Computational Antenna Analysis and Optimization

Introduction and Context

Modern antenna engineering is increasingly a computational discipline. Although analytical solutions remain essential for canonical structures such as infinitesimal dipoles, circular apertures, and uniform arrays, most practical antennas contain finite conductors, dielectric substrates, feeding networks, radomes, connectors, mounting platforms, and manufacturing details that make closed-form analysis impossible. Computational electromagnetic (CEM) methods provide numerical approximations to Maxwell’s equations and allow engineers to predict, visualize, and optimize antenna behavior before fabrication.

Computational antenna analysis is not merely the use of simulation software. It involves selecting an appropriate electromagnetic model, defining physically meaningful boundary conditions and material properties, discretizing the problem with controlled error, validating the numerical solution, and interpreting quantities such as impedance, gain, efficiency, radiation pattern, bandwidth, and specific absorption rate. Optimization adds a further layer: design variables are systematically changed to achieve specified performance objectives while satisfying physical, manufacturing, and regulatory constraints.

At postgraduate level, the subject requires both mathematical understanding and engineering judgment. A simulation result is credible only when its assumptions, numerical convergence, and agreement with theory or measurement have been examined.

Learning Objectives

After studying this chapter, the reader should be able to:

1. Explain the role of computational electromagnetics in antenna design.
2. Distinguish the principal numerical methods used for antenna analysis.
3. Formulate antenna geometries, materials, sources, boundaries, and observables correctly.
4. Understand meshing, convergence, numerical error, and model validation.
5. Interpret computational results including S-parameters, impedance, efficiency, gain, and fields.
6. Formulate single-objective and multi-objective antenna optimization problems.
7. Select appropriate gradient-based, evolutionary, surrogate-assisted, or topology-optimization methods.
8. Apply a rigorous computational workflow to a practical antenna-design problem.
9. Recognize limitations, reproducibility requirements, and responsible use of simulation and AI tools.

1. Electromagnetic Basis of Computational Analysis

A computational antenna solver approximates Maxwell’s equations in a specified region. In differential form, the time-domain equations are

Here E is electric field intensity in V/m, H is magnetic field intensity in A/m, D is electric flux density in C/m², B is magnetic flux density in Wb/m², J is current density in A/m², and ρ is charge density in C/m³. For linear isotropic materials,

where ε is permittivity in F/m, μ is permeability in H/m, and σ is conductivity in S/m. In an isotropic but lossy dielectric, ε may be represented by a complex permittivity. Using the phasor convention e^jωt, Maxwell’s curl equations become

where ω = 2πf is angular frequency in rad/s, f is frequency in Hz, and J_s represents an impressed source current.

In a homogeneous source-free region, the electric field satisfies the vector Helmholtz equation

under the appropriate divergence and boundary conditions, where k = ω√(με) is the wavenumber in a lossless medium. In free space, k₀ = 2π/λ₀ and λ₀ = c/f, with c approximately 3 × 10⁸ m/s.

The computational problem is therefore a boundary-value or initial-boundary-value problem. Geometry, excitation, constitutive properties, and truncation conditions define the problem; the numerical method converts it into a finite algebraic system.

2. Principal Computational Methods

2.1 Method of Moments

The Method of Moments (MoM) is primarily an integral-equation technique. It is especially effective for conducting wires, metallic surfaces, and open-region radiation problems. An unknown current is expanded in basis functions:

where fₙ(r) are known basis functions and Iₙ are unknown coefficients. Testing the integral equation with weighting functions produces a matrix system

where Z is the impedance matrix, I is the vector of unknown current coefficients, and V is the excitation vector. Once I is found, near fields, far fields, input impedance, and radiated power can be calculated.

MoM has the advantage that radiation conditions are naturally included through Green’s functions, so an infinite free-space region need not be discretized. Its disadvantages include dense matrices, memory requirements that can grow approximately with the square of the number of unknowns, and difficulty in handling strongly inhomogeneous volumetric dielectrics unless specialized formulations are used.

