:20 BER:0.001749
SNR:20 NoAvrg_BER:0.003268 SNR:20 NoAvrg_BER:0.003248
SNR:21 BER:0.000966 SNR:21 BER:0.000998
SNR:21 NoAvrg_BER:0.001677 SNR:21 NoAvrg_BER:0.001636
SNR:22 BER:0.001733 SNR:22 BER:0.001778
SNR:22 NoAvrg_BER:0.002772 SNR:22 NoAvrg_BER:0.002883
SNR:23 BER:0.004920 SNR:23 BER:0.004914
SNR:23 NoAvrg_BER:0.007638 SNR:23 NoAvrg_BER:0.007743
SNR:24 BER:0.000089 SNR:24 BER:0.000098
SNR:24 NoAvrg_BER:0.000220 SNR:24 NoAvrg_BER:0.000234
SNR:25 BER:0.000001 SNR:25 BER:0.000001
SNR:25 NoAvrg_BER:0.000052 SNR:25 NoAvrg_BER:0.000052
SNR:26 BER:0.000408 SNR:26 BER:0.000393
SNR:26 NoAvrg_BER:0.000695 SNR:26 NoAvrg_BER:0.000646
SNR:27 BER:0.000583 SNR:27 BER:0.000600
SNR:27 NoAvrg_BER:0.001222 SNR:27 NoAvrg_BER:0.001242
SNR:28 BER:0.000352 SNR:28 BER:0.000381
SNR:28 NoAvrg_BER:0.000609 SNR:28 NoAvrg_BER:0.000625
SNR:29 BER:0.000107 SNR:29 BER:0.000124
SNR:29 NoAvrg_BER:0.000365 SNR:29 NoAvrg_BER:0.000384
SNR:30 BER:0.000367 SNR:30 BER:0.000351
SNR:30 NoAvrg_BER:0.000720 SNR:30 NoAvrg_BER:0.000695

SNR Vs. BER values for Experiment 1

Mean BER std BER
-0.8814 0.0006
-0.9711 0.0012
-1.0623 0.0012
-1.0602 0.0009
-1.0884 0.0009
-1.5517 0.0000
-1.7840 0.0002
-1.8977 0.0024
-1.4374 0.0014
-1.5802 0.0021
-1.8490 0.0001
-2.1231 0.0005
-1.8110 0.0028
-2.2696 0.0083
-2.0571 0.0039
-1.8799 0.0019
-2.7203 0.0046
-2.5391 0.0035
-3.2039 0.0151
-2.2033 0.0031
-2.7511 0.0086
-3.0078 0.0099
-2.7556 0.0079
-2.3083 0.0004
-4.0308 0.0285
-6.1938 0
-3.3978 0.0118
-3.2281 0.0086
-3.4365 0.0247
-3.9387 0.0460
-3.4455 0.0137

Mean BER & std BER values for Experiment 1

1.2.2    Results of Second Experiment of Modified Model:

Figure 6.8, 6.9 and 6.10 gives us the results for the experiment 2, when user is travelling at the speed of 1km/s and the no of iterations taken here are 500 with repeat cycle i.e. N=2. Table 6.6 and 6.7 gives us the values of SNR vs. BER and the Mean BER and standard BER respectively.
  • Global BER comes out for experiment 2 was –inf.

Results for Experiment 2 (Speed=1km/s, Iterations=500, N=2)

Results for Experiment 2 (Speed=1km/s, Iterations=500, N=2)

Results for Experiment 2 (Speed=1km/s, Iterations=500, N=2)

