Chapter 7 Review · Wireless and Mobile Networks

Link layer ★ Chapter review Kurose & Ross p. 596 · ~13 min read

  • handoff
  • home network
  • indirect routing

Where you are

  • Application layer
  • Transport layer
  • Network layer
  • Link layer you are here
  • Physical layer

This chapter is two chapters. One is about a link you cannot see or trust. The other is about a device that will not stay still. They meet only on the last two pages.

What this page is

  • The chapter in one picture — every idea, linked to the section that taught it.

  • The argument, in four steps, which is how the chapter hangs together.

  • The two industries scored side by side, on the questions both had to answer.

  • Every number worth remembering, sortable.

  • The book’s five slips in this chapter, all verified on the page images.

  • A closing quiz on the chapter as a whole.

Where you are

Wireless networks and mobile networks are not the same subject, and the first thing this chapter did was separate them.

A wireless link is difficult because of physics: signal strength falls with distance, reflections arrive late, other sources interfere, and a sender cannot hear a collision it is causing. None of that requires anything to move.

Mobility is difficult because of addressing: a device that changes its point of attachment while its connections are open must still be findable. That requires nothing to be wireless — it just usually is.

Keep the two apart and the chapter is straightforward. Mix them and nothing in it makes sense.

The chapter in one picture

Chapter 7 in one picture

Hover an idea to see what it connects to. Click one to go back to the section that taught it.

Two halves that meet only at the end. Everything left of the dashed line is about a link that is wireless; everything right of it is about a device that moves. Click any box.

Read this map as a list
  • Four kinds of wireless network (§7.1 · infrastructure or not, one hop or many) — connects to Why the link is hard
  • Why the link is hard (§7.2 · path loss, multipath, interference, hidden terminals) — connects to Four kinds of wireless network, 802.11 architecture, TCP assumes loss means congestion
  • 802.11 architecture (§7.3.1 · the BSS and the access point) — connects to Why the link is hard, 802.11 MAC: CSMA/CA, 4G LTE and 5G
  • 802.11 MAC: CSMA/CA (§7.3.2 · avoid, do not detect) — connects to 802.11 architecture, The frame, and moving inside a subnet
  • The frame, and moving inside a subnet (§7.3.3–7.3.4 · four address fields) — connects to 802.11 MAC: CSMA/CA, Rate adaptation, power, Bluetooth
  • Rate adaptation, power, Bluetooth (§7.3.5–7.3.6) — connects to The frame, and moving inside a subnet
  • 4G LTE and 5G (§7.4 · gateways, a subscriber database, smaller cells) — connects to 802.11 architecture, 4G/5G in practice, and handover
  • Not everything that moves is mobile (§7.5 · a spectrum, not a switch) — connects to Home network and visited network
  • Home network and visited network (§7.5 · one place to ask, one place to coordinate) — connects to Not everything that moves is mobile, Indirect against direct routing
  • Indirect against direct routing (§7.5.3 · the triangle, and the price of removing it) — connects to Home network and visited network, 4G/5G in practice, and handover
  • 4G/5G in practice, and handover (§7.6.1 · four steps, then seven messages) — connects to Indirect against direct routing, Mobile IP, 4G LTE and 5G, TCP assumes loss means congestion
  • Mobile IP (§7.6.2 · the same design, never deployed) — connects to 4G/5G in practice, and handover
  • TCP assumes loss means congestion (§7.7 · and wireless breaks the assumption) — connects to Why the link is hard, 4G/5G in practice, and handover, Scarce bandwidth, location-aware apps
  • Scarce bandwidth, location-aware apps (§7.7) — connects to TCP assumes loss means congestion

The argument, in four steps

1 — The link is unreliable in ways a wire is not

Section 7.2 is the physical foundation for everything after it. Signal strength falls with distance, reflections cause multipath, other sources interfere, and two senders can each be audible to a receiver while being inaudible to one another.

That last case is the hidden terminal problem, and it is the reason section 7.3.2’s MAC (Media Access Control) protocol avoids collisions rather than detecting them. A wired sender can listen while transmitting. A wireless sender cannot: its own signal is overwhelmingly the loudest thing at its own antenna.

2 — Two industries answered the same questions differently

Sections 7.3 and 7.4 look like two unrelated technologies. They are one comparison.

