EC-11.6 Wireless Links, Measurement and First Security

en

What this is and why it exists

Almost every network a reader uses has a radio hop at one end. Almost every networking course treats radio as an afterthought.

This topic closes that gap, and then ends with measurement and security. Those are the two things a graduate is most often asked to do on someone else's network.

The vocabulary

  • Path loss — the fall in signal strength with distance.
  • Fading — variation in signal strength caused by movement or reflection.
  • Collision avoidance — asking permission before sending, rather than detecting a clash after.
  • Cell — the area covered by one base station.
  • Frequency reuse — using the same frequencies in cells far enough apart.
  • Handover — moving a live connection from one cell to the next.
  • Packet capture — recording the frames that actually crossed a link.
  • Confidentiality — hiding the contents of a message.
  • Authentication — establishing who the other party actually is.

The mental model

Three things change when a link becomes radio, and each of them breaks an assumption made on cable.

The medium is shared with everyone in range, including networks you do not control and equipment that is not a network at all. The error rate is orders of magnitude higher and varies with position, weather and who walks between the two ends. And capacity is not symmetric, because the two ends rarely have the same power and the same antenna.

Local wireless networking therefore avoids collisions rather than detecting them. A sender that wants to transmit waits for the medium to be idle, then waits an additional random time, and only then sends. For larger frames it may ask permission first, so that stations which cannot hear it still learn to keep quiet.

Every frame is acknowledged, which is unnecessary on cable and essential here. Loss is normal rather than exceptional. Recovering it locally is far cheaper than letting the transport layer notice a missing byte a round trip later.

Cellular networks solve a different problem: coverage over a whole country for devices that move. The area is divided into cells, each with a base station. The same frequencies are reused in cells far enough apart that they do not interfere.

Handover is what makes movement possible. The device measures the signal from neighbouring cells, reports what it sees, and the network moves the live connection when another cell is better. The connection survives; the radio link underneath it is replaced. That separation is another instance of the layering argument from the first topic.

Now measurement, which is the practical half of this topic. A packet capture shows what actually crossed the link, as opposed to what a configuration file says should have.

That distinction settles arguments. Round-trip timing tells you the delay and its variation. Path tracing shows where a packet stops. Loss statistics distinguish a capacity problem from a configuration problem, because the first appears only under load and the second appears always.

The habit worth building is to measure before theorising. It is the same habit the measurement module argued for, and it is as often skipped here.

Finally, security, introduced only as far as this module can honestly go. Encrypting a link hides the contents from someone listening on that link. That is all it does.

It does not hide who is talking to whom, because addresses must remain readable for the packet to be delivered. It does not protect the message beyond the far end of that link. And on its own it does not prove who sent anything, which is a separate property requiring authentication.

Three attacks are worth understanding at this stage, because all three are cheap on a network that authenticates nothing. Listening to an unencrypted medium. Sending frames that carry another machine's address as their own. And flooding a service with requests until it stops answering everyone else.

The first two are why encryption and authentication are added. The third is not solved by either, and saying so plainly is more useful than implying that security is one thing you switch on.

What you should now be able to explain or do

  • Name the three properties of a radio link that break assumptions made on cable.
  • Explain collision avoidance and why every wireless frame is acknowledged.
  • Describe cells, frequency reuse and what happens during a handover.
  • Use capture, timing and tracing to locate a network problem before theorising about it.
  • State precisely what encrypting a link protects and what it leaves exposed.
  • Describe three cheap attacks on an unauthenticated network and say which ones encryption addresses.

Check yourself

Because loss is normal on radio rather than exceptional. Recovering it locally costs one frame; letting the transport layer notice costs a whole round trip and a retransmission from the far end.

The connection survives. The radio link underneath it is replaced by one to a different base station, which is exactly the separation layering was designed to allow.

Who is talking to whom, since addresses must stay readable. It also does not protect the message beyond that link, and by itself it does not establish who sent anything.

Flooding a service until it stops answering. The requests can be perfectly authentic; the problem is their number, and that needs a different kind of defence entirely.

Go deeper

Back to Wireless Links, Measurement and First Security: work through the checklist