Narrowband Interference Rejection

A time-domain approach that dynamically places a narrow attenuation band at the instantaneous frequency of the jammer. An adaptive algorithm continuously tunes the filter coefficients to track a non-stationary interferer.

• LMS-based tracking: The Least Mean Square algorithm adjusts a second-order IIR filter in real time
• Key trade-off: Deeper notches remove more interference but introduce phase distortion at band edges
• Effective against: Continuous Wave (CW) tones and slowly varying FM jammers
• Limitation: Multiple jammers require cascaded notch stages, increasing group delay

Converts blocks of received samples into the frequency domain via FFT, identifies narrowband energy spikes exceeding an adaptive threshold, and zeroes or clamps those bins before inverse transformation.

• Excision strategies: Hard thresholding (zeroing bins) vs. soft clamping (limiting amplitude)
• Windowing: Applying a window function before the FFT reduces spectral leakage and improves jammer localization
• Overlap-add processing: Reconstructs the continuous time signal without block-edge discontinuities
• Advantage: Naturally handles multiple simultaneous jammers without additional filter stages

Exploits the fact that a narrowband jammer has high sample-to-sample correlation while the wideband DSSS signal appears noise-like. A Wiener filter predicts the jammer's next sample and subtracts the prediction from the received signal.

• Principle: The filter learns the jammer's autocorrelation structure to form a one-step predictor
• Adaptation: Filter taps update continuously using the RLS or LMS algorithm
• Residual: The prediction error contains the desired wideband signal plus thermal noise
• Strength: Preserves signal energy in the jammer's band rather than deleting it entirely

Decomposes the received signal using a wavelet packet transform to localize interference in both time and frequency simultaneously. This joint resolution enables precise excision of transient or pulsed jammers.

• Multi-resolution analysis: Deeper decomposition levels provide finer frequency resolution
• Threshold selection: Universal thresholding or level-dependent thresholds based on noise variance estimation
• Basis selection: Choosing the optimal wavelet family (Daubechies, Symlet) for the jammer's shape
• Best for: Non-stationary jammers that sweep in frequency or pulse on and off rapidly

Estimates the jammer's signal subspace from the received data's covariance matrix and projects the received vector onto the orthogonal noise subspace, effectively nulling the interference.

• Eigenvalue decomposition: Large eigenvalues correspond to the jammer subspace; small eigenvalues to noise and desired signal
• Rank estimation: MDL or AIC criteria determine the number of jammer components
• Robustness: Works well even when the jammer's frequency is unknown and time-varying
• Computational cost: Higher than filtering methods due to matrix operations on each block

Leverages knowledge of the spreading code to regenerate and subtract the desired signal, leaving a cleaner estimate of the jammer for more effective cancellation in an iterative loop.

• Successive Interference Cancellation (SIC): Alternates between despreading the desired signal and re-estimating the jammer
• Parallel architecture: Multiple cancellation stages can handle several jammers simultaneously
• Requirement: Requires coarse code synchronization before cancellation begins
• Performance: Approaches theoretical maximum signal-to-interference ratio improvement

A closed-loop filtering technique that dynamically tracks and suppresses narrowband interference by placing a deep spectral null at the instantaneous frequency of the jammer. The filter coefficients update continuously using algorithms like Least Mean Squares (LMS) to follow a non-stationary tone. Key characteristics:

• Tracks frequency-hopping jammers in real time
• Introduces minimal distortion to the wideband desired signal
• Typically implemented as an IIR lattice structure for stability
• Convergence speed trades off against notch depth and bandwidth

A block-processing method that converts the received signal to the frequency domain via FFT, identifies and zeroes out bins containing interference energy, then reconstructs the time-domain signal via IFFT for despreading. This approach excels against multiple simultaneous tones. Implementation considerations:

• Requires overlap-add or overlap-save to prevent block-edge artifacts
• Threshold setting is critical: too aggressive removes signal energy
• Can use wavelet transforms for non-stationary interference
• Well-suited to FPGA implementation with pipelined FFT cores

The fundamental purpose of interference rejection: restoring the spread-to-despread gain ratio that would otherwise be lost when a strong jammer saturates the receiver front-end. Without excision, the jammer power dominates the correlator input, collapsing the effective Eb/N0. Key metrics:

• Gp = Bspread / Binfo — theoretical maximum gain
• Excision loss must be subtracted from total link budget
• Residual interference after filtering still degrades bit error rate (BER)
• Adaptive systems aim to maximize signal-to-interference-plus-noise ratio (SINR) at the despreader input

The power ratio of the interfering tone to the desired spread spectrum signal at the receiver input, expressed in dB. This parameter dictates the required rejection depth and filter selectivity:

• JSR > 30 dB: Requires high-order notch filters or transform-domain excision
• JSR > 60 dB: May necessitate analog pre-filtering before the ADC to prevent saturation
• JSR estimation is a blind process — the receiver must characterize the interference without prior knowledge
• Directly determines the jamming margin of the overall system

The decision logic within transform-domain excision that separates interference bins from signal-plus-noise bins. Common algorithms include:

• N-sigma thresholding: Excision when bin power exceeds mean noise floor by N standard deviations
• Median filtering: Robust to outliers; threshold set as a multiple of the median bin power
• Order-statistic CFAR: Constant false alarm rate detection adapted from radar processing
• Iterative clipping: Repeatedly zero the strongest bins until residual power stabilizes Threshold selection balances probability of detection against probability of false alarm on signal bins.

The critical coupling between the interference rejection stage and the Delay Lock Loop (DLL) used for code synchronization. Excision filtering introduces group delay and phase distortion that can bias the DLL's timing error discriminator. Mitigation strategies:

• Use linear-phase FIR structures where possible
• Compensate for known filter delay in the code NCO
• Monitor S-curve zero-crossing for bias after filter insertion
• In extreme cases, implement joint interference suppression and code tracking as a unified estimation problem