Matched filtering is often introduced as the gold standard for radar detection, the optimal solution when the signal is known and the noise is white and Gaussian. However, in many real-world scenarios, matched filtering can become more of a constraint than a solution, such as when strong targets hide the returns of weaker targets. Enter, mismatched filtering.

In this workshop, participants will gain hands-on experience with designing and evaluating their own mismatched filters. Through prepared MATLAB coding examples and in-person explanations, participants will learn how filter design parameters impact detection performance in challenging scenarios. We’ll focus on quantifying the benefits and limitations of certain designs and utilize powerful visualization tools to help in decision making.

Attendees will learn many different strategies in this workshop, from choosing the right window/taper to finding optimal solutions with custom constraints. Throughout all of this, we will make sure to play close attention to the costs of each design decision to make sure our solution remains robust to the challenges of the field.

This guide was created with the help of AI, based on the presentation's transcript. Its goal is to give you useful context and background so you can get the most out of the session.

What this presentation is about and why it matters

This talk revisits a fundamental building block of radar and many sensing systems: matched filtering. Matched filters are the textbook optimal detector for a known transmitted waveform in white Gaussian noise. But in real systems the dominant problem is often not noise — itis interference and the sidelobes produced by the matched-filter autocorrelation. Strong reflectors (a large aircraft, building, or sea clutter) can mask nearby weak targets (small drones, birds, or subtle features). Spencer Markowitz shows why the textbook optimal solution can be a practical liability, and demonstrates practical alternatives (mismatched filters, spectral tapering, least-squares filter design and beam spoiling) that trade SNR for better detectability of weak, nearby returns. Engineers will leave with intuition and hands-on MATLAB workflows for making these trade-offs.

Who will benefit the most from this presentation

• DSP engineers and students who work on radar, sonar, lidar or ultrasonic sensing and need to understand range processing trade-offs.
• System designers deciding between waveform types (LFM chirps vs phase-coded) and evaluating detection performance in clutter-limited scenarios.
• Practitioners who want practical, reproducible MATLAB examples to experiment with windowing, mismatched filters, and filter-length choices.
• Anyone curious about how a small, deliberate degradation of an optimal detector can improve real-world performance.

What you need to know

This presentation assumes basic familiarity with sampled signals, FFTs and correlation. Key concepts to review beforehand will help you get more from the demos:

• Matched filter & autocorrelation: The matched filter for a transmit waveform $s[n]$ is the time-reversed conjugate, and matched filtering is equivalent to autocorrelation. It concentrates energy into a peak so range delay is easier to detect.
• Range mapping: Delay to range conversion is simple: if a return appears $n$ samples late at sample rate $f_s$, then $\Delta t = n/f_s$ and Range = $c\,\Delta t/2$, where $c$ is the speed of light. (So $\mathrm{range}=c\,n/(2f_s)$.)
• Sidelobes vs noise: Matched filters maximize peak SNR, but sidelobes in the autocorrelation can be the limiting factor — a large targetcan mask weak ones even when noise is low.
• Windowing / tapering: Applying a spectral (frequency-domain) window to the matched-filter template reduces time-domain sidelobes, at the cost of wider main lobe and some SNR (mismatch) loss. For chirp (LFM) signals equivalent effects can be obtained by time-domain tapering.
• Mismatched filters by least squares: You can design a filter longer than the waveform and solve a least-squares problem to push sidelobes toward an ideal impulse response. This often reduces integrated sidelobe energy drastically but may introduce artifacts and reduce Doppler tolerance.
• Beam spoiling: Intentionally allow a slightly wider main lobe in the design objective so the optimizer can further suppress sidelobes; a small widening can give large sidelobe benefits in many codes.
• Doppler and ambiguity: Moving targets cause Doppler shifts that change matched-filter responses. Consider Doppler tolerance when designing windows or mismatched filters. The ambiguity function summarizes joint range/Doppler response.

Glossary

• Matched filter – A filter matched to the transmitted waveform (time-reversed conjugate) that maximizes peak SNR for white Gaussian noise.
• Sidelobes – Secondary peaks in the autocorrelation / impulse response that can mask nearby weak targets.
• LFM (chirp) – Linear frequency modulated pulse; frequency sweeps linearly across the band during the pulse.
• Phase-coded waveform – A pulse made of discrete chips with controlled phases (e.g., Barker, P1) for better autocorrelation properties.
• Window (taper) – A spectral or time-domain weighting used to reduce sidelobes at the cost of main-lobe widening and mismatch loss.
• Mismatched filter – Any filter not equal to the matched filter (includes windowed matched filters and least-squares designs); trades optimality for other benefits.
• Integrated sidelobe level (ISL) – Total energy in sidelobes compared to main lobe; a holistic measure versus the worst-case PSL.
• Peak sidelobe level (PSL) – The amplitude (usually in dB) of the largest sidelobe relative to the main peak.
• Ambiguity function – Two-dimensional response showing matched-filter output across range delay and Doppler shift.
• Beam spoiling – Designing the filter objective to accept a wider main lobe so sidelobes can be further suppressed.

Parting words

SpencerMarkowitz blends clear intuition with reproducible MATLAB demos, showing that sometimes "perfect" (the matched filter) is suboptimal for the goals you actually care about. The talk is practical, experiment-driven and focused on trade-offs you will encounter in real radar or sensing systems. If you want hands-on ways to tune sidelobes, understand SNR cost, and try least-squares filter design or simple spectral tapering, this workshop is a great use of your time.