Reference no: EM133196926
Question 1 For this question you cannot use the Matlab-ready code to produce AWGN and equally probable 16-QAM symbols, to produce the MIMO Channel, or to perform detection. Instead you can write your own script based on the rand and randn functions. When you use results from the literature do not forget to add the corresponding references).
We consider a 5G multiuser-MIMO system with 2 base-station antennas and 2 single-antenna users that concurrently transmit (in the uplink) independent and identically distributed transmitted 16-QAM symbols, with no channel coding, using spatial multiplexing.
1) After searching the corresponding literature,
a) Explain in less than 500 words (and provide references to the corresponding literature) all the steps required in order to perform minimum mean square error (MMSE), "hard" detection in order calculate what the transmitted symbol is.
b) Provide a Matlab script that simulates the average symbol-error-rate (SER) performance of MMSE detection, in logarithmic scale, as a function of the signal-to-noise ratio (SNR), in dB, when assuming an independent and identically distributed (i.i.d.) Rayleigh fading MIMO transmission channel. All the details of your code should be explained. Plot the simulation results for a range of SNRs such that the SER ranges from approximately 10-1 to approximately 10-4 (Hint: verify your results with the literature).
2) After searching the corresponding literature,
a) Explain in less than 500 words (and provide references to the corresponding literature) the steps required to perform, "hard" Maximum-Likelihood (ML) MIMO detection by exhaustive search (i.e., not with sphere decoding). Calculate, and explain, what is the complexity of such an approach in terms of complex multiplications for the above 2x2 multi-user MIMO system.
b) Provide a Matlab script that simulates the average symbol-error-rate (SER) performance for "hard" ML detection through exhaustive search as a function of the signal-to-noise ratio (SNR), when assuming an i.i.d. Rayleigh fading MIMO transmission channel, similarly to part 1.b. All the details of your code should be explained. Plot the simulation results together with the results of part 1.b and compare the results. (Hint: Verify your results with the literature).
3) Assume that the number of base-station antennas increases from 2 to 64 (i.e., as in the case of Massive MIMO).
a) Explain in detail how this change affects the received SNR, in comparison to the case of 2 base-station antennas.
b) For this case with 64 base-station antennas, simulate and plot the SER as a function of the SNR for hard MMSE (similarly to 1.b) and ML (similarly to 2.b) detection for the case of 2 users, and again assuming a Rayleigh MIMO channel. Discuss the changes you have made to the previous code.
c) Based on the previous results comment on what detector would you choose between MMSE and ML in the case the base-station has two antennas and in the case, it has 32 antennas. Justify your answer.
Question 2
In this part of the coursework assignment, we implement either i) a 5G OFDM system with LDPC codes or ii) an IEEE 802.11a based OFDM system with convolutional codes. The impact of different design parameters, e.g., modulation level, code rate, sampling rate, equalizer, etc. on the system performance will be investigated. For those who would like to learn more about 5G and LDPC codes, you are encouraged to choose the first option. LDPC tutorials are provided in the accompanying PPT.
The following m-files can be found on SurreyLearn:
test_ldpc: Main program for Option i)
test_conv: Main program for Option ii)
modulator: Perform modulation, i.e., bit-to-symbol mapping
demodulator: Perform demodulation, i.e., symbol-to-bit mapping
getchannel: Generate IEEE 802.11 multipath channel
onetap_equalizer: Implement one-tap equalization
fig: Serve as a template to make matlab plots based on simulation results
The system parameters are specified in Table 1.
The following issues will be investigated in this part of the assignment:
1) One-tap equalization design in OFDM systems
Let us assume the receiver has perfect knowledge of the channel for equalization. Simulation the system with i) no equalization; ii) ZF equalization; iii) MMSE equalization, and plot a BER vs. SNR curve in each scenario and compare their performance in one figure. Note that you need to implement ZF equalizer by yourself. In the report, give the mathematic expressions for ZF and MMSE equalizers and your matlab implementation of ZF equalizer, describe what can be observed by comparing the performance of 3 different schemes.
2) Effect of channel coding for OFDM systems
2.1 Remove channel coding/decoding, interleaving/de-interleaving in the main program and simulate the performance of an uncoded OFDM system. In the report, show your matlab implementation of the uncoded system, and compare the performance of coded and uncoded OFDM systems in the same figure, discuss what can be concluded from the simulation results and explain why.
Hints: For the uncoded systems, you need to make hard decisions on the transmitted bits based on the equalizer outputs.
2.2. Repeat Task 1 for uncoded OFDM systems, and compare the performance of ZF and MMSE equalization in one figure. Explain the rationale why the comparison result seems unusual and departs from people's common belief. This requires rigorous thinking and analysis.
3) OFDM systems with different data rates
3.1 Perform the following tasks for 3 OFDM systems with different data rates:
i) Implement a baseline system employing QPSK modulation, channel coding and interleaving; ii) Design and implement an OFDM system which doubles the data rate compared to the baseline system; iii) Design and implement an OFDM system which triples the data rate compared to the baseline system.
In the report, specify which part of the simulation code is changed in different systems, compare the performance of 3 systems in the same figure, discuss the price to pay for the increasing the data rate as well as the engineering tradeoff.
3.2 Perform the following tasks for 4 OFDM systems with different data rate:
i) Implement a baseline system employing QPSK modulation, the LDPC or convolutional code with code rate ½ and interleaving; ii) Design and implement an OFDM system that achieves about 33.3% higher data rate than the baseline system; iii) Design and implement an OFDM system that achieves 50% higher data rate than the baseline system; iv) Design and implement an OFDM system that achieves 60% higher data rate than the baseline system;
In the report, specify which part of the simulation code is changed in each system, compare the performance of 4 different systems in the same figure, discuss the price to pay for the increasing the data rate as well as the engineering tradeoff.
Hints: For the LDPC code, check the accompanying PPT to see how the code rate can be changed; for the convolution code, check matlab help manual to see how puncturing can be incorporated in the functions convenc (for encoding) and vitdec (for decoding). You need to adjust the number of data blocks in order to have integer number of transmitted bits.
4) Effect of channel on OFDM system performance
Change the FFT size is changed to 512 in this experiment. Execute the following: i) run the system with sampling frequency equal to 20MHz; ii) run the system with sampling frequency equal to 100MHz, adjust the length of CP properly according to the length of the channel and re-run the system. In the report, i) explain what happens to the channel's time-domain impulse response when the sampling frequency increases, e.g., in terms of the length of the channel; ii) calculate the SNR loss rate in these 2 scenarios; iii) discuss what can be observed from the performance comparison and explain why. The SNR loss rate is defined as β=Tcp/(T+Tcp), where T is the FFT interval and Tcp is the length of CP.
Hints: In-depth knowledge in wireless communications is needed to explain the channel effect on OFDM system performance, try to think about effective means of combating the effect of fading in wireless systems.