Published: 2026-05-25 Moving Target Indication (MTI) is a radar signal-processing technique that suppresses echoes from stationary or slowly-moving clutterâsuch as ground terrain, buildings, vegetation, sea surface, and weather phenomenaâwhile preserving and displaying returns from moving targets of interest. The fundamental physical principle exploited by MTI is the Doppler effect: a target in motion relative to the radar produces an echo whose carrier frequency is shifted by an amount proportional to its radial velocity, whereas stationary clutter produces echoes at the nominal carrier frequency (or, more precisely, at frequencies that are coherently repeated from pulse to pulse). By examining the phase progression across successive transmitted pulses, an MTI radar distinguishes moving targets from clutter and achieves what is termed clutter rejection. The concept was pioneered during World War II, when early coherent radar systems introduced the idea of comparing the phase of returns from one pulse to the next. Modern MTI encompasses a family of implementation strategiesâfrom simple single- and double-delay-line cancellers to sophisticated staggered-PRF (Pulse Repetition Frequency) and adaptive filter architecturesâbut the underlying principle remains constant: exploit inter-pulse phase coherence to separate moving targets from the clutter background. A closely related concept is Moving Target Detection (MTD), which extends MTI by incorporating bank-of-filters Doppler processing (typically via FFT) to provide velocity discrimination and improved detection performance. In many contemporary systems, MTI preprocessing and MTD filtering are cascaded; the glossary entry here focuses on the classical MTI principles. The simplest MTI filter is the single-delay-line canceller (also called a two-pulse canceller). It computes the difference between two successive pulse returns: $$y(n) = x(n) - x(n-1)$$ where $x(n)$ is the complex (I/Q) video sample from the $n$-th pulse. For stationary clutter, $x(n) \approx x(n-1)$, so $y(n) \approx 0$. For a moving target, the phase difference produces a nonzero output. In the frequency domain, this operation corresponds to a filter with transfer function $$H(f) = 1 - e^{-j2\pi f / f_r} = 2j\sin!\left(\frac{\pi f}{f_r}\right) e^{-j\pi f/f_r}$$ which has nulls at DC and all integer multiples of $f_r$âprecisely where stationary clutter resides. However, the response also goes to zero at blind speeds, and the passband shape (a sinusoid) provides only moderate clutter attenuation for near-zero-Doppler clutter spread. Cascading two single cancellers yields the double-delay-line canceller (three-pulse canceller): $$y(n) = x(n) - 2x(n-1) + x(n-2)$$ with transfer function $|H(f)| \propto \sin^2(\pi f/f_r)$. The double canceller provides a sharper null around DC, significantly improving rejection of clutter with finite spectral width, and a flatter passband for moving targets. Its impulse response $[1, -2, 1]$ is the binomial coefficients with alternating signs, recognisable as a second-order finite impulse response (FIR) filter. Further cascading or using IIR (feedback) structures can deepen the clutter null or shape the passband more favourably. An $N$-pulse canceller (order $N-1$) has an $\sin^{N-1}(\pi f/f_r)$ response. However, higher-order cancellers also widen the blind-speed notches and may amplify blind-phase effects. Recursive (feedback) MTI filters can achieve sharper clutter rejection with fewer delay elements but are more sensitive to phase instabilities and may exhibit ringing. By alternating between two or more PRIs, the blind-speed locations become non-coincident, so that a target falling at a blind speed for one PRF will generally be visible at another. The effective unambiguous velocity range is extended, and the average clutter rejection is improved. The design of stagger patterns involves careful selection of PRI ratios to ensure adequate first-line clutter suppression while controlling the residual blind-speed residues. The improvement factor $I$, denoted Improvement Factor in Chinese radar terminology, is defined by the IEEE as: The signal-to-clutter ratio at the output of the clutter filter, divided by the