2.2 Finite-Difference Time-Domain Method

The Finite-Difference Time-Domain (FDTD) method discretizes Maxwell’s equations in space and time. Electric and magnetic field components are placed on a staggered Yee grid. Time-domain excitation produces a broadband response in one simulation, after which Fourier transformation yields frequency-domain quantities.

For a one-dimensional illustrative case, a stability condition has the form

where ∆t is the time step, ∆x is the smallest spatial cell dimension, and v is wave velocity. In three-dimensional Cartesian grids, the Courant condition is more restrictive; for unequal cell dimensions it is commonly written

for an explicit standard Yee scheme. The inequality is a numerical stability condition, not a guarantee of accuracy. Accuracy also requires sufficient spatial resolution, generally related to the shortest wavelength and material features.

FDTD is well suited to broadband, transient, and electrically moderate-size problems. Staircasing errors can occur for curved conductors and oblique boundaries. Memory consumption may become severe when a fine grid is needed around small geometrical features.

2.3 Finite Element Method

The Finite Element Method (FEM) divides the computational domain into elements, often tetrahedra in three dimensions, and approximates the field using local shape functions. A weak or variational form of Maxwell’s equations is constructed and assembled into a sparse system. FEM is particularly powerful for complex geometries, anisotropic materials, dielectric loading, and local mesh refinement.

For a frequency-domain electric-field formulation, a simplified matrix representation is

where K represents curl-curl or stiffness terms, M is a mass matrix, C represents conductive or other loss terms, E is the vector of unknown field coefficients, and F is the source vector. Exact matrix definitions depend on the formulation, element type, material model, and boundary conditions.

Open-region antenna problems require an absorbing boundary, perfectly matched layer (PML), or coupling to an exterior-region formulation. Poorly designed absorbing boundaries can reflect energy back into the computational domain.

2.4 Finite Integration and Hybrid Methods

The finite integration technique and related discrete methods preserve integral forms of Maxwell’s equations on paired grids. Hybrid methods combine complementary strengths, such as FEM for complex near-field geometry and MoM or physical optics for open-region radiation. Domain decomposition can reduce memory requirements and enable parallel computation.

No method is universally best. The selection depends on frequency range, electrical size, geometry, material complexity, required bandwidth, and available computational resources.

3. Building a Trustworthy Computational Model

3.1 Geometry and Parameterization

The geometry must represent the electromagnetic behavior relevant to the design objective. Small mechanical details may be omitted when they are electrically insignificant, but a feed gap, via, shorting pin, connector, radome, or mounting bracket can strongly affect resonance and matching.

A parameterized model expresses geometry through a vector

where xᵢ may represent length, width, slot position, substrate thickness, flare angle, or array spacing. Good parameterization reduces the search space and ensures that generated designs remain manufacturable. Parameters should be physically interpretable whenever possible.

3.2 Materials

The solver requires ε, μ, and σ, potentially as functions of frequency, temperature, and direction. Substrate datasheets often report nominal relative permittivity and loss tangent, but actual values vary with batch, frequency, resin content, moisture, and fabrication process. For a dielectric with relative permittivity εᵣ and loss tangent tanδ, dielectric loss is represented approximately by

where ε′ and ε″ are the real and loss-related components under the chosen phasor convention. Material uncertainty should be included in sensitivity studies when performance margins are small.

3.3 Excitation and Ports

A port must reproduce the physical feeding mechanism. Common models include wave ports, lumped ports, coaxial feeds, voltage gaps, and impressed current sources. Incorrect port placement or an unrealistic reference plane may produce a plausible but physically misleading S-parameter.

For a one-port antenna, the reflection coefficient is

where Z_in is input impedance and Z₀ is the reference impedance, commonly 50 Ω. The return loss is often defined as

and the voltage standing-wave ratio is

These expressions assume a consistent reference impedance and a linear time-invariant port model.

