Subspace projection and time-varying AR modeling for anti-jamming DS-CDMA communications

Summary form only given. Direct-sequence spread-spectrum (DS/SS) communication systems have certain degree of inherent immunity to intentional and/or unintentional jammers. However, in some applications, the jammer might be much stronger than the desired DS/SS signal, and the processing gain due to spreading might be insufficient to provide enough jamming resistance for decoding the useful signal reliably. In such cases, jammer suppression needs to be done prior to symbol detection. For stationary interference, many jammer mitigation techniques have been developed to remove its effect adequately. However, the nonstationary interferences cannot be adequately suppressed using a single domain mitigation algorithm due to the fact that signal´s parameters are time-varying. Frequency modulated (FM) interferers are such kind of examples. Many interference suppression techniques are reported and most of the techniques use IF estimation and time-frequency analysis (TFA). The disadvantage of TFA is its large computation burden and slow convergence. To increase convergence and lower the computational complexity, this paper proposed to estimate the IF, based on time-varying autoregressive (TV-AR) modeling. One of the important IF-based interferer rejection techniques is to uses the jammer IF to construct a time-varying excision notch filter that effectively removes the interference. However, this notch filtering technique may cause significant distortion to the desired signal, resulting in poor receiver performance. Recently, subspace projection techniques based on IF estimation have been devised for FM interference excision in DS/SS communications. It is demonstrated recently that the subspace projection method can improve the output SNR significantly. In this work, we model the received data as a linear combination of FM interferences with time-varying IFs, thermal noise and the desired DS/SS CDMA signal using a time varying autoregressive (TV-AR) model. We then use the orthogonal and the oblique projection techniques to suppress the jammer based on the estimated IF from the TV-AR model. Finally, from simulation results, comparing the proposed subspace jammer suppression method with the time-varying notch filter method, the proposed method induces less distortion to t- he desired signal and can improve the performance of the whole system.