UNIT 2: COMPUTER NETWORKS - SHORT NOTES

I. NETWORK MODELS AND ARCHITECTURE

A. ISO-OSI Reference Model

Definition: A 7-layer theoretical framework for network communication, developed by ISO. Each layer provides services to the layer above it and uses services from the layer below.

Service Primitives: Request, Indication, Response, Confirm (used between adjacent layers).

[!TIP] CRITIQUE: OSI is reference model only; no widespread real-world implementation. Protocols were designed after the model, causing mismatches. TCP/IP was developed concurrently and gained adoption.

B. TCP/IP Reference Model

Layers:

1. Network Interface (Link): Combines OSI's Physical & Data Link. Handles hardware addressing (MAC).

2. Internet: OSI's Network layer. Core protocol: IP. Handles logical addressing & routing.

3. Transport: OSI's Transport. Protocols: TCP (reliable), UDP (unreliable).

4. Application: Combines OSI's Application, Presentation, Session. Protocols: HTTP, DNS, SMTP.

Advantages: Practical, implemented first, scalable, protocol-independent. Disadvantages: Less rigid layering, no clear separation of services (e.g., no distinct session/presentation layers).

C. OSI vs. TCP/IP Comparison

II. PHYSICAL LAYER AND TRANSMISSION MEDIA

A. Guided Transmission Media

B. Unguided Transmission Media

C. Multiplexing Techniques

D. Line Coding Techniques

Maps binary data to physical signals.

E. Transmission Capacity (Nyquist & Shannon)

1. Nyquist Theorem (Noiseless Channel):

Maximum bit rate = $$\displaystyle 2 \times B \times \log_2(M) $$

• $B$ = Bandwidth (Hz)

• $M$ = Number of discrete signal levels (voltage levels)

Example: $$\displaystyle B = 3 $$ kHz, $$\displaystyle M = 4 $$ (2 bits/level). Max rate = $$\displaystyle 2 \times 3000 \times \log_2(4) = 2 \times 3000 \times 2 = 12,000 $$ bps.

If $M$ increases to 8 (3 bits/level), rate = $$\displaystyle 2 \times 3000 \times 3 = 18,000 $$ bps.

2. Shannon's Theorem (Noisy Channel):

Channel Capacity $$\displaystyle C = B \log_2(1 + \text{SNR}) $$

• $B$ = Bandwidth (Hz)

• $\text{SNR}$ = Signal-to-Noise Ratio (linear, not dB)

Example: $$\displaystyle B = 4 $$ kHz, $$\displaystyle \text{SNR} = 1000 $$ (30 dB). $$\displaystyle C = 4000 \log_2(1001) \approx 4000 \times 9.97 \approx 39,880 $$ bps.

Key Relationship: Capacity increases with bandwidth and SNR. Nyquist gives an upper bound for a given $M$; Shannon gives the absolute maximum for a noisy channel.

III. DATA LINK LAYER

A. Functions & Services

• Framing: Packet (network layer) → Frame (adds header/trailer).

• Error Control: Detection (CRC) & Correction (Hamming).

• Flow Control: Prevent fast sender from overwhelming slow receiver (Stop-and-Wait, Sliding Window).

• Access Control: MAC protocols for shared media (Ethernet, Wi-Fi).

B. Error Detection & Correction

CRC Computation Steps:

1. Append $r$ zeros to data $D(x)$ ($r$ = degree of $G(x)$).

2. Divide $$\displaystyle D(x) \cdot x^r $$ by $G(x)$ (modulo-2).

3. Remainder $R(x)$ is the CRC checksum.

4. Transmit $$\displaystyle D(x) \cdot x^r + R(x) $$.

5. Receiver divides by same $G(x)$. Zero remainder = no error.

Example (Jun 2025): $$\displaystyle D(x)=x^5+x^4+x^2+1 $$ (binary 110101), $$\displaystyle G(x)=x^3+x+1 $$ (binary 1011). After division, remainder 011. Codeword: 110101011.

C. Data Link Protocols

D. Framing & Stuffing

Example (Nov 2022): Bit string 011110111110111101111110111101111110. After bit stuffing (insert '0' after 5 '1's): 011110111110111101011111101111010111110.

E. Flow Control Protocols

1. Stop-and-Wait:

• Sender sends 1 frame, waits for ACK before next.

