What Are IIR Filters?

Infinite Impulse Response (IIR) filters are a class of digital filters characterized by their use of feedback. Unlike Finite Impulse Response (FIR) filters, which rely solely on feedforward paths, IIR filters incorporate previous output samples into their current output calculation. This recursive structure allows IIR filters to achieve a given frequency response—such as a sharp cutoff—with far fewer coefficients than an equivalent FIR design. The result is a computationally efficient filter that requires less memory and fewer multiplications per sample, a significant advantage in real-time systems where processing resources are constrained.

Mathematically, an IIR filter is described by a difference equation that includes both input and output terms. The transfer function in the z-domain contains poles as well as zeros, which gives the filter its infinite impulse response. The placement of these poles and zeros determines the filter’s frequency selectivity, stability, and phase characteristics. For radar and sonar applications, the ability to design complex frequency responses with minimal hardware is a driving factor.

Key Design Considerations for IIR Filters in Radar and Sonar

Stability Analysis

The most critical design requirement for IIR filters in radar and sonar is stability. Because of the feedback loop, an IIR filter becomes unstable if any of its poles lie outside the unit circle in the z-plane. In practice, finite word length effects from fixed-point arithmetic can shift pole locations, potentially causing instability. Designers must account for coefficient quantization and use structures such as cascaded second-order sections (biquads) to maintain stability even under limited precision.

Phase Linearity Trade-Offs

Radar and sonar systems often rely on the phase of the signal for coherent processing, such as Doppler estimation or beamforming. FIR filters can provide exact linear phase, preserving signal shape. IIR filters, however, inherently introduce nonlinear phase distortion. While this is acceptable in many amplitude-focused tasks (e.g., noise reduction), it can degrade performance in coherent integration. Engineers often mitigate this by placing IIR filters in the magnitude path only, or by using all-pass phase equalizers to linearize the phase response over the frequency band of interest.

Coefficient Sensitivity

IIR filters are more sensitive to coefficient quantization than FIR filters. Small changes in coefficients can lead to large shifts in pole locations, affecting both frequency response and stability. In high-dynamic-range environments like radar, where signals span many tens of decibels, this sensitivity requires careful fixed-point scaling or the use of floating-point arithmetic when possible.

Application in Radar Systems

Clutter Suppression

Ground clutter, weather clutter, and sea clutter are major sources of interference in radar. These clutter signals are often stationary or slowly moving, with Doppler frequencies near zero. IIR high-pass filters, commonly known as Moving Target Indication (MTI) filters, can effectively block clutter while passing signals from moving targets. A simple single-delay canceler is an FIR filter, but more sophisticated designs—such as the cascaded integrator-comb (CIC) or recursive MTI filters—use IIR structures to achieve deeper clutter rejection with fewer stages. An example is the infinite memory MTI filter, which uses feedback to create a very narrow notch at zero Doppler, greatly reducing stationary clutter.

Pulse Doppler Processing

In pulse-Doppler radar, the received pulse train is processed to extract range and velocity information. IIR filters are used in the Doppler filter bank, often implemented as a set of narrowband bandpass filters. Compared to FIR-based filter banks, IIR designs require fewer filter taps, which reduces the computational load when many Doppler channels are needed. The trade-off is the nonlinear phase response, which may cause slight range walk, but this is often acceptable for moderate target velocities.

Adaptive Interference Cancellation

Modern radars employ adaptive filtering to combat rapidly changing interference. Adaptive IIR filters can track the frequency or statistics of interfering signals and adjust their coefficients accordingly. For instance, a notch filter tuned to a jammer’s carrier frequency can be implemented as an IIR structure that updates its pole location in real time. While adaptive IIR filters are more challenging to stabilize than their FIR counterparts, advances in algorithms like the recursive least squares (RLS) and lattice structures have made them viable in operational systems.

Application in Sonar Systems

Reverberation Rejection

Sonar systems face severe reverberation from the sea surface, seabed, and volume scatterers. Reverberation often exhibits a decaying tail that can mask weak target echoes. IIR filters can be designed as pre-whitening filters that flatten the reverberation spectrum, making it easier to detect targets. Alternatively, IIR-based matched filters can be used to maximize the signal-to-reverberation ratio. The recursive nature of IIR filters allows them to model the reverberation process efficiently, especially when the reverberation has a long time constant.

