Sunteți pe pagina 1din 5

Laboratory #4

Orthogonal Frequency Division Multiplexing

SYSC 4607 Wireless Communications Department of Systems and Computer Engineering Faculty of Engineering Carleton University March 2014

Department of Systems & Computer Engineering

Page 2 of 5

Purpose and Objectives


The purpose of this laboratory experiment is to investigate the power spectral density and peakto-average power ratio of OFDM signals, and to demonstrate the resilience of OFDM signals to the effects of frequency selective channels. In this experiment we will implement an OFDM transceiver using parameters loosely based on the IEEE 802.11n standard for wireless LANs.

System Model
In an OFDM system, the data to transmit is first mapped to modulation symbols. The 802.11n standard supports the use of BPSK, 4-QAM, 16-QAM, and 64-QAM, but we will only implement 16-QAM. We will use a 20 MHz band divided into subcarriers, with subcarrier { }. Only indices subcarriers are used to carry information, while the others are left blank to simplify the requirements on the analogue hardware components. In our system a message word of 22,400 bits will be mapped to 5,600 modulation symbols using 16-QAM. These modulation symbols will be transmitted in 100 ODFM symbols, with each } are the 56 16-QAM symOFDM symbol containing 56 modulation symbols. If { bols to transmit in one ODFM symbol, then the symbol to transmit on subcarrier k is { This mapping is illustrated below:
0 0 0 0 k -32 -31 -30 -29 -28 -1 0 1 28 29 30 31

IDFT

0 0 0

The time-domain samples of the OFDM symbol are determined from the crete-time Fourier transform (IDFT) according to:

-point inverse dis-

SYSC 4607 Wireless Communication

Laboratory #3

Department of Systems & Computer Engineering

Page 3 of 5

A cyclic prefix, consisting of a replica of the last samples produced by the IDFT, is inserted before the samples to generate the full OFDM symbol which contains samples, as shown below:

For the transmitted signal to fit within a bandwidth of 20 MHz the sample period is nsec, so the ODFM symbol has a duration of . In a normal 802.11n system the number of OFDM symbols per frame is variable, depending on how much data needs to be transmitted, and each frame is preceded by several training and control symbols, but we will ignore these and only transmit data symbols. Transmission will be carried out in frames of OFDM symbols, so a total of samples will be th transmitted for each frame. Let be the n sample. Passing the samples through an analogue-to-digital converter gives the transmitted complex lowpass equivalent signal ( ) where ( ) = 10 MHz. ( )

is the normalized impulse response of an ideal lowpass filter with bandwidth The transmitted signal passes through a frequency selective channel, giving ( ) where ( ) ( ) ( )

( ) is the channel impulse response, and

( ) is AWGN.

At the receiver the signal is sampled at the sampling rate, giving samples ( )

} is the discrete-time channel imwhere L is the channel delay spread (in samples), { pulse response, and { } are noise samples, modelled as independent Gaussian random variables with zero mean and variance for the real and imaginary parts. The received samples corresponding to the transmission of the cyclic prefixes are discarded, leaving samples for each OFDM symbol. Let { } be the samples for one OFDM symbol. The -point discrete Fourier transform of the samples is calculated to con-

SYSC 4607 Wireless Communication

Laboratory #3

Department of Systems & Computer Engineering vert to the frequency domain. The received value for subcarrier } is {

Page 4 of 5

To estimate the transmitted data, calculate the decision variable where

Then discard the symbols received on the unused carriers, giving {

which can be used to estimate the bits in .

Laboratory
Part I: Implementation Implement the system described by the model given above in MATLAB. It is recommended that you build your simulator in steps, like you did in Lab #1: 1) Map 24,400 random bits to 5600 symbols using 16-QAM, and then use your decision device to convert the symbols back to bits. Make sure the BER is zero. 2) Map the 16-QAM symbols to subcarriers and back again. Check the BER. 3) Take the IDFT of the symbols in each OFDM symbol to produce the time-domain samples, and convert them back to the frequency domain using a DFT. You should use the MATLAB function ifft( ) to calculate the IDFT, but read help ifft and note the differences between it and the IDFT that you need. In particular, ifft( ) divides by instead of , and expects the symbols to be in the following order: . You can use fftshift( ) to fix the order of the symbols. You should also use fft( ) to calculate the DFT at the receiver, again paying attention to the differences. 4) Insert the cyclic prefixes and produce one long OFDM frame of Modify you receiver accordingly. 5) Generate the analogue waveform, ( ). Use an over-sampling factor of use the interp( ) function for this. The correct syntax is:
>> vt = interp(v, eta, 11, 57/64) / sqrt(Ts);

samples. . You can

where vt is the oversampled signal, v are the samples of the OFDM frame, eta is the oversampling factor, and Ts = . Use vt(1:eta:end) * sqrt(Ts) to recover the transmitted samples. SYSC 4607 Wireless Communication Laboratory #3

Department of Systems & Computer Engineering

Page 5 of 5

6) You do not need to implement the AWGN or frequency selective channel until Part IV. Part II: Spectral Analysis Plot the PSD of ( ), using the technique used in Lab #1. You should average the results of 100 frames. Part III: Peak-to-Average Power Ratio Find the average and peak instantaneous power of ( ) over 100 frames. Calculate the peak-toaverage power ratio (PAPR) in dB. Part IV: BER Performance AWGN Channel Generate a graph of the BER vs. for between 0 and 14 dB by simulating 1000 frames in an AWGN channel. To speed up your simulations, you should not include the A/D converter (interp) and downsampler in your simulations. Your received samples are just equal to your transmitted samples ( ) with complex Gaussian noise added. The variance of the real and imaginary parts of the noise should be , where depends on and . To calculate , note that each frame conveys bits, and the total transmitted energy for the frame is

where is the average energy of each 16-QAM symbol. Compare your results with the nonOFDM transmission of 16-QAM you studied in Lab #2, and justify any differences. Part V: BER Performance Frequency Selective Channel Modify your simulator to use a frequency selective channel with impulse response { You may want to use the MATLAB conv( ) function to perform the convolution of the channel impulse response with the transmitted samples to give the received samples. You can use the fft( ) function to calculate , but make sure you use an -point FFT. Use to determine the decision variables. Plot the BER vs. for between 0 and 18 dB, limiting the graph to BERs greater than . Why is the BER different than in Part IV?

SYSC 4607 Wireless Communication

Laboratory #3

S-ar putea să vă placă și