• Efficiency $$\displaystyle \eta = \frac{1}{1 + 2a} $$, where $$\displaystyle a = \frac{\text{Propagation Delay}}{\text{Transmission Time}} $$.

• Problem: Low utilization for long propagation delays (high $a$).

2. Sliding Window:

• Sender Window ($$\displaystyle W_s $$): Max unacknowledged frames.

• Receiver Window ($$\displaystyle W_r $$): Max frames receiver can accept (usually 1 for Go-Back-N, >1 for Selective Repeat).

• Piggybacking: ACK carried in data frame going opposite direction (improves efficiency).

F. Multiple Access Protocols (MAC Sublayer)

Static Channel Allocation: FDM/TDM. Inefficient for bursty traffic.

Dynamic Allocation (Random Access):

Binary Exponential Backoff (BEB):

After $k$-th collision, wait random slots from $0$ to $$\displaystyle 2^k - 1 $$.

Example: 1st collision → 0 or 1 slot; 2nd → 0-3 slots; max $$\displaystyle k=10 $$ (1023 slots).

IV. LOCAL AREA NETWORKS (LAN) STANDARDS

A. IEEE 802 Standards Overview

Defines LAN/MAN standards. Sublayers:

• LLC (Logical Link Control): Common interface to network layer (IEEE 802.2).

• MAC (Media Access Control): Media-specific access method.

B. Ethernet (IEEE 802.3)

Access Method: CSMA/CD. Frame Format (64-1518 bytes):

[Preamble 7B][SFD 1B][Dest MAC 6B][Src MAC 6B][Type 2B][Data 46-1500B][FCS 4B]

• Preamble/SFD: Clock synchronization.

• Type: Indicates upper-layer protocol (e.g., 0x0800 = IPv4).

• FCS: CRC-32.

Physical Variants:

• 10BASE-T: 10 Mbps, Cat 3 TP, star topology, max segment 100m.

• 100BASE-TX: Fast Ethernet, 100 Mbps, Cat 5 TP.

• 1000BASE-T: Gigabit Ethernet, all 4 pairs, full-duplex.

Performance: Excellent for light load; efficiency drops with heavy load due to collisions. Minimum Frame Size: Ensures collision detection before transmission ends.

C. Token Bus (IEEE 802.4)

• Access: Token passing on a logical bus (physical can be bus/star).

• Operation: Station with token can transmit. Token passed by address (highest to lowest).

• Performance: Deterministic, good for factory automation. Complex token management.

D. Token Ring (IEEE 802.5)

• Access: Token passing on logical ring (physical star via MAU).

• Frame Format:

[SD][FC][DA][SA][DSAP][SSAP][Control][Data][FCS][ED][FS]

• SD (Starting Delimiter): 0xF7 (violates bit stuffing rule).

• FC (Frame Control): Indicates frame type.

• Ring Maintenance: Active Monitor (AM) handles token recovery, beaconing on failures.

• Priority: 8 priority levels via reservation bits.

• Performance: Deterministic, but complex and slower than Ethernet.

E. FDDI (Fiber Distributed Data Interface)

• Dual ring: Primary & secondary. Counter-rotating.

• Token passing: Like Token Ring, but higher speed (100 Mbps).

• Reliability: Single break doesn't fail network (ring wraps).

• Topology: Dual ring of stations (dual-attached).

F. Wireless LAN (IEEE 802.11)

MAC Sublayer:

• DCF (Distributed Coordination Function): Mandatory, CSMA/CA with RTS/CTS.

• PCF (Point Coordination Function): Optional, AP-centric, contention-free.

Frame Types:

• Management: Authentication, association (Beacon frames).

• Control: RTS, CTS, ACK.

• Data: Carries payload.

Interframe Spaces (IFS):

• SIFS (Short IFS): Highest priority (ACK, CTS).

• PIFS (PCF IFS): PCF frames.

• DIFS (DCF IFS): DCF data frames.

G. LAN Standards Comparison (802.3 vs 802.4 vs 802.5)

V. NETWORKING DEVICES AND BRIDGING

[!TIP] KEY DIFFERENCE: Hubs/Switches/Bridges operate at L2 (MAC addresses). Routers at L3 (IP addresses). Gateways at L7.

VI. NETWORK LAYER

A. Design Issues

Routing, congestion control, addressing, fragmentation/reassembly, error handling (ICMP).