3.4 Boundaries and Radiation

Because an antenna radiates into an unbounded region, a finite computational domain must emulate open space. Common approaches are absorbing boundary conditions, PMLs, and integral-equation exterior formulations. The outer boundary should be sufficiently separated from the radiating structure or appropriately transformed so that artificial reflections do not contaminate the antenna response.

3.5 Meshing and Convergence

Discretization error is controlled by mesh density, element quality, basis order, and local refinement. A useful engineering principle is to resolve the shortest wavelength in the material, not merely the free-space wavelength. In a dielectric with relative permittivity εᵣ, the approximate wavelength is

for a homogeneous, nondispersive, nonmagnetic medium.

Mesh refinement should be concentrated near narrow gaps, sharp edges, thin layers, ports, vias, and regions with strong field gradients. A convergence study compares a quantity of interest across increasingly refined meshes. If Q₁ and Q₂ are results from successive meshes, a relative change may be reported as

The quantity Q may be resonant frequency, input impedance, gain, or another design metric. A single converged quantity does not prove that all quantities are converged. For example, resonant frequency may stabilize while peak local electric field remains mesh-sensitive.

4. Antenna Quantities Extracted from Simulation

4.1 Input Impedance and S-Parameters

The input impedance is

where R_in is resistance and X_in is reactance, both in ohms. At resonance, X_in is approximately zero, although practical antennas may have multiple resonances and frequency-dependent loss.

For a multiport structure, the scattering relation is

where a and b are incident and outgoing wave-amplitude vectors and S is the scattering matrix. S₁₁ measures input reflection at port 1 when other ports are terminated according to the reference condition.

4.2 Power, Efficiency, Gain, and Directivity

Radiation efficiency is

where P_rad is radiated power and P_acc is accepted power. Total efficiency may include mismatch:

where P_inc is incident power at the port, assuming a single-port antenna and consistent power-wave definitions.

Directivity is

where U is radiation intensity in W/sr. Gain is

and realized gain includes mismatch under the stated convention:

The distinction is important: a design can have high directivity but poor realized gain because of mismatch or loss.

4.3 Far-Field and Near-Field Results

The far-field approximation is valid when the observation distance r is sufficiently large relative to the antenna maximum dimension D and wavelength. A common criterion is

with additional practical separation requirements. Near fields are essential for evaluating coupling, electromagnetic compatibility, biological exposure, and placement on platforms. Far-field patterns should be checked for coordinate conventions, polarization basis, normalization, and angular sampling.

5. Formulating Antenna Optimization

Analysis predicts performance for a specified design; optimization searches for a design that satisfies objectives. Let x denote design variables, p denote uncertain or fixed parameters, and gᵢ and hⱼ denote inequality and equality constraints. A general constrained problem is

The objective F may combine several normalized measures. For example,

where wᵢ are nonnegative weights. Normalization is necessary because gain, dimensions, and reflection coefficient have different scales and units. A poor choice of weights can cause one objective to dominate the optimization.

Typical constraints include:

• frequency range for |S₁₁| below a specified threshold;
• minimum realized gain or efficiency;
• maximum physical dimensions and mass;
• minimum trace width and fabrication clearance;
• maximum SAR or near-field intensity;
• symmetry, polarization, and pattern requirements;
• thermal, structural, or regulatory limits.

5.1 Single-Objective and Pareto Optimization

A weighted sum converts several objectives into one scalar objective, but it may hide trade-offs. Multi-objective optimization instead seeks Pareto-optimal designs: a design is Pareto-optimal if no objective can be improved without worsening at least one other objective. The resulting Pareto front helps designers choose among bandwidth, gain, size, efficiency, cost, and robustness.

5.2 Gradient-Based Optimization

If the solver is differentiable or adjoint sensitivities are available, gradient-based methods can be highly efficient. A generic update is

where αₖ is a step size and ∇F is the gradient with respect to design variables. These methods are effective for smooth continuous variables but can become trapped in local minima and are less convenient for discrete topology changes.

