Ferkans — Interactive Telecom Tutor

The Golden Thread: SAR as $\mathbf{y} = \mathbf{A}\mathbf{c} + \mathbf{w}$

Every imaging system in this book reduces to the same linear model: $\mathbf{y} = \mathbf{A}\mathbf{c} + \mathbf{w}$ . In SAR, the sensing matrix $\mathbf{A}$ is built from the platform trajectory and the transmitted waveform. Platform motion creates the large synthetic aperture that gives SAR its extraordinary cross-range resolution — the same physics exploited by the multi-view sensing geometries of Chapter 8, but now with a single moving antenna.

Definition:
The Synthetic Aperture Concept

A radar mounted on a platform moving at velocity $v_p$ along the $x$ -axis transmits pulses at positions $x_n = v_p \eta_n$ , $n = 0, 1, \ldots, N_a - 1$ , where $\eta_n$ are the slow-time (azimuth) sample instants. The synthetic aperture length is:

$L_{\text{sa}} = v_p \cdot T_{\text{sa}},$

where $T_{\text{sa}}$ is the total observation (dwell) time.

By coherently combining the returns from all $N_a$ pulses, the system behaves as if it had a physical antenna of length $L_{\text{sa}}$ — typically tens to hundreds of meters for airborne systems and kilometers for spaceborne systems.

Definition:
Range Resolution in SAR

Range resolution is determined by the transmitted bandwidth $B$ , identically to conventional pulsed radar (Chapter 7):

$\Delta R = \frac{c}{2B}.$

SAR systems use linear frequency modulation (LFM) chirp pulses with large time-bandwidth products $BT_p \gg 1$ , achieving range resolutions from centimeters (wide-bandwidth systems) to meters (spaceborne systems with limited bandwidth).

Theorem: SAR Cross-Range Resolution

For a stripmap SAR with synthetic aperture length $L_{\text{sa}}$ , carrier wavelength $\lambda$ , and broadside range $R_0$ , the cross-range (azimuth) resolution is:

$\Delta x = \frac{\lambda R_0}{2 L_{\text{sa}}}.$

The best achievable cross-range resolution, attained when the target is illuminated for the maximum possible time, is:

$\boxed{\Delta x_{\min} = \frac{D}{2}},$

where $D$ is the physical antenna length. This result is independent of range and wavelength.

A smaller antenna has a wider beam, illuminating the target longer and creating a longer synthetic aperture. The gain from longer coherent integration exactly compensates for the smaller physical aperture — yielding the "resolution paradox" where a smaller antenna gives better cross-range resolution.

Proof

Step 1: Synthetic aperture beamwidth

The synthetic aperture of length $L_{\text{sa}}$ produces an effective beamwidth $\theta_{\text{sa}} = \lambda/(2L_{\text{sa}})$ (the factor of 2 arises from the round-trip phase). The corresponding cross-range resolution at range $R_0$ is $\Delta x = R_0 \theta_{\text{sa}} = \lambda R_0/(2L_{\text{sa}})$ .

Step 2: Maximum aperture from beam footprint

The real antenna beamwidth $\theta_{3\text{dB}} = \lambda/D$ determines how long a target at range $R_0$ stays illuminated. The maximum synthetic aperture is $L_{\text{sa,max}} = R_0 \theta_{3\text{dB}} = R_0\lambda/D$ .

Step 3: Best achievable resolution

Substituting $L_{\text{sa,max}}$ into the resolution formula: $\Delta x_{\min} = \lambda R_0 / (2 \cdot R_0\lambda/D) = D/2$ . $\blacksquare$

The SAR Resolution Paradox

The result $\Delta x_{\min} = D/2$ appears paradoxical: a smaller antenna gives better resolution. The explanation is that a smaller antenna has a wider beam, illuminating the target for a longer time and creating a longer synthetic aperture.

In practice, resolution is limited by motion errors and signal-to-noise ratio rather than by antenna size. This is why autofocus (Section s02) is so critical for real SAR systems.

Definition:
SAR Operating Modes

SAR systems operate in several modes that trade resolution for swath width and coverage rate:

Stripmap SAR: The antenna beam is fixed broadside. The platform motion sweeps the beam along the ground, producing a continuous strip image. Cross-range resolution: $\Delta x = D/2$ .

