TSB-88 coverage acceptance, explained for the shop that has to pass it

The criterion comes first

Everything in a TSB-88 study is measured against a channel performance criterion (CPC): the quality of signal a user has to receive for a location to count as covered. For voice it is usually stated as a delivered audio quality (DAQ), a scale that runs from unusable to perfectly clear; for data it is a throughput or error rate.

A CPC is turned into something a radio can be measured against by the signal-to-interference-plus-noise ratio the receiver needs to deliver it with a given modulation, which the standard tabulates. That ratio, not a raw signal level, is what decides whether a tile passes.

The criterion is a design decision, and it belongs to the contract. A reliability figure computed against a criterion nobody chose is a figure nobody can defend, which is why a study should make you state it rather than fill one in quietly.

Reliability at a point, and over an area

Signal strength at a location is not one number. It varies from spot to spot within a small area and from moment to moment, so a prediction gives a median level and a spread around it. Reliability at a point is the probability that the criterion is met there.

Area reliability is that idea across the whole service area: the share of the area where the criterion is met with the required probability. A contract states a target for it, and the target matters as much as the criterion. The same system can read very differently against two targets.

Tiles

TSB-88 divides the service area into a grid of uniform tiles. A prediction gives each tile its own median signal and reliability, and an acceptance test samples each tile, so prediction and measurement are compared tile for tile rather than as one coverage outline.

The service area should be the one the system actually serves: a city boundary, a county, a district, traced or loaded as a boundary file. A circle around the tower counts tiles nobody asked to cover and dilutes the ones they did.

Interference is part of it

Because the criterion is a ratio to interference plus noise, a system that is loud enough can still fail where a co-channel or adjacent-channel signal is strong. A study that models only noise answers a narrower question. Adjacent-channel interferers are weighed by the receiver's own rejection, not treated as if they were on channel.

Acceptance testing

Acceptance is measured, not predicted. A test plan agreed in the contract sets how many tiles are tested, how many samples are taken in each, and how the result is scored, so that the pass or fail carries statistical confidence rather than depending on one lucky or unlucky drive. Portable and in-building coverage is tested against the building losses the contract defines, which is why those losses have to be agreed before the design, not after.

What a prediction can and cannot do

• It can size a system, choose between sites, show where the weak tiles will be, and explain to a customer why a design is what it is.
• It rests on its inputs: the terrain and land-cover data, the antenna pattern, the building losses and the criterion. Each should be stated in the report so the prediction can be judged.
• It does not replace the acceptance test, and it does not replace licensed engineering review on a public-safety design.