The adjoint method obtains the gradient of many objectives with a computational cost that is often comparable to a small number of full simulations, largely independent of the number of design variables. This is valuable in shape and topology optimization.

5.3 Evolutionary and Population-Based Methods

Genetic algorithms, differential evolution, particle swarm optimization, and covariance-adaptation methods do not require gradients. They can explore discontinuous, multimodal, and mixed-variable design spaces. Their disadvantages are many solver evaluations, sensitivity to parameter settings, and difficulty guaranteeing a global optimum.

5.4 Surrogate-Assisted Optimization

When a full-wave simulation is expensive, a surrogate model approximates the response. Given training data {xᵢ, yᵢ}, the surrogate predicts y at untested designs. Gaussian-process models, radial-basis functions, polynomial response surfaces, neural networks, and support-vector regression are common choices.

A surrogate must not be treated as automatically reliable outside its training domain. Cross-validation, adaptive sampling, and final full-wave verification are essential. Active learning can select new simulations where predicted uncertainty or expected improvement is high.

5.5 Robust Optimization and Uncertainty

A nominal optimum may fail after fabrication or under environmental variation. Let p be a random vector of uncertain parameters. Robust design may minimize expected loss and variability:

where E denotes expectation, Var denotes variance, and β controls the penalty on variability. Worst-case formulations instead optimize performance over a defined uncertainty set. Robustness generally costs some nominal performance but improves production yield.

Figure: General closed-loop workflow for computational antenna optimization.

6. A Rigorous Design Workflow

A defensible workflow contains the following stages:

1. Translate requirements into metrics. Define frequency bands, allowable reflection coefficient, gain, polarization, scan range, size, power, and environmental conditions.
2. Select a parameterization. Use a compact design vector with meaningful bounds and fabrication rules.
3. Choose the computational method. Consider electrical size, materials, bandwidth, and available resources.
4. Build and inspect the baseline model. Check units, ports, symmetry, coordinate systems, material data, and boundary placement.
5. Perform mesh and solver convergence studies. Record tolerances, mesh statistics, and stopping criteria.
6. Validate against theory or measurement. Use canonical benchmarks, circuit estimates, energy balance, or a prototype.
7. Select the optimizer. Use gradients for smooth high-dimensional problems, evolutionary methods for discontinuous spaces, and surrogates for expensive simulations.
8. Monitor feasibility during optimization. Penalize or reject designs that violate manufacturing, physical, or regulatory constraints.
9. Re-simulate the final design independently. Use a refined mesh, alternative solver settings, and preferably a second formulation or measurement.
10. Report uncertainty and reproducibility information. Archive geometry, material data, mesh rules, solver version, scripts, random seeds, and objective definitions.

7. Worked Example: Resonance Adjustment of a Printed Monopole

Consider an illustrative quarter-wave printed monopole intended for 2.45 GHz on a substrate with relative permittivity εᵣ = 4.0. A first estimate of guided wavelength is

Using c = 3 × 10⁸ m/s and f = 2.45 × 10⁹ Hz,

A simple quarter-wave estimate is therefore

This is only an initial estimate. The actual resonant length depends on fringing fields, conductor width, substrate thickness, ground-plane dimensions, feed geometry, and effective rather than bulk permittivity.

Suppose a full-wave simulation with L = 15.3 mm gives a resonance at 2.32 GHz. A local linear estimate assumes that frequency is approximately inversely proportional to length over a small range:

Thus,

A new simulation is necessary; the estimate is not a substitute for solving Maxwell’s equations. If the revised design resonates at 2.43 GHz, a further local correction gives

The designer should then evaluate not only resonance, but also |S₁₁| across the required band, realized gain, efficiency, radiation pattern, current distribution, and sensitivity to substrate thickness and etching tolerance. A design that achieves 2.45 GHz but has inadequate efficiency or an unstable pattern is not a successful optimization.