Spotlight SAR: The antenna is steered to keep the beam on a fixed patch of ground for an extended dwell time, creating a longer synthetic aperture. Resolution: $\Delta x = D/2$ if the full beamwidth is used, but spotlight can exceed this by collecting data over $> 1$ beamwidth via electronic steering.

ScanSAR (Wide-swath): The antenna beam is periodically switched between multiple swaths, trading reduced dwell time per swath for wider total coverage. Cross-range resolution degrades by the number of sub-swaths (typically 3--5 $\times$ ).

SAR Operating Modes Comparison

Mode	Cross-Range Resolution	Swath Width	Application
Stripmap	$D/2$	Medium (single beam)	Standard mapping, surveillance
Spotlight	$< D/2$ (steered dwell)	Small (fixed patch)	High-resolution target imaging
ScanSAR	$\sim 3$ -- $5 \times D/2$	Wide (multiple sub-swaths)	Ocean monitoring, wide-area mapping

Example: Resolution of ESA Sentinel-1

Compute the resolution of ESA's Sentinel-1 SAR satellite in Interferometric Wide Swath (IW) mode:

Carrier frequency: $f_0 = 5.405$ GHz ( $\lambda = 5.55$ cm).
Antenna length: $D = 12.3$ m.
Bandwidth: $B = 56.5$ MHz.
Orbit altitude: $H = 693$ km, slant range $R_0 \approx 800$ km.

Solution

Range resolution

$\Delta R = c/(2B) = 3 \times 10^8 / (2 \times 56.5 \times 10^6) = 2.65$ m.

Best cross-range resolution (stripmap)

$\Delta x_{\min} = D/2 = 12.3/2 = 6.15$ m.

Real-aperture comparison

Without SAR: $\Delta x_{\text{real}} = \lambda R_0 / D = 0.0555 \times 800{,}000 / 12.3 = 3{,}610$ m. SAR improves cross-range resolution by a factor of $\sim$ 587.

SAR Resolution vs Bandwidth and Aperture Length

Explore how SAR resolution depends on system parameters.

Left panel: Range resolution $\Delta R = c/(2B)$ (dashed, constant with range) and cross-range resolution $\Delta x = \lambda R/(2L_{\text{sa}})$ (solid, increasing with range).

Right panel: Resolution cell area $\Delta R \times \Delta x$ .

Switch between stripmap, spotlight, and ScanSAR modes to see how the resolution trade-off changes.

Parameters

Bandwidth

B

(MHz)200

Synthetic aperture

L_{\text{sa}}

(m)50

Carrier frequency (GHz)10

SAR mode

Theorem: Degrees of Freedom in SAR Imaging

For a stripmap SAR system with range swath $W_R$ and azimuth swath $W_x$ , the number of independent resolution cells (degrees of freedom) is:

$N_{\text{DOF}} = \frac{W_R}{\Delta R} \times \frac{W_x}{\Delta x} = \frac{2B W_R}{c} \times \frac{2 L_{\text{sa}} W_x}{\lambda R_0}.$

This equals the time-bandwidth product in range times the space-bandwidth product in azimuth.

Proof

Range DOF

The range DOF follows from Shannon sampling at bandwidth $B$ : $N_R = W_R / \Delta R = 2BW_R/c$ .

Azimuth DOF

The synthetic aperture of length $L_{\text{sa}}$ at range $R_0$ subtends an angular bandwidth $\Delta\theta = L_{\text{sa}}/R_0$ . By the Fourier uncertainty principle, $N_x = (2/\lambda)\Delta\theta \cdot W_x = 2L_{\text{sa}}W_x/(\lambda R_0)$ .

Total DOF

$N_{\text{DOF}} = N_R \times N_x$ . $\blacksquare$

SAR as a Special Case of the Imaging Model

The SAR measurement model fits directly into the framework of Chapter 8:

$\mathbf{y} = \mathbf{A}_{\text{SAR}} \, \mathbf{c} + \mathbf{w},$

where $\mathbf{A}_{\text{SAR}}$ has Kronecker structure:

$\mathbf{A}_{\text{SAR}} = \mathbf{A}_{\text{az}} \otimes \mathbf{A}_{\text{rg}},$

with $\mathbf{A}_{\text{rg}}$ encoding the range (frequency) dimension and $\mathbf{A}_{\text{az}}$ the azimuth (slow-time position) dimension. This Kronecker factorization enables efficient computation of both the forward operator and its adjoint — the classical RDA is nothing but $\mathbf{A}^{H} \mathbf{y}$ applied via the Kronecker factors.