Target Detection in Noise

Active sonar pings propagate through water and reflect off objects. The returned signal is corrupted by ambient noise, flow noise, and self-noise from the platform. IIR low-pass or bandpass filters can reduce noise outside the frequency band of interest. Shipboard sonar systems often use IIR filters to suppress low-frequency machinery noise while preserving higher-frequency target echoes. The computational savings of IIR are especially important in multi-channel sonar systems with hundreds of hydrophones.

Doppler Estimation in Sonar

Moving targets—such as submarines, torpedoes, or marine mammals—induce a Doppler shift in the reflected sonar ping. IIR filters are employed in the Doppler processing chain to isolate specific velocity ranges. Because sonar signals are at much lower frequencies than radar (tens of kilohertz versus GHz), the filter order required for a given sharpness is lower, making IIR an even more attractive option. Recursive notch filters can track the Doppler frequency of a maneuvering target with low latency.

Advantages and Limitations

Advantages

  • Computational Efficiency: IIR filters require far fewer coefficients for the same passband/stopband steepness, reducing multiply-accumulate operations per sample. This is critical in real-time systems with high sample rates.
  • Low Memory Footprint: With fewer coefficients and fewer delay elements (especially compared to FIR), IIR filters consume less on-chip memory, making them suitable for FPGA and DSP implementations.
  • High Attenuation: Recursive designs can achieve very deep nulls (e.g., -60 dB) with low order, ideal for rejecting strong clutter or narrowband interference.

Limitations

  • Stability Risk: As noted, IIR filters can become unstable due to coefficient quantization or unexpected signal conditions. This requires careful design and often additional stability monitoring.
  • Nonlinear Phase: The phase response is not linear, which can distort transient signals and reduce the effectiveness of coherent processing. Equalization adds complexity.
  • Group Delay Variation: IIR filters exhibit group delay that varies with frequency. For some radar waveforms (e.g., linear frequency modulation), this can cause pulse stretching or compression, degrading range resolution.

Adaptive IIR Filters

Adaptive filters are increasingly used in both radar and sonar to cope with non-stationary environments. Adaptive IIR filters have the advantage of being able to model poles (i.e., resonances) in the interference, which is difficult with FIR as it requires very high order. Algorithms such as the adaptive lattice IIR filter or the equation-error approach provide stability monitoring and global convergence properties. In radar, they are used for adaptive clutter rejection; in sonar, for adaptive noise cancellation on towed arrays.

FPGA and Hardware Implementation

Field-programmable gate arrays (FPGAs) are the platform of choice for many real-time radar and sonar processors. IIR filters are implemented using pipelined multiplier-accumulator (MAC) structures. The recursive loop limits the maximum clock frequency, but techniques such as residue number system arithmetic or block processing (e.g., vector IIR by convolution) can mitigate this. Modern FPGAs with integrated DSP slices make high-speed IIR filters practical for sample rates above 100 MHz, supporting wideband radar chirps and high-resolution sonar.

Hybrid FIR-IIR Systems

To combine the strengths of both filter types, system designers often use a hybrid approach: FIR filters handle phase-sensitive processing (e.g., pulse compression), while IIR filters handle magnitude-only tasks (e.g., clutter cancellation). This partitioning maintains coherent processing performance while leveraging IIR efficiency. An example is the use of an FIR matched filter followed by an IIR MTI filter in a typical radar receiver chain.

Conclusion

IIR filters remain a fundamental building block in modern radar and sonar signal processing. Their ability to provide sharp frequency selectivity with low computational cost makes them indispensable for real-time applications where every millisecond matters. From clutter rejection in air defense radars to reverberation suppression in submarine sonars, IIR filters enhance detection range and reliability. While challenges such as stability and phase linearity must be managed through careful design and modern adaptive techniques, the ongoing evolution of digital signal processing hardware and algorithms continues to expand the role of IIR filters. As radar and sonar systems push toward higher bandwidths and more complex environments, the efficient yet powerful IIR filter will remain a key tool in the engineer’s arsenal.

External Resources