1 What it means for a network to sense

1.1 A radio that is already there

A mobile network fills a place with radio. That is its whole job: a base station transmits, a phone receives, and between them the air is never empty.

Anything solid in that air changes the signal. It bounces some of it back, bends some of it, casts a shadow behind itself. A receiver that only wants the message throws all of that away as noise.

Sensing is the proposal that the network stop throwing it away. Read the change instead of discarding it, and the same radio that carried a phone call tells you that something is out there, roughly where, and roughly how fast it is moving.

The short name for this in 3GPP is ISAC — integrated sensing and communication. "Integrated" is the load-bearing word: not a radar bolted on beside the network, but the network's own signal doing two jobs at once.

1.2 What 3GPP means by the word

The requirements document defines the feature before anything else does.

Three things in that sentence are worth slowing down for. It is a feature of the 5G system, not of the radio alone. The list of characteristics is wide — shape, size, orientation, speed, location, distance, motion. And the signals are NR radio signals, meaning the 5G radio the network already runs.

Two more definitions separate the raw material from the product. "3GPP sensing data" is data derived from 3GPP radio signals that an object or environment has reflected, refracted or diffracted ^[4]. A "sensing result" is that data once it has been processed for whoever asked for it ^[5].

That split matters later, because the groups that argue about sensing data and the groups that argue about sensing results are not the same groups.

1.3 Four statements of the same problem

This subject is not one work item. The local record ties 18 entries in the 3GPP work plan to it ^[6], under 14 different acronyms ^[7] — three entries share the acronym Sensing and three share Sensing-CT, which is why those two figures differ.

Between them those entries have produced 3492 meeting documents in the local record ^[8]. 205 of them were filed in 2022 ^[9], the oldest year the record holds, and 1031 in 2026 ^[10], the newest.

Four of those work items open by saying why anybody should bother, and the four say it differently enough to be worth reading side by side.

The first one, from the requirements group, states the general idea. Wireless sensing acquires information about a remote object or environment, and the data received about an object and its surroundings can be analysed to obtain meaningful information about it ^[11].

The radio study states a gap instead. 5G-Advanced was designed around carrying data; the network can already work out where a connected device is, but the specifications offer no built-in way to detect an object that is not connected to the network at all ^[2].

The architecture study states an ambition. It calls integrating sensing with communication one of the promising technologies of the moment, and names the industries it could change: cars, healthcare, drones, smart cities ^[12].

The second radio study states the mechanism, and only the mechanism. The same radio signal is used for sensing and for communication at the same time ^[13]. What that is supposed to buy, not one of the four documents says.

1.4 What is actually being built first

The four statements above are the ambition. The specification currently being written is far narrower, and the gap between the two is the single most useful thing to know about this area.

Aerial objects, and nothing else. And the sensing itself is done by the base station: "gNB-based Sensing" means the gNB — the 5G base station — is the thing acting as the sensing entity, and the definition adds that this implies nothing about where the result is worked out ^[14].

The requirements document above it is deliberately unspecific about applications. It says of itself only that it describes functional and performance requirements for the 5G wireless sensing service ^[15].

So the shape of this area is a wide requirement, a wide study, and a first build that takes one slice: find drones with base stations. Every count in this course sits somewhere between those two.

1.5 Why reuse the network instead of building a radar

Radar exists. Nothing in 3GPP had to be invented for somebody to detect a drone; a radar does that, and has done for decades.

The argument for doing it in the network is one of reuse, and the second radio study states only the bare mechanism: the same radio signal serves sensing and communication at the same time ^[13].

That is as far as the record behind this course goes, and the gap matters. Not one document in it claims a benefit — not better use of spectrum, not lower delay, not higher reliability, not lower cost. Whether reusing the network beats building a radar is the question the studies were opened to settle, and their conclusions are not in this data.

So the honest answer to "why not just use a radar" is that nobody has written one down where this course can read it. What is written down is the mechanism above, and the first narrow build: find drones, with base stations.

The counter-argument is in the numbers rather than in any document here: a network built to carry data is not built to hear echoes, and the whole first radio study was spent on how to even model the problem — RAN1 and the radio problem is about that.

1.6 What this course cites, and why

There is no single specification behind this course. An area of work is not a document, so nothing here can point at a clause the way a course about TS 23.501 can.

Instead every statement points at an entry in a source register built from files on this machine: the work plan, the meeting-document database, the specification catalogue and the parsed text of the specifications themselves. A link that reads like a number is a link into that register.

Two warnings about the numbers belong here rather than at the end, because they apply to every figure in every chapter.

• Every count is a floor, never a total. Documents were counted only when they carry one of the 14 work item acronyms. 440 documents whose title says "integrated sensing" or "ISAC" carry no acronym at all ^[16], and none of them is in the 3492.

• The word "sensing" is used by other work entirely. 2053 documents in the local record say sensing and mean something else — sidelink partial sensing, listen-before-talk channel sensing, or the separate 6G work ^[17]. Nothing here was found by searching for the word.

1.7 Where to read the real thing

This school does not carry the text of any 3GPP specification for this subject. Every chapter ends by naming the document to go and open instead.

For this chapter there are two. The requirements are TS 22.137, which is short and defines the words the rest of the area uses TS 22.137. The stage 2 specification being written now is TS 23.137 TS 23.137.

Next comes what people actually want this for: the use cases the first study collected, in What people want it for, and why that question came first.