8. Applied Case: Robust Wearable Antenna Optimization

Background

The following is an illustrative postgraduate design case. A wearable dual-band antenna is required for a body-area wireless device operating near 2.45 GHz and 5.8 GHz. The antenna must fit within 35 mm × 35 mm, use a flexible substrate, and maintain acceptable performance when separated from or placed near the human body.

Problem

A nominal simulation shows satisfactory matching in free space but a large frequency shift and reduced efficiency when the antenna is placed over a multilayer tissue model. The design is also sensitive to bending radius and substrate thickness.

Analysis

The design vector includes patch length, slot length, slot width, feed position, ground-plane length, and spacing from the body. The objectives are to minimize a weighted matching error at both bands, maximize realized gain, and reduce sensitivity to uncertain spacing and bending. Constraints include the package dimensions, minimum manufacturable slot width, and a maximum exposure metric specified by the applicable safety standard.

A direct evolutionary search would require thousands of full-wave simulations. Instead, an initial design of experiments samples the bounded design space. A Gaussian-process surrogate is trained for resonance frequency, efficiency, and realized gain. Cross-validation identifies regions where prediction error is high; additional full-wave simulations are then placed in those regions.

Approach and Solution

The optimization proceeds in three stages. First, a coarse mesh and simplified tissue model identify promising geometries. Second, the best candidates are re-evaluated with dispersive multilayer tissue properties, refined meshing, and several bending radii. Third, a robust objective penalizes designs whose performance varies excessively across the uncertainty set.

The final candidate is not selected solely by the lowest nominal objective. A slightly less efficient design is preferred because its resonant frequencies vary less with spacing and bending. The final full-wave model uses a refined mesh and an independent boundary configuration. Fabricated prototypes are measured using a calibrated vector network analyzer, and the antenna is tested on a representative phantom.

Outcome

The principal outcome is a design with acceptable dual-band matching and lower performance spread than the nominally optimized alternative. Measurement reveals a small frequency offset caused by substrate permittivity and fabrication tolerance. Because uncertainty had been included in the workflow, the measured result remains within the design margin.

Learning Points

1. The best nominal simulation is not always the best engineering design.
2. Human-body loading must be modeled with frequency-dependent and position-dependent properties.
3. Surrogate models reduce computational cost but require validation.
4. Manufacturing and measurement uncertainties should be considered before prototype fabrication.
5. Exposure and privacy requirements are part of responsible wearable-antenna engineering.

9. Verification, Validation, and Error Sources

Verification asks whether the equations were solved correctly; validation asks whether the model represents the real physical system. Important error sources include:

• inadequate mesh density or poor element quality;
• incorrect port calibration or reference-plane definition;
• insufficient absorbing-region performance;
• inaccurate or frequency-independent material data;
• omitted connectors, cables, enclosures, or mounting structures;
• inconsistent far-field normalization;
• inappropriate symmetry assumptions;
• solver tolerances that are too loose;
• optimization overfitting to numerical artifacts.

Energy conservation provides one useful diagnostic. For a lossless one-port model, accepted power should equal radiated power apart from numerical error. For lossy structures, accepted power should be partitioned among radiated, conductor-loss, dielectric-loss, and other modeled channels. Large unexplained discrepancies indicate a modeling or post-processing problem.

Validation should use at least one independent line of evidence: analytical benchmark, alternative numerical method, circuit model, or measurement. Agreement at only one frequency is weak evidence; trends across frequency, geometry, and excitation are more informative.

10. Contemporary Developments and Responsible Use of AI

High-performance computing, GPU acceleration, model-order reduction, differentiable solvers, and cloud-based workflows are expanding the size of feasible antenna studies. Digital twins can combine electromagnetic models with thermal, structural, and manufacturing models. Topology optimization can generate nonintuitive metallic or dielectric distributions, while additive manufacturing enables complex three-dimensional radiators.

