Homework answers / question archive / COMM

COMM

Project Management

Share With

COMM.SYS.450 Multicarrier and Multiantenna Techniques
Mandatory project work
Alternative 1: OFDM link simulation
Here the task is to implement in Matlab a baseband simulation model for a very basic CP-OFDM
transmission link, very similar to the Matlab exercises that you have done so far. You have to
implement the corresponding transmitter, channel model and receiver. The system is to be designed
such that the following specifications are met:
• 20 MHz bandwidth channel is available for transmission.
• Use 15 kHz subcarrier spacing.
• Due to ACLR requirements, a 10% of the available bandwidth is left as guard band in order
to reduce the emissions onto the adjacent channel. (5% on each side of the main carrier)
• Peak rate of at least 80 Mbps is to be supported.
You have to choose a convenient CP length, long enough to prevent ISI from happening due to the
multipath propagation, but as short as possible such that unnecessary overheads are avoided.
Choose the QAM modulation order in such a way that the data rate target is reached.
Consider ZF equalization at the receiver.
The transmission is carried out over a block-fading multipath channel model with given power-delay
profile. The channel characteristics depends on the last digit of your student number:
For channel estimation purposes
• Use random binary PSK as subcarrier symbols for constructing the OFDM symbol for
training.
• Use a frame structure containing 16 OFDM symbols: the training symbol + 15 data symbols
• Assume stationary channel during each frame (block-fading model), that is, the channel
remains unchanged within the duration of a frame.
Last digit of
student number
Channel model
0 Veh-AE
1 Veh-A
2 ETU
3 Ped-B
4 Veh-A
5 Veh-AE
6 Ped-B
7 ETU
8 Veh-AE
9 Veh-A
The mentioned channel models are widely used in the development and standardization of mobile
wireless systems. The power delay profiles are defined by the tap gains and multipath delays:
case 'Ped-B'
gains = [0 -0.9 -4.9 -8 -7.8 -23.9]; % in dB
delays = [0 200 800 1200 2300 3700]*1e-9; % in seconds
case 'Veh-A'
gains = [0 -1 -9 -10 -15 -20]; % in dB
delays = [0 310 710 1090 1730 2510]*1e-9; % in seconds
case 'Veh-AE'
gains = [0 -1 -9 -10 -15 -20]; % in dB
delays = [0 310 710 1090 1730 10000]*1e-9;% in seconds
case 'ETU'
gains = [-1 -1 -1 0 0 0 -3 -5 -7]; % in dB
delays = [0 50 120 200 230 500 1600 2300 5000]*1e-9;% in seconds
The channel model can be conveniently implemented in Matlab using the
comm.RayleighChannel function as follows:
fd = 0; % Doppler spread, zero in block-fading model
chan = comm.RayleighChannel( ...
'SampleRate', Fs, ...
'MaximumDopplerShift', fd, ...
'PathDelays', delays, ...
'AveragePathGains', gains);
channel_output = chan(channel_input(:));
% The following kind of line could be included in the simulation loop.
% It creates a new channel instance in each call:
release(chan);
The bit error rate (BER) over signal-to-noise ratios (SNRs) will serve to evaluate the system
performance. In order to calculate the BER, some facts need to be carefully considered:
• Use sufficient number of channel instances (at least 1000) to get smooth plot down to BER level of
10-3. Simulating one frame per channel instance is enough.
• Use reasonable SNR range, the (uncoded) BER range of 0.1 … 0.001 is interesting. SNR resolution of
2 dB is sufficient.
• SNR scaling tends to be a tricky issue in simulation studies. Notice that SNR is defined at the receiver
input as the ratio of the average received signal power and the noise power in the used bandwidth
(active subcarriers in the OFDM case).
• The AWGN power added in the simulations is white, i.e., constant in all subcarriers corresponding to
the used FFT size. For the useful signal power, the IFFT includes some scaling of the signal level. The
ratio of total to active subcarriers influences both the noise and signal power scaling.
• Easy solution: Use the awgn function of Matlab to reach targeted SNR for each received frame. This
corresponds to having ideal power control in the system, because the received signal power is actually
different for different channel instances.
• Also the comm.RayleighChannel can be configured to give constant channel power gain for
different instances, but then you need to figure out the different scaling factors to the received power.
• Alternatively, the simulations can be done by assuming constant average receive power (over high
number of channel instances). By scaling the power delay profile properly, the average channel power
gain can be set to unity.
• For calculating the BER, you may find the comm.ErrorRate object or biterr function useful.
But using either is not strictly required.
• For improving the BER, you may find using Gray coding in data modulation useful.
What to return
• Matlab script
• BER over SNR plot
• Brief description and analysis of the implemented system.
• Short explanation of the main characteristics of OFDM systems, advantages and
disadvantages compared to single carrier waveforms.