Moving Target Indicator (MTI) Radar

Unit Overview

This study guide supports Unit: Moving Target Indicator (MTI) Radar of ECE 521E. Every surveillance radar that looks toward the earth's surface must extract aircraft, ship, or vehicle echoes from echoes that are vastly stronger — returns from hills, buildings, sea waves, rain and chaff. This undesired return is clutter. MTI processing is the classical, and still indispensable, technique for rejecting zero-Doppler clutter while passing moving targets.

Indicative Study Time

Section 1 — Doppler Foundations of MTI

1.1 The Doppler Shift

Relative radial motion between radar and target shifts the echo frequency by the two-way Doppler amount:

• v_r > 0 toward the radar → positive shift (echo frequency rises).
• v_r < 0 receding → negative shift.
• Stationary clutter on a stationary platform → f_d = 0.

1.2 Phase Progression Pulse-to-Pulse

The echo phase is φ = 4πR/λ. Between pulses spaced T_r = 1/f_r apart the target moves ΔR = v_rT_r, so

MTI filtering is fundamentally the measurement of this pulse-to-pulse phase rotation at a fixed range cell. If Δφ is a multiple of 2π, successive echoes are identical and the target masquerades as clutter — the blind-speed condition (Section 4).

1.3 Sampling and Doppler Ambiguity

Pulsed radar samples the target's motion at f_r. Hence Doppler is only observable modulo f_r: frequencies f_d and f_d + nf_r are indistinguishable. The first blind speed corresponds exactly to f_d = f_r.

Section 2 — The MTI Concept

2.1 Signal Model at the Receiver

• s(n) — target echo: phase rotates by 2πf_dT_r per pulse.
• c(n) — clutter echo: essentially constant pulse-to-pulse (zero Doppler).
• w(n) — receiver noise.

An MTI filter H(z) is designed so that H(e^{jω}) ≈ 0 at ω = 0 (the clutter frequency) while passing signal energy at ω_d = 2πf_dT_r.

2.2 Coherent versus Non-Coherent MTI

2.3 Canonical Coherent MTI Chain

TX pulse ──► Antenna ◄── echo (target + clutter + noise)
                │
                ▼
        Mixer (STALO) ──► IF ──► Phase-sensitive detector ──► I/Q baseband
                                                              │
                                                              ├──► [delay Tᵣ] ──►(−)──► Σ ──► |·| ──► threshold ──► MTI video
                                                              └────────────────────────────┘

The delay-and-subtract element is the heart of MTI: identical (stationary) echoes cancel, phase-rotated echoes survive.

Section 3 — Delay-Line Cancellers

3.1 Single Canceller

Deep null at f = 0; nulls repeat at every multiple of f_r (blind speeds).

3.2 Double Canceller

3.3 Why the Double Canceller Wins Against Real Clutter

Real clutter is not perfectly stationary — vegetation sways, sea moves. Its spectrum has finite width σ_c. Near DC, the power response of the double canceller falls as f⁴ versus f² for the single canceller, so far more clutter power is suppressed for the same notch centre (see Fig. C in Section 8).

3.4 Binomial N-Pulse Canceller

Weights from Pascal's triangle give a maximally flat notch; N = 2 (three pulses) is the most common practical choice; N = 3 used when weather/chaff is severe. Costs: more pulses per dwell, longer transient, hardware taps.

3.5 Implementation History

Section 4 — Blind Speeds

4.1 Definition and Derivation

Blind speeds are target velocities at which the MTI response vanishes — the moving target cancels exactly like clutter:

Physical picture: the target moves ΔR = v_nT_r = nλ/2 between pulses — exactly n half-wavelengths. The echo is then identical on every pulse.

4.2 The Range–Doppler Dilemma

With one PRF, pushing blind speeds up (high f_r) shortens the unambiguous range R_u = c/(2f_r). Both cannot be large simultaneously — the fundamental motive for staggered PRF (Section 5) and PRF diversity.

Section 5 — Staggered PRF

5.1 Idea

Alternate PRIs T₁, T₂, … cyclically. Blind speeds of the individual intervals no longer coincide, so no velocity (except zero) is blind in every interval. The composite response for a two-interval double canceller:

5.2 Choosing the Stagger Ratio

5.3 Engineering Costs

• Range-gate alignment: target echoes arrive at different times per PRI → digital interpolation before differencing.
• Cancellation loss from timing jitter and misregistration.
• CFAR normalisation must cope with unequal dwell statistics.

Section 6 — MTI Filter Design

6.1 Transversal (FIR) Structure

Design objectives: deep notch at DC, maximally flat or equiripple rejection band, controlled passband gain. The binomial family is the simplest; Chebyshev/equiripple and optimum (Wiener) weights — matched to the measured clutter covariance — give the best subclutter visibility.

6.2 Recursive (IIR) Structure

Sharper notch per pulse; must be checked for stability (poles inside unit circle) and short-dwell transients. Practical radars cascade a wide-notch FIR (bulk weather/sea) with a narrow-notch IIR (residual ground clutter).

6.3 Adaptive and Clutter-Locked MTI

• Clutter locking: estimate the mean clutter Doppler and recenter the notch on it — vital when the platform moves (shipborne, airborne).
• TACCAR (Time-Averaged-Clutter Coherent Airborne Radar): the analog ancestor of clutter locking; a phase shifter driven by a feedback loop centres the clutter spectrum at 0 Hz.
• Fully adaptive MTI: per-cell weights from the sample covariance matrix — the scalar-time ancestor of STAP (Section 9).