Machine learning is useful for surrogate modeling, inverse design, defect detection, and rapid tolerance analysis. However, a neural network trained on narrow data may extrapolate incorrectly, violate physical constraints, or conceal uncertainty. Physics-informed learning, dimensional analysis, passivity checks, symmetry constraints, and final full-wave verification improve reliability.

Responsible computational practice includes documenting software versions and licenses, protecting proprietary geometries and measurement data, reporting limitations, and avoiding claims of experimental performance based solely on unverified simulation. Designs must also comply with spectrum regulations, electromagnetic exposure requirements, product safety rules, and export or security controls where applicable.

11. Best Practices

• Start with analytical estimates and benchmark structures.
• Keep units explicit and consistent throughout the model.
• Use parameterized geometry rather than manual editing.
• Separate design variables from uncertain parameters.
• Refine the mesh where fields vary rapidly.
• Test mesh, boundary, and solver convergence independently.
• Normalize multi-objective functions carefully.
• Enforce manufacturability during, not after, optimization.
• Prefer robust performance over fragile nominal optima.
• Verify optimized designs with independent settings or tools.
• Compare simulation with calibrated measurement whenever feasible.
• Archive scripts, meshes, material data, and random seeds for reproducibility.

Chapter Summary

Computational antenna analysis converts Maxwell’s equations, material laws, excitations, and boundary conditions into numerical predictions of currents, fields, impedance, efficiency, gain, and radiation patterns. The Method of Moments, FDTD, FEM, and hybrid approaches have different computational strengths and limitations. Trustworthy results depend on accurate geometry, material characterization, port definition, radiation treatment, mesh convergence, and validation.

Optimization formulates antenna design as a constrained search over geometry, materials, and feeds. Gradient methods are efficient for smooth differentiable problems; evolutionary methods explore discontinuous spaces; surrogate and adjoint methods reduce the cost of repeated full-wave analysis. Multi-objective and robust optimization expose the trade-offs among bandwidth, gain, efficiency, size, manufacturability, and tolerance.

The central professional principle is that optimization does not remove the need for engineering judgment. A credible final design must be numerically converged, physically validated, manufacturable, compliant with relevant standards, and appropriately documented.

Key Terms

Computational electromagnetics: Numerical solution of electromagnetic field problems.

Method of Moments: Integral-equation method that converts unknown current distributions into a matrix equation.

FDTD: Time-domain finite-difference method using a spatial and temporal grid.

FEM: Element-based numerical method using local basis functions and a weak formulation.

Perfectly matched layer: Artificial absorbing region designed to minimize reflections.

Mesh convergence: Stabilization of a computed quantity as discretization is refined.

Port: Mathematical or physical representation of an antenna feed or terminal.

S-parameter: Scattering parameter describing incident and reflected traveling waves.

Reflection coefficient: Ratio of reflected to incident wave amplitude at a port.

Radiation efficiency: Ratio of radiated power to accepted power.

Directivity: Radiation concentration relative to an isotropic radiator.

Realized gain: Gain including mismatch loss under a stated convention.

Design variable: Adjustable parameter controlled by an optimization algorithm.

Constraint: Requirement limiting allowable designs or responses.

Pareto front: Set of nondominated solutions in a multi-objective problem.

Surrogate model: Approximation of an expensive simulation response.

Adjoint method: Sensitivity technique that efficiently computes gradients of objectives.

Robust optimization: Optimization accounting for uncertainty and performance variation.

Verification: Demonstration that numerical equations were solved correctly.

Validation: Demonstration that the model represents the physical system adequately.

Review Questions

Short Answer Questions

1. Why are analytical antenna solutions insufficient for many practical structures?
2. State two advantages and two limitations of the Method of Moments.
3. What is the purpose of a perfectly matched layer?
4. Distinguish radiation efficiency from realized gain.
5. Why is mesh convergence not equivalent to physical validation?

