Key Techniques Of Radio Classification Of Intrusion Detection System

In order to implement the radio intrusion detection and classification system on the Software-defined Radio System (SRS) of DSP centre, a project from Ngee Ann Polytechnic, several key techniques are investigated in this thesis, including spectrum sensing, modulation recognition, wide-band I/Q demodulator, FMC150control IP and timing recovery algorithm.Firstly, the spectrum sensing and modulation recognition algorithms are designed in MATLAB. Frequency domain approximate entropy-based algorithm is proposed for these applications. In addition, Monte-Carlo simulations are employed to evaluate their performance. According to ANOVA tests, the spectrum sensing detector can detect the2ASK signal with the signal to noise ratio (SNR) being as low as-28dB. When the SNR is-20dB, the proposed method has a detection rate of91.4%with the false alarm rate of10%. The modulation recognition scheme can classify the2ASK,4ASK,8ASK,2FSK,4FSK and8FSK signals with the correct rate of100%when the SNR is OdB. The average correct recognition rate for2PSK,4PSK and8PSK signals is91.55%.Secondly, an I2C-controlled I/Q demodulator is designed. The noise figure of the demodulator is about21dB according to the testing results. The output range of local oscillator is from137.5MHz to4.4GHz. The input frequency of the demodulator ranges from70MHz to2GHz. Furthermore, the FMC150IP core is customized and AXI-Lite interface is added. An AXI-based system on chip (SOC) is designed to integrate the control subsystem for the whole system.Finally, timing recovery algorithm is implemented on FPGA by using the tool of Xilinx System Generator. The algorithm can correct the timing error which is less than1%. From the constellation of the recovered signal, it is demonstrated that the proposed method can recover the symbol value. This module can also be used as teaching materials to help students to understand the application of digital signal processing techniques in digital communications.The research works have shaped the radio intrusion detection and classification system. With the hardware and algorithms, the system can detect modulated signals and classify them at a low SNR. The timing recovery algorithm can recover the constellation when there is timing error between transmitters and receivers. The research outcomes meet the system design requirements.