6.4 Blind Phases and I/Q Processing

Single-channel (envelope) differencing can cancel a target whose two successive echo magnitudes coincide — "blind phase." Coherent I/Q samples preserve phase, so cancellation depends only on f_d; blind phases vanish. This is why all modern MTI is implemented at complex baseband.

Section 7 — Performance Measures

7.1 Definitions

7.2 Improvement Factor of the Canceller (Gaussian clutter)

With clutter rms spectral width σ_c, normalised σ_n = σ_c/f_r:

The σₙ⁻⁴ scaling of the double canceller explains its dominance: for narrow clutter, each canceller order multiplies rejection dramatically.

7.3 What Limits I

Section 8 — Radar Clutter and Its Models

8.1 Clutter Taxonomy

8.2 Amplitude Statistics

Section 9 — Limitations and the Path to MTD

9.1 Limitation Summary

1. Blind speeds (mitigated, never eliminated, by stagger).
2. Wide-clutter breakthrough — finite notch width.
3. No velocity measurement — detection-only output.
4. Platform motion sensitivity — needs clutter lock/STAP.
5. Stability ceiling on achievable cancellation.
6. Range-ambiguous clutter folding into the dwell.
7. Scalloping — gain varies with f_d, causing fluctuating target returns.

9.2 From MTI to MTD and STAP

Section 10 — Tutorial Problems

Attempt all problems before consulting the solutions. Values: c = 3×10⁸ m/s.

1. An L-band MTI radar (f₀ = 1.25 GHz) uses PRF = 900 Hz. Find (a) λ, (b) the first three blind speeds, (c) the unambiguous range.
2. A single delay-line canceller is used with f_r = 1 kHz. Compute |H| at f_d = 100, 250 and 500 Hz. Repeat for the double canceller.
3. Show from first principles that v₁R_u = λc/4. Hence choose f_r for λ = 0.15 m so that the first blind speed exceeds 100 m/s, and state the resulting R_u.
4. A two-PRF stagger uses f₁ = 950 Hz and f₂ = 1050 Hz at λ = 0.1 m. Compute the first blind speed and the stagger ratio.
5. Ground clutter has σ_c = 0.5 Hz and f_r = 800 Hz. Estimate the ideal improvement factors (dB) of single and double cancellers.
6. Define subclutter visibility. A system quotes I = 48 dB and total processing losses of 6 dB — what SCV should the data sheet claim?
7. Explain why an airborne early-warning radar flying at 250 m/s cannot use a fixed zero-Doppler notch at X-band, and name two techniques that rescue it.
8. A coastal radar must reject both sea clutter (σ_c = 20 Hz) and rain (σ_c = 40 Hz) at f_r = 1200 Hz. Sketch (describe) a suitable cascaded filter architecture.
9. Given I/Q samples x(n) = {3+j2, −1+j4, −2−j3, 1−j1}, compute the output of a double canceller for n = 2 and n = 3.
10. Write a one-page comparison of coherent and non-coherent MTI with a table of four differences and two application scenarios for each.

Section 11 — Glossary of Key Terms

Blind speed

Radial velocity v_n = nλf_r/2 at which MTI cancellation is complete; the target is invisible.

Clutter

Unwanted radar echoes from terrain, sea, rain, chaff or man-made structures.

Clutter attenuation (CA)

Ratio of input to output clutter power through the MTI filter.

COHO / STALO

Coherent oscillator / stable local oscillator providing the phase reference for coherent MTI.

Delay-line canceller

Filter subtracting the echo sample of the previous pulse from the current one.

Improvement factor (I)

Factor by which signal-to-clutter ratio is enhanced by MTI processing.

MTD

Moving Target Detector: MTI plus Doppler filter bank, per-bin CFAR and clutter maps.

MTI

Moving Target Indicator: processing that rejects stationary clutter and indicates moving targets.

PRF / PRI

Pulse repetition frequency f_r and its inverse T_r.

Staggered PRF

Cyclic alternation of PRIs to raise blind speeds without shortening range coverage proportionally.

Subclutter visibility (SCV)

How far (dB) a target may lie below clutter and still be detected.

TACCAR

Time-Averaged-Clutter Coherent Airborne Radar — clutter-locking phase compensation for platform motion.

STAP

Space-Time Adaptive Processing — joint adaptive spatial and Doppler filtering for airborne radars.

Section 12 — References and Further Reading

1. M. I. Skolnik, Introduction to Radar Systems, 3rd ed., McGraw-Hill, 2001 — Ch. 4 (MTI and pulse-Doppler radar).
2. M. I. Skolnik (ed.), Radar Handbook, 3rd ed., McGraw-Hill, 2008 — MTI, clutter, and Doppler processing chapters.
3. N. Levanon, Radar Principles, Wiley-Interscience, 1988.
4. B. R. Mahafza, Radar Systems Analysis and Design Using MATLAB, 3rd ed., CRC Press, 2013 — canceller simulation examples.
5. F. E. Nathanson, Radar Design Principles, 2nd ed., McGraw-Hill, 1991 — clutter statistics and MTI hardware limits.