Long Answer Questions

1. Compare MoM, FDTD, and FEM for a broadband antenna containing metallic conductors and lossy dielectric materials.
2. Explain how geometry, materials, ports, boundaries, and meshing jointly determine the credibility of a computational antenna model.
3. Derive and interpret the relationship among reflection coefficient, VSWR, and return loss for a one-port antenna.
4. Discuss gradient-based, evolutionary, surrogate-assisted, and adjoint optimization methods, including their appropriate application domains.
5. Explain how uncertainty and robust optimization should be incorporated into the design of a wearable antenna.

Analytical / Application Questions

1. A one-port antenna has Z_in = 25 + j10 Ω with a 50 Ω reference impedance. Calculate the magnitude of the reflection coefficient and determine whether the antenna satisfies a requirement of |Γ| ≤ 0.333.
2. Design a convergence study for an antenna whose resonant frequency changes from 2.41 GHz to 2.44 GHz after mesh refinement. Specify what additional quantities should be checked before accepting the result.
3. Formulate a multi-objective optimization problem for a 5.8 GHz compact antenna requiring minimum size, at least 6 dBi realized gain, and |S₁1| ≤ −10 dB over a specified bandwidth. Identify at least four design variables and three practical constraints.

Multiple Choice Questions with Answers

Question 1

Which computational method is especially natural for thin conducting surfaces in open-region radiation problems?

A. Method of Moments

B. Finite-Difference Time-Domain only

C. Finite-volume thermal analysis

D. Lumped circuit analysis

Correct Answer: A. Method of Moments

Question 2

For a one-port antenna, which expression gives the reflection coefficient relative to reference impedance Z₀?

A. Γ = (Z_in + Z₀)/(Z_in − Z₀)

B. Γ = (Z_in − Z₀)/(Z_in + Z₀)

C. Γ = Z_in Z₀

D. Γ = Z₀/Z_in

Correct Answer: B. Γ = (Z_in − Z₀)/(Z_in + Z₀)

Question 3

What is the primary purpose of mesh convergence testing?

A. To increase the antenna’s physical bandwidth

B. To determine whether a computed quantity is stabilizing with refinement

C. To eliminate the need for material data

D. To replace experimental validation

Correct Answer: B. To determine whether a computed quantity is stabilizing with refinement

Question 4

Which method naturally provides broadband information from a transient simulation?

A. FDTD

B. Static circuit analysis

C. Single-frequency MoM only

D. Geometrical optics only

Correct Answer: A. FDTD

Question 5

If radiation efficiency is 0.80 and |Γ| = 0.5, what is the total efficiency under the stated single-port convention?

A. 0.20

B. 0.40

C. 0.60

D. 0.80

Correct Answer: C. 0.60

Question 6

A Pareto-optimal antenna design is one for which:

A. All objectives have identical numerical values

B. No objective can improve without worsening at least one other objective

C. The design has no manufacturing constraints

D. The optimizer has performed only one iteration

Correct Answer: B. No objective can improve without worsening at least one other objective

Question 7

Which optimization approach is generally most suitable for a smooth, differentiable high-dimensional design with available sensitivities?

A. Gradient-based optimization

B. Random guessing without evaluation

C. Manual tuning only

D. A fixed lookup table unrelated to geometry

Correct Answer: A. Gradient-based optimization

Question 8

Why can a nominal optimum be unsuitable for manufacturing?

A. Manufacturing variation can shift its electromagnetic performance

B. Nominal designs cannot be simulated

C. Manufacturing always increases gain

D. Optimization removes all tolerances

Correct Answer: A. Manufacturing variation can shift its electromagnetic performance

Question 9

What does verification primarily examine?

A. Whether the numerical equations were solved correctly

B. Whether the antenna complies with every national law

C. Whether the product is aesthetically acceptable

D. Whether the optimizer used a neural network

Correct Answer: A. Whether the numerical equations were solved correctly

Question 10

For a dielectric with relative permittivity 4, the approximate wavelength under the simple homogeneous, nondispersive assumption is:

A. λ₀/4

B. λ₀/2

C. λ₀

D. 2λ₀

Correct Answer: B. λ₀/2