Network Layer Fundamentals

Network Layer Overview

The network layer is the third layer of the OSI reference model. It is mainly responsible for transferring packets between different networks and providing end-to-end communication.

---

1. Core Functions of the Network Layer

1.1 Heterogeneous Internetworking: the “Translator” of the Network World

Why Is Heterogeneous Internetworking Needed?

Many kinds of network exist in the real world:

These networks are like different languages that cannot talk to each other directly. Through the uniform IP protocol, the network layer makes the languages interoperable.

The IP Protocol: the “Common Language” of the Network

Key Technologies for Heterogeneous Internetworking

1. Address translation

Different networks use different address formats:

• Ethernet: MAC address (48 bits, for example 00:1A:2B:3C:4D:5E)
• IP networks: IP address (32-bit IPv4 or 128-bit IPv6)
• Telephone networks: telephone number

The network layer performs address translation through the ARP protocol: IP address → ARP protocol → MAC address, for example: 192.168.1.1 → query → 00:1A:2B:3C:4D:5E.

2. Data encapsulation

Different networks use different data formats, so the network layer encapsulates them uniformly:

Like a courier parcel:

• the contents are the goods (raw data)
• the outer layer is the courier box (the IP packet)
• the outer layer of that is the transport container (the Ethernet frame)
• the container changes with the transport method

3. Protocol conversion

The network layer converts between different protocols:

• TCP/IP ↔ IPX/SPX
• IPv4 ↔ IPv6
• Wired ↔ wireless

---

1.2 Routing and Packet Forwarding: the “Navigation System” for Packets

How Routing Works

An analogy with navigation software:

The routing process:

The Packet Forwarding Mechanism

A forwarding example:

Suppose you are sending a parcel from Beijing to Guangzhou:

The packet leaves your computer and passes through the Beijing, Shijiazhuang, and Wuhan routers in turn before reaching the Guangzhou server.

The forwarding procedure at each router:

1. Receive the packet
2. Extract the destination IP address
3. Look up the routing table
4. Choose the next-hop router
5. Encapsulate the packet
6. Send it to the next hop

The Structure of the Routing Table

A router maintains a routing table that records information about the network topology:

What the routing table columns mean:

• Destination network: the network address to be reached
• Next hop: the address of the next router
• Interface: the port the packet leaves through
• Priority: the entry with the higher priority is chosen when several paths exist

---

1.3 Congestion Control: the “Traffic Controller” of the Network

Causes of Congestion

An analogy with traffic congestion:

How congestion shows up:

• 📉 The transmission rate drops
• 📈 The packet loss rate rises
• ⏱️ The delay increases
• 🔁 The number of retransmissions grows

Congestion Control Strategies

1. Congestion detection

Monitor the state of the network:

• router queue lengths
• packet loss rate
• average delay
• link utilization

2. Congestion avoidance

Prevent congestion before it happens:

• rate shaping: control the sending rate
• load balancing: spread traffic over several paths
• route optimization: steer traffic around congested areas
• resource reservation: reserve bandwidth for important traffic

3. Congestion recovery

Handle the aftermath of a congestion event:

• loss notification: ICMP source quench messages
• rate reduction: the sender slows down
• path switching: move to a different path
• priority scheduling: handle important data first

4. Traffic shaping

Control the traffic entering the network:

• token bucket: limit the sending rate
• leaky bucket: smooth the traffic
• Random Early Detection (RED): drop packets proactively to prevent congestion

---

2. Types of Service Offered by the Network Layer

2.1 Connectionless Service (Datagram Service)

Characteristic Analysis

An analogy with posting letters:

How It Works

Sender:

1. Prepare the packet (adding the destination IP address)
2. Send it straight to the network
3. Do not wait for an acknowledgment

Network:

1. Every router routes independently
2. Packets may take different paths
3. Packets may be lost, delayed, or reordered

Receiver:

1. Receive the packet
2. Check whether the destination address matches
3. Process the data or discard the packet

Applicable Scenarios

Suitable for real-time applications:

• 📞 Voice calls: a little packet loss does not affect call quality
• 📹 Video streaming: a small amount of data loss can be tolerated
• 🎮 Online games: real-time responsiveness matters more than reliability
• 📡 Video conferencing: real-time interaction, latency sensitive

---

2.2 Connection-Oriented Service (Virtual Circuit Service)

Characteristic Analysis

An analogy with making a phone call:

How It Works

Connection establishment:

1. The sender sends a call request
2. The network establishes the virtual circuit (assigning a virtual circuit number)
3. Every router on the path records the virtual circuit information
4. The receiver confirms the connection

Data transfer:

1. Every packet carries the virtual circuit number
2. Routers forward based on the virtual circuit number
3. All packets follow the same path
4. Order and reliability are guaranteed

Connection release:

1. The sender sends a release request
2. The network releases the virtual circuit resources
3. The routers on the path clear their entries
4. The connection ends

Applicable Scenarios

Suitable for applications that need reliable transfer:

• 📁 File transfer: FTP and file downloads, where data must not be lost
• 🌐 Web browsing: HTTP and HTTPS, which need complete pages
• 📧 Mail transfer: SMTP and POP3, where message content must stay intact
• 📂 Database synchronization: data has to be exactly right

---

2.3 Comparison of the Two Services

---

3. The Main Protocols of the Network Layer

3.1 The IP Protocol: the Foundation of the Internet

Characteristics of the IP Protocol

Connectionless:

• every packet is handled independently
• no connection has to be established in advance
• as simple as posting a letter

Best effort:

• the data is delivered with the greatest effort possible
• reliability is not guaranteed
• packets may be lost, delayed, or reordered

Uniform addressing:

• every device has a unique IP address
• a globally scoped address system
• support for IPv4 and IPv6

The Structure of an IP Packet

IP packet = IP header + data portion

The key fields:

---

3.2 Supporting Protocols: the “Helpers” of IP

The ARP Protocol: Address Resolution

How it works:

Scenario: the computer wants to send data to the server with the IP address 192.168.1.100.

