What Anritsu means by an AI-enabled vector network analyser

Anritsu describes Tensor as the world’s first AI-enabled vector network analyser. The important distinction is not that artificial intelligence performs the underlying RF measurement, but that AI-assisted functions are being added to the instrument’s setup, workflow and interpretation tools.

A vector network analyser measures how an RF or microwave device changes signals passing through it. These results are normally expressed as S-parameters, including reflected and transmitted magnitude and phase.

The measurements are used to characterise amplifiers, filters, antennas, mixers, cables, PCB interconnects, connectors and waveguides.

Source-per-port architecture

Tensor uses what Anritsu calls a source-per-port architecture. Each measurement port has access to its own signal source rather than relying on one source switched sequentially between ports.

This can reduce switching between measurement states and support more flexible excitation of multiport devices. The architecture is relevant to differential components, phased-array systems and devices whose behaviour changes depending on which ports are driven.

Anritsu says the platform can also work with third-party millimetre-wave frequency-extension heads. The company lists supported measurement applications extending to 1.1THz, although the frequency range of an individual installation will depend on the selected base instrument, extenders, probes and calibration equipment.

Where the AI is used

The AI-assisted features are intended to guide instrument configuration and automate parts of the measurement workflow. Anritsu says the system can simplify amplifier, mixer, S-parameter, calibration, limit-line, SCPI and reporting tasks.

This is distinct from replacing calibration or measurement traceability with a generated answer. The measured RF data still comes from the instrument’s receivers, sources and calibration model.

For an AI-assisted recommendation to be useful, the operator must still define the device, frequency range, power level, calibration plane and required accuracy correctly. Automated setup cannot compensate for an unsuitable fixture, damaged cable or calibration standard used outside its specified range.

The AI engine operates within the instrument without requiring an internet connection, according to Anritsu. The analyser can also run applications internally, allowing it to act as both the measurement platform and control computer in some configurations.

Likely applications

Anritsu lists high-frequency device characterisation, aerospace and defence systems, power-amplifier and mixer testing, signal-integrity measurements and PCB, connector and waveguide validation among the target applications.

Multiport measurements are a particular focus. Modern RF systems may require several ports to be measured coherently, while high-speed digital interconnects often need differential and mode-conversion parameters as well as basic insertion and return loss.

The value of Tensor’s AI functions will ultimately depend on how well they handle real test configurations, how clearly their recommendations are explained and whether automated workflows remain repeatable between operators.

Anritsu is demonstrating Tensor at European Microwave Week in London from 6 to 8 October 2026.