ISAR (Inverse SAR)

ISAR — When the Target Moves Instead of the Radar

Inverse SAR (ISAR) reverses the SAR concept: a stationary (or slowly moving) radar images a rotating target. The target's rotation provides the synthetic aperture, enabling cross-range resolution of ships, aircraft, satellites, and vehicles. The fundamental mechanism is identical to SAR — relative motion creates a synthetic aperture — but the aperture is now controlled by the (unknown) target motion rather than the (known) platform trajectory.

Definition:

ISAR Geometry and Turntable Model

A radar at fixed position illuminates a target rotating about an axis perpendicular to the line of sight with angular velocity ω\omega. Over observation time TT, the total rotation angle is:

Δθ=ωT.\Delta\theta = \omega T.

Each scatterer at position (xk,yk)(x_k, y_k) relative to the rotation center traces a range history:

Rk(η)≈R0+xk+ykωη,R_k(\eta) \approx R_0 + x_k + y_k \omega \eta,

for small rotation angles ωη≪1\omega\eta \ll 1. The cross-range coordinate yky_k maps to a Doppler frequency:

fD,k=2ωykλ,f_{D,k} = \frac{2\omega y_k}{\lambda},

enabling Doppler-based cross-range imaging. The ISAR image is a 2D map of target reflectivity in the range-Doppler (range-cross-range) plane.

Theorem: ISAR Cross-Range Resolution

For a target rotating at rate ω\omega observed for time TT, the cross-range resolution is:

Δy=λ2Δθ=λ2ωT.\Delta y = \frac{\lambda}{2\Delta\theta} = \frac{\lambda}{2\omega T}.

This is identical to the SAR cross-range result with Δθ\Delta\theta replacing Lsa/R0L_{\text{sa}}/R_0.

The Doppler resolution is δfD=1/T\delta f_D = 1/T. Since cross-range maps to Doppler via fD=2ωy/λf_D = 2\omega y/\lambda, finer Doppler resolution (longer observation) gives finer cross-range resolution.

Proof
Doppler resolution

The Doppler resolution from an observation of duration TT is δfD=1/T\delta f_D = 1/T.

Cross-range mapping

Since fD=2ωy/λf_D = 2\omega y/\lambda, the cross-range resolution is Δy=λδfD/(2ω)=λ/(2ωT)=λ/(2Δθ)\Delta y = \lambda \delta f_D / (2\omega) = \lambda/(2\omega T) = \lambda/(2\Delta\theta). ■\blacksquare

Definition:

ISAR Image Formation Pipeline

ISAR processing follows a pipeline similar to SAR with additional motion compensation:

  1. Range compression: Matched filtering in fast time (identical to SAR).

  2. Translational motion compensation (TMC):

    • Range alignment: Correct the bulk range shift of the target center using cross-correlation of range profiles: Δ^m=arg⁡max⁡Δ∣⟨rm,rm−1(⋅−Δ)⟩∣\hat{\Delta}_m = \arg\max_\Delta |\langle \mathbf{r}_m, \mathbf{r}_{m-1}(\cdot - \Delta)\rangle|.
    • Phase adjustment: Remove the residual phase from translational motion using a prominent scatterer as reference (dominant scatterer algorithm).

  3. Azimuth compression: FFT along slow time (Doppler processing) to form the range-Doppler image.

The cross-range axis of the ISAR image is calibrated using the estimated rotation rate: y=λfD/(2ω)y = \lambda f_D / (2\omega).

ISAR Range-Doppler Image

Simulates ISAR imaging of a rotating ship-like target. The radar is at 10 GHz with 500 MHz bandwidth. Scatterers are placed at hull edges, masts, and superstructure positions.

Adjust the rotation rate ω\omega and observation time TT to see how cross-range resolution changes. Lower rotation rates require longer observation for the same resolution.

The cross-range axis is in meters (calibrated by ω\omega).

Parameters
5
2
20

Example: ISAR Imaging of a Ship

A shore-based radar at f0=10f_0 = 10 GHz (λ=3\lambda = 3 cm) observes a ship rolling at ω=0.05\omega = 0.05 rad/s for T=10T = 10 s with bandwidth B=500B = 500 MHz.

Compute the ISAR image resolution.

Solution
Total rotation angle

Δθ=ωT=0.5\Delta\theta = \omega T = 0.5 rad =28.6°= 28.6°.