For N = 1 For N = 2
SNR: 0 BER:0.144963 SNR: 0 BER:0.144537
SNR: 0 NoAvrg_BER:0.200382 SNR: 0 NoAvrg_BER:0.200617
SNR:1 BER:0.103536 SNR:1 BER:0.103496
SNR:1 NoAvrg_BER:0.153318 SNR:1 NoAvrg_BER:0.153312
SNR:2 BER:0.081079 SNR:2 BER:0.081874
SNR:2 NoAvrg_BER:0.123080 SNR:2 NoAvrg_BER:0.123966
SNR:3 BER:0.056279 SNR:3 BER:0.056618
SNR:3 NoAvrg_BER:0.096223 SNR:3 NoAvrg_BER:0.096636
SNR:4 BER:0.070647 SNR:4 BER:0.070241
SNR:4 NoAvrg_BER:0.103436 SNR:4 NoAvrg_BER:0.103023
SNR:5 BER:0.063094 SNR:5 BER:0.063427
SNR:5 NoAvrg_BER:0.089725 SNR:5 NoAvrg_BER:0.090577
SNR:6 BER:0.020785 SNR:6 BER:0.021318
SNR:6 NoAvrg_BER:0.039970 SNR:6 NoAvrg_BER:0.040469
SNR:7 BER:0.024660 SNR:7 BER:0.024455
SNR:7 NoAvrg_BER:0.040979 SNR:7 NoAvrg_BER:0.041170
SNR:8 BER:0.032986 SNR:8 BER:0.032662
SNR:8 NoAvrg_BER:0.052140 SNR:8 NoAvrg_BER:0.052100
SNR:9 BER:0.023306 SNR:9 BER:0.022988
SNR:9 NoAvrg_BER:0.037168 SNR:9 NoAvrg_BER:0.037283
SNR:10 BER:0.009120 SNR:10 BER:0.008878
SNR:10 NoAvrg_BER:0.017749 SNR:10 NoAvrg_BER:0.017499
SNR:11 BER:0.023258 SNR:11 BER:0.023224
SNR:11 NoAvrg_BER:0.034964 SNR:11 NoAvrg_BER:0.034473
SNR:12 BER:0.023534 SNR:12 BER:0.023745
SNR:12 NoAvrg_BER:0.034579 SNR:12 NoAvrg_BER:0.034325
SNR:13 BER:0.000103 SNR:13 BER:0.000101
SNR:13 NoAvrg_BER:0.000588 SNR:13 NoAvrg_BER:0.000648
SNR:14 BER:0.000016 SNR:14 BER:0.000010
SNR:14 NoAvrg_BER:0.000196 SNR:14 NoAvrg_BER:0.000231
SNR:15 BER:0.000009 SNR:15 BER:0.000014
SNR:15 NoAvrg_BER:0.000209 SNR:15 NoAvrg_BER:0.000240
SNR:16 BER:0.001996 SNR:16 BER:0.002008
SNR:16 NoAvrg_BER:0.003367 SNR:16 NoAvrg_BER:0.003535
SNR:17 BER:0.002367 SNR:17 BER:0.002430
SNR:17 NoAvrg_BER:0.003467 SNR:17 NoAvrg_BER:0.003535
SNR:18 BER:0.000002 SNR:18 BER:0.000004
SNR:18 NoAvrg_BER:0.000010 SNR:18 NoAvrg_BER:0.000018
SNR:19 BER:0.001298 SNR:19 BER:0.001367
SNR:19 NoAvrg_BER:0.002071 SNR:19 NoAvrg_BER:0.002116
SNR:20 BER:0.009918 SNR:20 BER:0.009850
SNR:20 NoAvrg_BER:0.014701 SNR:20 NoAvrg_BER:0.014585
SNR:21 BER:0.000472 SNR:21 BER:0.000521
SNR:21 NoAvrg_BER:0.000769 SNR:21 NoAvrg_BER:0.000854
SNR:22 BER:0.001085 SNR:22 BER:0.001169
SNR:22 NoAvrg_BER:0.001855 SNR:22 NoAvrg_BER:0.001917
SNR:23 BER:0.001360 SNR:23 BER:0.001495
SNR:23 NoAvrg_BER:0.002240 SNR:23 NoAvrg_BER:0.002427
SNR:24 BER:0.000595 SNR:24 BER:0.000621
SNR:24 NoAvrg_BER:0.001258 SNR:24 NoAvrg_BER:0.001321
SNR:25 BER:0.000873 SNR:25 BER:0.000820
SNR:25 NoAvrg_BER:0.001457 SNR:25 NoAvrg_BER:0.001422
SNR:26 BER:0.000003 SNR:26 BER:0.000003
SNR:26 NoAvrg_BER:0.000199 SNR:26 NoAvrg_BER:0.000201
SNR:27 BER:0.000326 SNR:27 BER:0.000342
SNR:27 NoAvrg_BER:0.000637 SNR:27 NoAvrg_BER:0.000651
SNR:28 BER:0.000198 SNR:28 BER:0.000216
SNR:28 NoAvrg_BER:0.000270 SNR:28 NoAvrg_BER:0.000262
SNR:29 BER:0.000000 SNR:29 BER:0.000000
SNR:29 NoAvrg_BER:0.000000 SNR:29 NoAvrg_BER:0.000000
SNR:30 BER:0.000000 SNR:30 BER:0.000000
SNR:30 NoAvrg_BER:0.000071 SNR:30 NoAvrg_BER:0.000078

SNR Vs. BER values for Experiment 2

Mean BER Std BER
0.8394 0.0009
0.9850 0.0001
1.0890 0.0030
1.2483 0.0018
1.1522 0.0018
1.1989 0.0016
1.6767 0.0078
1.6098 0.0026
1.4838 0.0030
1.6355 0.0042
2.0458 0.0083
1.6337 0.0005
1.6264 0.0027
3.9915 0.0072
4.8895 0.1323
4.9486 0.1512
2.6985 0.0018
2.6201 0.0081
5.5089 0.1569
2.8755 0.0159
2.0051 0.0021
3.3048 0.0302
2.9484 0.0231
2.8460 0.0291
3.2160 0.0133
3.0727 0.0192
5.4949 0
3.4765 0.0154
3.6850 0.0273
Inf NaN
Inf NaN

Mean BER and std BER values for Experiment 2

Remarks

This thesis presents the design and Implementation of DVB-T2 system in MATLAB software. The basic purpose of this thesis is to check the bit error ratio (BER) and signal to noise ratio (SNR) for DVB-T2 system so that the system could be improved to a better quality. DVB-T2 system is evaluated for mobile users moving at different speeds. It is clearly shown that the mobility has an impact on the received signal, where the SNR goes to zero in some points. This behavior will generate high BER. If the figures for impulse responses are checked for all the three experiments then it is depicted that the Impulse is high for the third experiment where the mobility speed is higher than the first two experiments. The packet data loss is almost zero for the first experiment while it’s increasing in the second and is higher in the third. The number of packet lost confirms this behavior that high losses occurred in the case of high mobility.

Syed Zain Nasir

I am Syed Zain Nasir, the founder of <a href=https://www.TheEngineeringProjects.com/>The Engineering Projects</a> (TEP). I am a programmer since 2009 before that I just search things, make small projects and now I am sharing my knowledge through this platform.I also work as a freelancer and did many projects related to programming and electrical circuitry. <a href=https://plus.google.com/+SyedZainNasir/>My Google Profile+</a>

Share
Published by
Syed Zain Nasir