NN-Based 8FSK Demodulator for the Covert Channel.

Area of Science:

• Information Theory
• Digital Communications
• Signal Processing

Background:

• Covert communication channels are crucial for secure data transmission.
• 8-Frequency Shift Keying (8FSK) modulation is a candidate for covert waveforms.
• Channel estimation errors and imperfect successive interference cancellation (SIC) degrade performance.

Purpose of the Study:

• To propose and examine a superposition-based covert channel and its demodulator.
• To investigate the impact of channel estimation error and imperfect SIC on demodulation.
• To introduce a neural network (NN)-based demodulator to mitigate these imperfections.

Main Methods:

• Utilized an 8FSK modulation scheme for the covert waveform.
• Analyzed the effects of channel estimation errors and imperfect SIC.
• Developed and evaluated a neural network (NN)-based demodulator.
• Compared the NN demodulator against a traditional 8FSK correlator-based receiver.
• Assessed performance using hard- and soft-decision detectors.

Main Results:

• The NN-based demodulator generally outperforms the traditional correlator-based 8FSK demodulator.
• Even with perfect SIC, the NN demodulator does not achieve zero Bit Error Rate (BER).
• With additional covert channel coding, the NN demodulator achieves error-free demodulation.
• Error-free demodulation is possible even with a four-times greater channel gain error.

Conclusions:

• The proposed NN-based demodulator offers a superior solution for 8FSK covert communication compared to traditional methods.
• The NN demodulator effectively mitigates performance degradation caused by channel estimation errors and imperfect SIC.
• The combination of NN demodulation and channel coding enables robust and error-free covert data transmission.