Cross-range resolution

Δy=λ/(2Δθ)=0.03/(2×0.5)=3\Delta y = \lambda/(2\Delta\theta) = 0.03/(2 \times 0.5) = 3 cm.

Range resolution

ΔR=c/(2B)=3×108/(2×500×106)=30\Delta R = c/(2B) = 3 \times 10^8/(2 \times 500 \times 10^6) = 30 cm.

Assessment

The 3×303 \times 30 cm resolution resolves individual structural features. However, the ship also translates (heave, surge), requiring range alignment and phase autofocus before Doppler processing.

SAR vs ISAR — Key Differences

Feature SAR ISAR
Moving element Radar platform Target
Aperture control Known (navigation) Unknown (estimated)
Scene Stationary terrain Rotating target
Motion compensation Autofocus (residual errors) Essential (TMC first)
Cross-range axis Azimuth (along-track) Doppler ↔\leftrightarrow cross-range
Application Earth observation, surveillance Ship/aircraft classification

Both produce 2D coherent images from 1D radar returns — the fundamental mechanism is identical: relative motion creates a synthetic aperture.

Definition:

Cross-Range Scaling in ISAR

The ISAR image cross-range axis is in Doppler frequency fDf_D, not physical meters. Converting to meters requires knowledge of the rotation rate ω\omega:

y=λfD2ω.y = \frac{\lambda f_D}{2\omega}.

Since ω\omega is unknown for non-cooperative targets, it must be estimated from the data. Common methods:

  • Two sub-aperture correlation: Split the data into two halves, form images, and estimate the rotation from the relative shift.
  • Minimum entropy over ω\omega: Search for the ω\omega that minimizes image entropy.
  • Known target dimension: If the target's extent is known (e.g., from a database), ω\omega can be calibrated from the Doppler extent.

Common Mistake: The Uniform Rotation Assumption in ISAR

Mistake:

ISAR processing assumes the target rotation is uniform (ω=const\omega = \text{const}) over the observation interval. Real targets exhibit non-uniform rotation, 3D tumbling, and translational acceleration.

Correction:

For non-uniform motion, use time-frequency analysis (short-time Fourier transform, Wigner-Ville distribution) to form time-varying ISAR images, or apply iterative autofocus adapted from SAR processing. The sparsity-based joint autofocus approach of Section s05 handles non-uniform rotation naturally.

Why This Matters: ISAR for Non-Cooperative Target Recognition

ISAR is a critical technology for maritime and air surveillance. Shore-based or shipborne radars image vessels at ranges of tens of kilometers, producing range-cross-range maps that reveal the ship's silhouette, mast positions, and superstructure — features used for automatic target recognition (ATR).

In the ISAC context ([?ch34]), ISAR imaging can be performed using OFDM communication waveforms: the OFDM subcarriers provide range resolution (bandwidth), while target rotation provides cross-range resolution.

See full treatment in Chapter 34، Section 2

Historical Note: From Turntable Radar to ISAR

1960s–2000s

The ISAR concept originated from turntable radar measurements in the 1960s, where targets were placed on a rotating platform and imaged by a stationary radar. The insight that natural target motion (ship roll, aircraft maneuvers) could serve the same purpose led to operational ISAR systems in the 1980s. Today, ISAR is standard equipment on naval vessels and coastal surveillance radars worldwide. The 2000s saw a resurgence of interest driven by space situational awareness — ISAR imaging of satellites and debris for orbital characterization.

Quick Check

An ISAR radar observes a target rotating at ω=0.1\omega = 0.1 rad/s for T=5T = 5 s at λ=3\lambda = 3 cm. The cross-range resolution is:

3 cm

30 cm

15 cm

6 cm

Correction:
3 cm

Δy=λ/(2ωT)=0.03/(2×0.1×5)=0.03\Delta y = \lambda/(2\omega T) = 0.03/(2 \times 0.1 \times 5) = 0.03 m = 3 cm.

Key Takeaway

ISAR reverses SAR: a stationary radar images a rotating target. Cross-range resolution Δy=λ/(2Δθ)\Delta y = \lambda/(2\Delta\theta) depends on the total rotation angle, which is unknown and must be estimated. Translational motion compensation is mandatory before Doppler processing. The same y=Ac+w\mathbf{y} = \mathbf{A}\mathbf{c} + \mathbf{w} model applies, with A\mathbf{A} now encoding the unknown target rotation rather than the known platform trajectory.

SAR Image Formation AlgorithmsTomographic SAR