Both had to decide who transmits next, how a device joins, who authenticates it, and what happens when it moves. 802.11 answers as a technology anyone can install; cellular networks answer as a service you subscribe to.

That single difference — subscription — is what gives cellular networks a home network . The home network is what makes the second half of the chapter possible at all.

3 — Mobility is a question about addressing

Section 7.5 removes the technology entirely and asks what must happen for data to reach something that moves. The answer is short: somebody has to know where the device is, and something has to carry data to that place.

Two designs follow. Indirect routing sends everything to the home network, which tunnels it onward — simple, and it walks two sides of a triangle. Direct routing asks where the device is first, then sends straight there — one side of the triangle, plus a location protocol, plus trouble when the device moves again.

Section 7.6 then shows that both real systems — 4G/5G, and the Internet Protocol’s own Mobile IP (Internet Protocol) — chose indirect.

4 — And then the two halves collide

Section 7.7 is two pages long and it is where the chapter’s two subjects finally meet each other, and meet chapter 3.

TCP (Transmission Control Protocol) reduces its sending rate whenever a segment is lost, because it assumes loss means a router queue overflowed. Step 1 of this argument said bit errors are common on wireless links. Step 3 said handover can lose segments too.

So on a mobile wireless path, TCP slows down for reasons that have nothing to do with congestion — while the router buffers may be completely empty.

The two industries, scored

The two industries, on the questions they both had to answer
802.11 WiFi§7.34G/5G cellular§7.4
Who may transmit next
How a device joins
Who the device proves itself to
Coverage of one attachment point
Moving between attachment points
Who owns the infrastructure

Cells marked ⓘ have a reason behind them — click to read it.

Sections 7.3 and 7.4 are one comparison in disguise. Neither column is better — they were designed for different scales and different owners. Click any cell.

Every number worth remembering

Every number worth remembering

Cells marked ⓘ have an explanation — click to read it. Sortable columns have a ↕ in the heading.

Sortable. Each one is either a design constant or the outcome of an argument the chapter makes.

The book’s slips in this chapter

The book's slips in this chapter
What the book printsWhat it should say

Cells marked ⓘ have an explanation — click to read it. Sortable columns have a ↕ in the heading.

Five, all verified against the printed page images rather than the text extraction.

In plain words

Four of the five are transcription damage in a figure or a table: transposed letters, a missing unit, a duplicated panel label.

The fifth is different and worse for a reader. The opening sentence of section 7.6 tells you to look in section 7.2.1 — twice — for material that is actually in 7.6.1 and 7.6.2. A reader who follows that pointer lands in the CDMA (Code Division Multiple Access) section and finds nothing about mobility management at all.

What the chapter does not settle

Two things are left genuinely open, and both are worth carrying forward.

Handover policy. The 4G and 5G standards fix the seven messages but specify no algorithm for deciding whether to hand over, or which target to choose. That is left to equipment vendors and remains an active research area.

TCP over wireless. The three classes of fix in section 7.7 are described as approaches, not as a settled answer. The one most widely deployed in cellular data networks — splitting the connection — is also the one that gives up the end-to-end connection the earlier chapters were built on.

Check yourself

Chapter 7 — check yourself

0 of 7 answered
  1. 1.The chapter separates two subjects at the start. What is the distinction?

  2. 2.Why does 802.11 avoid collisions rather than detect them, as Ethernet does?

  3. 3.What does a cellular network have that 802.11 does not, which makes sections 7.5 and 7.6 possible?

  4. 4.predictA roaming device hands over between two nearby base stations in the same visited network. Which of its two tunnels changes?

  5. 5.Mobile IP and split TCP connections had opposite fates. What does the chapter suggest decided them?

  6. 6.Which decision do the 4G and 5G standards deliberately leave unspecified?

  7. 7.Why does section 7.7 belong at the end of this chapter rather than in chapter 3?

The chapter in three lines

  • Wireless and mobile are two different problems. One is physics on a link, the other is addressing for a device that moves. A network can have either without the other.

  • The home network is the whole trick. It gives one place to ask where a device is and one place to coordinate through — and it exists because cellular networks have subscribers, which is exactly what 802.11 does not have.

  • Both real mobility systems chose indirect routing, and neither elegance nor standardization decided what got deployed. Mobile IP is clean and unused; split TCP connections break the end-to-end principle and are widely used.