Synthetic Aperture

An effectively large antenna aperture created by coherently combining radar returns collected at successive positions along the platform trajectory. The synthetic aperture length $L_{\text{sa}} = v_p T_{\text{sa}}$ determines the achievable cross-range resolution.

Cross-Range Resolution

The ability to distinguish targets separated in the direction perpendicular to the radar line of sight. In SAR, $\Delta x = \lambda R/(2L_{\text{sa}})$ ; the best achievable value is $D/2$ , half the physical antenna length.

Stripmap SAR

A SAR mode where the antenna beam points broadside (perpendicular to the flight path), producing a continuous strip image with cross-range resolution $D/2$ .

Spotlight SAR

A SAR mode where the antenna beam is steered to illuminate a fixed ground patch for extended dwell time, achieving finer cross-range resolution at the cost of reduced area coverage.

Historical Note: The Invention of SAR

1951–1970s

Synthetic aperture radar was invented independently by Carl Wiley at Goodyear Aircraft Corporation (1951) and by a team at the University of Illinois. Wiley's insight was that the Doppler history of a target illuminated by a side-looking radar contains the same information as a large physical antenna. The first airborne SAR images were produced in the mid-1950s using optical processing — the coherent combination was performed by recording the radar signal on film and processing it with a coherent optical system. Digital SAR processing became practical in the 1970s with the advent of FFT hardware.

Quick Check

A SAR system has antenna length $D = 10$ m. If the antenna is replaced with one of length $D = 2$ m (same carrier frequency), the best achievable cross-range resolution:

Worsens by a factor of 5

Improves by a factor of 5

Stays the same

Depends on the range

Correction:

Improves by a factor of 5

$\Delta x_{\min} = D/2$ . With $D = 2$ m, $\Delta x_{\min} = 1$ m versus 5 m for $D = 10$ m. The smaller antenna has a wider beam, illuminating the target longer and creating a longer synthetic aperture.

Key Takeaway

SAR creates a large virtual antenna through platform motion. Range resolution is $\Delta R = c/(2B)$ from bandwidth; cross-range resolution is $\Delta x = \lambda R/(2L_{\text{sa}})$ from the synthetic aperture. The remarkable result $\Delta x_{\min} = D/2$ means a smaller antenna paradoxically gives better SAR resolution. The entire SAR measurement fits the model $\mathbf{y} = \mathbf{A}\mathbf{c} + \mathbf{w}$ with Kronecker-structured $\mathbf{A}$ .

SAR Geometry and Resolution

The Golden Thread: SAR as y=Ac+w\mathbf{y} = \mathbf{A}\mathbf{c} + \mathbf{w}y=Ac+w

Definition: The Synthetic Aperture Concept

Definition: Range Resolution in SAR

Theorem: SAR Cross-Range Resolution

Step 1: Synthetic aperture beamwidth

Step 2: Maximum aperture from beam footprint

Step 3: Best achievable resolution

The SAR Resolution Paradox

Definition: SAR Operating Modes

SAR Operating Modes Comparison

Example: Resolution of ESA Sentinel-1

Range resolution

Best cross-range resolution (stripmap)

Real-aperture comparison

SAR Resolution vs Bandwidth and Aperture Length

Parameters

Theorem: Degrees of Freedom in SAR Imaging

Range DOF

Azimuth DOF

Total DOF

SAR as a Special Case of the Imaging Model

Synthetic Aperture

Cross-Range Resolution

Stripmap SAR

Spotlight SAR

Historical Note: The Invention of SAR

Quick Check

Key Takeaway

The Golden Thread: SAR as $\mathbf{y} = \mathbf{A}\mathbf{c} + \mathbf{w}$

Definition:
The Synthetic Aperture Concept

Definition:
Range Resolution in SAR

Definition:
SAR Operating Modes