1. The computer knows the destination IP: 192.168.1.100
2. But it does not know the MAC address (the physical address)
3. It broadcasts an ARP request: “Who is 192.168.1.100? Tell me your MAC address”
4. The server receives the ARP request and replies with an ARP response: “I am 192.168.1.100, and my MAC address is 00:1A:2B:3C:4D:5E”
5. The computer receives the response and records the IP-to-MAC mapping
6. It can then send data directly

The courier analogy:

• IP address = the recipient’s address (the name of the residential complex)
• MAC address = the exact door number (building and unit)
• The ARP protocol = the process of looking up the door number

The DHCP Protocol: Dynamic Address Assignment

How it works:

A device joins the network (the four-step DHCP exchange):

1. The device sends a DHCP Discover: “I am new here and need an IP address”
2. The DHCP server sends an Offer: “I am a DHCP server and I can offer an IP address”
3. The device sends a DHCP Request: “Please give me an IP address”
4. The DHCP server sends an Acknowledge: “Your IP is 192.168.1.100, with a lease of 24 hours”
5. The device starts using that IP
6. The DHCP server records it: “192.168.1.100 has been allocated to this device”

Benefits:

• ✅ Automatic configuration with no manual setup
• ✅ Address reuse, which raises utilization
• ✅ Central management, which avoids address conflicts
• ✅ Mobility: a device that moves to another network obtains a new address automatically

The ICMP Protocol: Network Diagnostics

Common applications:

The ping command:

ping 192.168.1.100

Purpose: test whether the network is reachable
Process: send an ICMP request and receive an ICMP response
Result: display latency, packet loss rate, and related information

The traceroute command:

traceroute www.baidu.com

Purpose: trace the path of a packet
Process: send a series of ICMP packets and record every router
Result: display all the routers traversed and the latency at each hop

---

4. Network Layer Devices

4.1 Router: the “Traffic Hub” of the Network

Router Functions

1. Connecting different networks:

• connecting LANs and WANs
• connecting wired and wireless networks
• connecting networks built on different technologies

2. Routing:

• maintaining the routing table
• choosing the best path
• updating routing information dynamically

3. Packet forwarding:

• receiving packets
• looking up the routing table
• sending them to the next hop

4. Other functions:

• NAT address translation
• ACL access control
• QoS quality assurance
• VPN support

How a Router Works

1. Extract the destination IP address
2. Look up the routing table
3. Match the routing entries
4. Choose the best route
5. Check the TTL (decrement it by 1)
6. Update the checksum
7. Encapsulate it into a new frame
8. Send it out of the designated interface

A routing table example:

Routing matching rules:

• Longest match: choose the route with the longest (most specific) netmask
• Priority: when several routes match, choose the one with the higher priority

---

4.2 Layer 3 Switch: a High-Performance Router

Characteristics of a Layer 3 Switch

Differences from a router:

How it works:

The first packet:

1. Software processing (like a router)
2. Look up the routing table
3. The hardware records the routing information

Subsequent packets:

1. The hardware forwards them directly (like a switch)
2. Forwarding is fast, using the hardware routing table
3. No software processing is required

The defining characteristic is: “route once, switch many times”.

---

Practical Examples

Example 1: The Process of Visiting a Website

Consider the network layer steps involved in loading a web page in a browser:

Your computer (192.168.1.10):

1. You enter the address www.baidu.com
2. DNS resolves it to an IP address: 220.181.38.148
3. It builds the IP packet: source IP 192.168.1.10, destination IP 220.181.38.148, with an HTTP request as the data

The local router: 4. Receives the packet 5. Looks up the routing table; the destination IP is not on the local network 6. Selects the default route and forwards it to the upstream router 7. Converts the source IP to a public IP using NAT

The ISP routers (several of them): 8. Each router looks up its routing table 9. Each router selects the next hop 10. The packet is forwarded hop by hop toward the Baidu server

The Baidu server (220.181.38.148): 11. Receives the IP packet 12. Checks that the destination IP matches 13. Extracts the HTTP request 14. Processes the request and returns the page

The return path: 15. The server builds the response packet (source IP 220.181.38.148, destination IP your public IP) 16. It returns hop by hop through the routers 17. NAT converts it back to the private IP 18. Your computer receives the response

Example 2: Interconnecting Different Networks

A home network interconnected with a corporate network:

Home network:

• Wi-Fi network: 192.168.1.0/24
• Wired network: 192.168.2.0/24
• Devices: phones, computers, televisions

Corporate network:

• Office network: 10.0.0.0/8
• Data center: 172.16.0.0/16
• Devices: servers, workstations

The interconnection process:

1. The home router connects to the ISP over fiber
2. The ISP connects to the corporate network over a leased line
3. The IP protocol encapsulates uniformly: home Wi-Fi frame → IP packet → fiber frame → ISP routing → IP packet → leased line frame → corporate router
4. Communication across the different networks becomes possible

---

Exercises

Exercise 1

What is heterogeneous internetworking, and why is it needed?

Exercise 2

What is the difference between routing and packet forwarding?

Exercise 3

What is the difference between the datagram service and the virtual circuit service, and what is each used for?

Exercise 4

What is the role of the ARP protocol? Briefly describe how it works.

---

Glossary