In the context of switching networks, the cost of a network is usually…

Concept

A complexity measure identifies the resource whose growth or quantity represents cost under a particular model. Depending on the model, that resource may be execution steps, switching elements or operations, or circuit gates and depth.

For a switching-network realization, implementation cost is associated with the switching resources needed to provide the required connections. The conventional name for that resource measure is switching complexity.

Application

The stem fixes the model as a switching network and asks for its usual cost measure. The relevant resource is therefore the switching structure used to realize the network, so the applicable measure is switching complexity.

Contrast

• Time complexity measures how an algorithm’s execution time or operation count grows with input size; its resource is computation time rather than switching hardware.

• Circuit complexity measures the gate count, size, or depth of a Boolean circuit model; it is a circuit-model measure rather than the specific switching-network term asked here.

• None of these would apply only if no listed category named the relevant resource measure, but switching complexity is one of the listed categories.

Cross-check

As a terminology check, an algorithmic running-time question would point to time complexity, while a Boolean-circuit size or depth question would point to circuit complexity. Retaining the switching-network context consistently points to switching complexity.

Result

The usual network-cost measure in this context is switching complexity.