BPSK and 16-QAM modulated OFDM signal transmission for IoT devices over Rayleigh flat fading channels in the Wi-Fi Halow frequency band

PythonCodeForDataGeneration.py: A comprehensive simulation framework designed to model IoT devices competing for channel access within the 902–928 MHz unlicensed band. 50,000 IoT devices are evenly divided into two groups. One group uses Binary Phase Shift Keying (BPSK), while the other uses 16-Quadrature Amplitude Modulation (16-QAM). Furthermore, the IoT devices are categorized into three data rate–based classes in line with common IoT standards.: 1000 kbps (Long Term Evolution Category 1, (LTE Cat1)), 160 kbps (Narrowband IoT, (NB-IoT)), and 72 kbps (Extended Coverage GSM (EC-GSM)). Signal transmission employs an Orthogonal Frequency Division Multiplexing (OFDM) system with parameters fixed as follows: cyclic prefix length of 16, Fast Fourier Transform (FFT) size of 64, and number of sub-carriers is 64. On the transmitter side, a random binary sequence to be transmitted is generated. Then, the binary sequence is modulated, either BPSK or 16-QAM, depending on the specifications of the device transmitting. Next, the modulated symbols are assigned to OFDM data sub-carriers and the cyclic prefix is added. To model realistic propagation conditions, a Rayleigh flat fading channel is assumed. To take into account the effect of the channel, each OFDM symbol is convolved with a 4-tap Rayleigh fading channel. After that, additive white gaussian noise (AWGN) is added. On the receiver side, for each sub-band, the following features from the data obtained after adding the AWGN is calculated: total signal power; sums of the real and imaginary components; variances of the real and imaginary parts; spans of the real and imaginary parts; as well as the skewness and kurtosis of both the real and imaginary components. Each 1 kHz sub-band 902–928 MHz unlicensed frequency band is labelled according to its usage status during the simulation: sub-bands carrying IoT device transmission subjected to Rayleigh flat fading and AWGN, are labelled as ‘1’, indicating signal presence; sub-bands containing only AWGN without any IoT transmission are labelled as ‘0’, indicating signal absence.

PandasCodeForDataProcessing.ipynb: Selects the first 5000 sub-bands labelled as '1' and the first 5000 sub-bands labelled as '0'. Then Min-Max normalisation is performed.

Final_Dataset_SNR_Minus5.csv, Final_Dataset_SNR_Minus10.csv, Final_Dataset_SNR_Minus15.csv, Final_Dataset_SNR_Minus20.csv and Final_Dataset_SNR_Minus25.csv: Simulations are conducted over a range of SNRs from -5 dB down to -25 dB by varying the transmitted signal power in PythonCodeForDataGeneration.py