Internet Protocol (IPv4): Broadcasting, Subnetting, NAT, ICMP & ARP

Broadcast Addresses

IPv4 supports broadcasting: a single datagram is delivered to all hosts/routers on the local subnet.
- Achieved by putting the broadcast address in the destination IP field.
How to form the broadcast address of a subnet:
- Keep the network prefix untouched.
- Set every bit in the host portion to $1$ .
- Example (Class C sized subnet): $223.1.1.0/24 \Rightarrow 223.1.1.255$ .
Delivery mechanics on Ethernet:
- IP datagram is wrapped in an Ethernet frame whose destination MAC is $ff:ff:ff:ff:ff:ff$ (all 1 bits at Layer-2).
Constraint: the broadcast address may appear only as a destination, never as a source, in an IP packet.

Directed vs Limited Broadcast

Directed broadcast
- Constructed exactly the same way—host bits $=1$ —but the address is specific to one subnet and therefore routable across the Internet.
- Example: for $223.1.1.0/24$ the directed broadcast is $223.1.1.255$ .
- Any host, anywhere, could send to that address; routers will forward until the packet reaches the target subnet, where it is broadcast locally.
Limited broadcast
- Fixed, subnet-independent address $255.255.255.255$ (all 1s in every octet).
- Routers never forward such packets; the scope is limited to the originating subnet.

IP Address Classes (OBSOLETE)

Historic, now discouraged way of carving the IPv4 space by inspecting the first few bits.
Encodings and implied prefix lengths:
- Class A: leading bit $0$ , implied prefix $/8$ .
- Class B: leading bits $10$ , implied prefix $/16$ .
- Class C: leading bits $110$ , implied prefix $/24$ .
- Class D: leading bits $1110$ , reserved for multicast.
- Class E: leading bits $1111$ , future use / experimental.
Modern practice: always state netmask or CIDR prefix explicitly; never rely on classful inference.

Subnetting Concepts

Subnetting = subdividing one larger IP block into several smaller, non-overlapping blocks.
Visualised as a binary aggregation tree; every left/right branch halves the available address space.
Examples:
- Splitting a $/26$ yields two $/27$ sub-prefixes.
- Combining two adjacent $/25$ networks recreates the parent $/24$ .
Key rule: choose the longest prefix that still offers ≥ required host addresses (remember 2 hosts lost to network & broadcast IDs).

Worked Subnetting Example (223.1.1.0/24)

Goal: create three usable subnets for $100$ , $50$ and $25$ hosts.
Steps (largest first):
1. $100$ hosts ⇒ need $\ge 100+2 = 102$ addresses ⇒ $/25$ (gives $2^{7}-2 = 126$ usable).
- Assigned range: $223.1.1.0 \text{ – } 223.1.1.127$ ⇒ $223.1.1.0/25$ .
1. $50$ hosts ⇒ need $/26$ ( $2^{6}-2 = 62$ usable).
- Assigned range: $223.1.1.128 \text{ – } 223.1.1.191$ ⇒ $223.1.1.128/26$ .
1. $25$ hosts ⇒ need $/27$ ( $2^{5}-2 = 30$ usable).
- Assigned range: $223.1.1.192 \text{ – } 223.1.1.223$ ⇒ $223.1.1.192/27$ .
Remaining addresses: $223.1.1.224/27$ (unused / future growth).

Subnetting with Private Address Blocks

RFC 1918 private ranges: $10.0.0.0/8$ , $172.16.0.0/12$ , $192.168.0.0/16$ .
When space is plentiful (e.g. using $172.16.0.0/12$ ): simply dedicate one $/24$ per building/department if ≤ $254$ hosts each.
- Example allocation:
- $172.16.1.0/24$ — subnet needing $100$ hosts.
- $172.16.2.0/24$ — subnet needing $50$ hosts.
- $172.16.3.0/24$ — subnet needing $25$ hosts.

Subnetting Exercises (Campus Scenarios)

Exercise 1: public block $192.40.8.0/21$ , reserve first $/24$ for core, then carve out ranges for Science ( $120$ hosts), Engineering ( $200$ ), Library ( $80$ ), Arts ( $50$ ), Administration ( $40$ ), Hostel ( $60$ ), Wi-Fi ( $350$ ) plus point-to-point router links.
Exercise 2: identical logical design but using private block $172.16.0.0/12$ .
Both emphasise choosing prefixes based on host counts and keeping aggregation contiguous.

Private vs Public Addressing & Allocation

Private addresses suffice only for internal communication.
To reach the global Internet, hosts need public, globally routable unicast addresses.
Public address procurement channels:
1. Direct assignment from Regional Internet Registries (RIRs) – usually paid, and IPv4 is exhausted.
2. Delegation from an ISP – ISP owns a larger block delegated by an RIR and loans you a subset.
Because ISPs rarely provide one public address per internal host, most organisations deploy NAT.