signal-to-clutter ratio at the input of the clutter filter, averaged uniformly over all target radial velocities from zero to the maximum unambiguous velocity. Formally: $$I = \frac{\overline{S_o/C_o}}{S_i/C_i}$$ where the overbar denotes averaging over all target Doppler frequencies. $I$ is a single figure of merit that characterises the average improvement in detectability provided by the MTI system. It captures both the degree of clutter suppression and the effect of the MTI filter on target signals (including losses at blind speeds). A related metric is the clutter attenuation (CA), which is the ratio of clutter power at the filter input to clutter power at the output (not averaged over target velocity). In general $I \leq \text{CA}$, with equality when the MTI filter does not attenuate the target on average. Oscillator phase noise is conventionally expressed as $\mathcal{L}(f_m)$, the single-sideband phase noise power spectral density (in dBc/Hz) at offset frequency $f_m$ from the carrier. In a coherent radar, the STALO and COHO phase noise is impressed on both the transmitted signal and the received signal. For a homodyne or zero-IF architecture, the common-path phase noise cancels to first order; however, the delay mismatch $\tau_d$ (the round-trip propagation delay to the clutter cell being processed) converts phase noise into residual amplitude/phase fluctuations after mixing. For clutter at range $R$, $\tau_d = 2R/c$. The residual phase error after mixing a delayed return with a reference is the phase noise difference $\Delta\phi(t) = \phi(t) - \phi(t - \tau_d)$. For a clutter cell at delay $\tau_d$, the mean-square residual phase error is: $$\sigma_{\Delta\phi}^2 = 2 \int_0^{f_r/2} |H_{\text{MTI}}(f_m)|^2 \cdot 2\mathcal{L}(f_m) \cdot [1 - \cos(2\pi f_m \tau_d)] \, df_m$$ where $H_{\text{MTI}}(f_m)$ is the MTI filter's baseband response. This integral shows that: 1. Near-carrier phase noise (small $f_m$) is suppressed by the factor $[1 - \cos(2\pi f_m \tau_d)]$, which vanishes for small delaysâhence close-in phase noise matters most for long-range (large $\tau_d$) clutter. 2. The MTI filter shape weights the noise: a deeper, wider null (higher-order canceller) suppresses more of the phase-noise-induced clutter residual, but also passes less target signal energy. 3. The resulting clutter residual due to phase noise limits the achievable cancellation ratio and, consequently, the improvement factor. When phase noise is the dominant source of imperfect cancellation, the improvement factor is bounded by: $$I_{\max} \approx \frac{1}{\sigma_{\Delta\phi}^2}$$ (in linear units), or equivalently $I_{\max}(\text{dB}) \approx -10\log_{10}(\sigma_{\Delta\phi}^2)$. In practical terms, to achieve an improvement factor of 50 dB (a factor of $10^5$), the rms residual phase error must be no greater than approximately $0.18°$ ($\approx 3 \times 10^{-3}$ rad). This translates to stringent requirements on oscillator spectral purity, particularly at offset frequencies corresponding to the MTI filter passband. For example, a ground-based radar processing clutter at ranges up to 200 km ($\tau_d \approx 1.33\,\text{ms}$) requires excellent phase noise performance at offset frequencies from tens of Hz to tens of kHz. To maximise the improvement factor: - Use low-noise oscillators: Modern designs employ crystal oscillators, SAW oscillators, or synthesizers with carefully optimised close-in phase noise specifications (often $<-110$ dBc/Hz at 1 kHz offset for high-performance radars). - Minimise delay mismatch: Monostatic radars inherently have delay; in some architectures (e.g., pseudo-coherent MTI using a COHO locked per pulse), the effective $\tau_d$ is reduced. - Match STALO/COHO paths: Any unshared path between the transmitter and receiver introduces independent phase noise contributions that do not cancel. - Select MTI filter order judiciously: Higher-order filters provide deeper nulls but may amplify phase-noise contributions in their passbands; the optimal order balances clutter suppression against noise enhancement. See also: MTD (Moving Target Detection); CFAR (Constant False Alarm Rate); Doppler filter bank; clutter map; coherent-on-receive; STALO / COHO architecture.