Network Address Translation (NAT) – Principles

Inside network uses private addresses ( $10/8$ , $172.16/12$ , $192.168/16$ ).
NAT router sits at the boundary and changes address fields on-the-fly:
- Outbound: replace source private IP with a public IP (or with one of many via pooling). Store mapping in translation table.
- Inbound: look up translation table and rewrite destination public IP back to original private IP.
Typical NAT table entry: outside address ↔ inside address (plus TCP/UDP port numbers when Port Address Translation (PAT) is used).

Detailed NAT Walk-through (Single Public IP)

Host $10.0.0.1$ sends datagram to $128.119.40.186$ .
NAT rewrites source $10.0.0.1 \rightarrow 138.76.29.7$ , stores mapping.
Reply arrives for $138.76.29.7$ .
NAT consults table, rewrites destination $138.76.29.7 \rightarrow 10.0.0.1$ , forwards internally.

Port multiplexing allows sharing one public IP among many internal hosts — NAT also rewrites TCP/UDP ports (covered later in TCP/UDP lecture).

Advantages of NAT

Only one (or a small pool of) public IPs needed — conserves scarce IPv4 addresses.
Internal addressing freedom: can renumber inside hosts, or switch ISPs, without visible change to the outside world.
Implicit security: inside hosts are not directly reachable, offering a basic firewall-like barrier.

Hands-On NAT Exercise (Wi-Fi Test)

Join $SLIIT\text{-}STD$ Wi-Fi, record device’s private IP.
Browse to https://focnet.sliit.lk/ie1030/nat-test.php.
Page echoes the public IP:PORT currently representing you on the Internet — precisely the NAT table mapping.

ICMP – Internet Control Message Protocol

Sits above IP (protocol number $1$ ) and is carried inside IP packets.
Core purposes:
- Error reporting (destination unreachable, fragmentation needed, TTL expired, etc.).
- Echo diagnostics (ping).
- Router discovery & advertisement.
Generic ICMP packet structure: Type | Code | Checksum | data (often the first $8$ bytes of the offending IP packet).

Common ICMP Types & Codes

Echo reply: Type $0$ Code $0$ .
Echo request: Type $8$ Code $0$ .
Destination unreachable: Type $3$ with multiple codes:
- $0$ network unreachable, $1$ host unreachable, $3$ port unreachable, $4$ fragmentation needed but DF set, $6$ network unknown, $10$ communication administratively prohibited, etc.
TTL expired: Type $11$ Code $0$ .
Bad IP header: Type $12$ Code $0$ .

Ping (ICMP Echo)

Sender emits Echo Request (Type $8$ ).
Receiver responds with Echo Reply (Type $0$ ).
Identifier + Sequence Number let sender match replies and measure RTT.

ICMP Destination Unreachable

Usually generated by an intermediate router.
Frame includes IP header + first $8$ bytes of original payload so the sender can identify the failed packet.

Traceroute Algorithm

Send 3 probes per hop, incrementing TTL: $1,2,3,\dots$ .
Each router that decrements TTL to $0$ returns ICMP Time exceeded (Type $11$ ).
When destination is finally reached, it replies with Port unreachable (Type $3$ Code $3$ ) — traceroute stops.
RTT for each hop measured from request ➔ corresponding ICMP reply.

Address Resolution Protocol (ARP)

Goal: map known IPv4 addresses to unknown MAC addresses on a local (broadcast) network.
ARP is considered layer 2.5 — messages are not encapsulated in IP; instead they ride directly inside Ethernet frames (Ethertype $0806$ ).

ARP Table (Cache)

Each host stores learned bindings: IP, MAC, TTL (≈ $20$ min default).

ARP Packet Format

Fixed header (8 bytes): Hardware Type, Protocol Type, HLEN, PLEN, Opcode.
- $0001$ = Ethernet, $0800$ = IPv4, HLEN $=6$ , PLEN $=4$ .
- Opcode $1$ = request, $2$ = reply.
Variable portion: Sender MAC, Sender IP, Target MAC, Target IP (total 20 bytes for Ethernet/IPv4).
Entire ARP message encapsulated in an Ethernet frame → FCS.

ARP Operational Sequence

Host A needs MAC of host B (IP known, MAC unknown, not in cache).
A broadcasts ARP Request with:
- Dest MAC $ff:ff:ff:ff:ff:ff$ , Target IP = B, Target MAC = $00:00:00:00:00:00$ .
Every LAN node examines; only B replies via unicast ARP Reply:
- Contains B’s MAC in Sender MAC/Target MAC fields.
A receives, updates its ARP table, sets TTL timer.
Subsequent IP packets to B are encapsulated using B’s resolved MAC.

Inter-Protocol Relationships (Big Picture)

IP delivers packets end to end; broadcast, subnetting & NAT modify addressing semantics.
ICMP gives the network a “voice” for diagnostics & error signalling.
ARP glues IP to the underlying data-link layer, enabling local delivery.
Combined, these protocols enable scalable, routable, and debuggable IPv4 networking in both private and public contexts.