B. IP Addressing

1. Classful Addressing (Legacy):

• Class A: 0 prefix, /8, 16M hosts. Range: 1.0.0.0 - 126.255.255.255

• Class B: 10 prefix, /16, 64K hosts. Range: 128.0.0.0 - 191.255.255.255

• Class C: 110 prefix, /24, 254 hosts. Range: 192.0.0.0 - 223.255.255.255

• Class D: 1110 prefix, multicast.

• Class E: 1111 prefix, experimental.

Limitations: Rapid depletion of Class B, inefficient use of addresses (many organizations needed <254 hosts but got /24).

2. CIDR (Classless Inter-Domain Routing):

• Notation: a.b.c.d/n (e.g., 192.168.1.0/24).

• Aggregation (Supernetting): Combine contiguous prefixes into larger block (e.g., 192.168.0.0/24 + 192.168.1.0/24 → 192.168.0.0/23).

• Solves routing table explosion & address waste.

3. Subnetting:

• Divide a network into smaller subnetworks.

• Subnet Mask: 32-bit number separating network/host bits (e.g., 255.255.255.0 = /24).

• Steps:

  1. Determine required subnets/hosts.

  2. Borrow bits from host part for subnet ID.

  3. New prefix = original prefix + borrowed bits.

  4. Subnet Increment = $$\displaystyle 2^{\text{borrowed bits}} $$.

  5. Calculate ranges: First usable = Network ID + 1; Last usable = Broadcast ID - 1; Broadcast = Network ID + Increment - 1.

Example (Jun 2023, May 2024): 192.168.10.0/24 into 4 subnets.

• Borrow 2 bits → new prefix /26 (mask 255.255.255.192).

• Increment = $$\displaystyle 2^2 = 64 $$.

• Subnets:

1. 192.168.10.0/26 → Range: 192.168.10.1 - 192.168.10.62, Broadcast: 192.168.10.63

2. 192.168.10.64/26 → Range: 65 - 126, Broadcast: 127

3. 192.168.10.128/26 → Range: 129 - 190, Broadcast: 191

4. 192.168.10.192/26 → Range: 193 - 254, Broadcast: 255

4. IPv4 vs IPv6:

C. Address Resolution

ARP (Address Resolution Protocol):

• Purpose: Map IP address → MAC address on local network.

• Operation:

  1. Host checks ARP cache.

  2. If missing, broadcasts ARP request: "Who has IP X? Tell Y (MAC of sender)."

  3. Owner replies with ARP reply (unicast): "IP X is at MAC Z."

  4. Both update ARP caches.

• RARP: Reverse ARP. Diskless node broadcasts its MAC, asks for IP (legacy, replaced by BOOTP/DHCP).

• InARP: Used in Frame Relay/ATM to map DLCI to IP.

• BOOTP/DHCP: Dynamic IP assignment (DHCP is enhanced BOOTP).

D. ICMP (Internet Control Message Protocol)

• Role: Network layer error-reporting & diagnostic protocol.

• Encapsulated in IP (Protocol number 1).

• Message Types:

  • Error Reporting:

    • Destination Unreachable (Port, Network, Host unreachable)

    • Time Exceeded (TTL expired, fragment reassembly timeout)

    • Parameter Problem (Invalid header field)

  • Query:

    • Echo Request/Reply (ping)

    • Timestamp Request/Reply

    • Router Advertisement/Solicitation (for autoconfiguration)

E. Routing Algorithms

1. Distance Vector Routing (Bellman-Ford):

• Principle: Each router knows distance (cost) to destination via each neighbor. Shares entire routing table with neighbors periodically.

• Algorithm: $$\displaystyle D_x(y) = \min_{v \in \text{neighbors}} \{ \text{cost}(x,v) + D_v(y) \} $$

• Example (Nov 2023): Given delay vectors from neighbors, router J computes new distances.

• RIP: Uses DV, hop count metric (max 15 hops). Problems: Count-to-infinity, slow convergence, routing loops.

• Solutions: Split horizon (don't advertise route back to source), Poison reverse (advertise infinite metric back), Hold-down timers.

2. Link State Routing (Dijkstra):

• Principle: Each router has complete map of network (link state database). Computes shortest path to all destinations using Dijkstra's algorithm.

• Steps:

  1. Flood LSA (Link State Advertisement) to all routers.

  2. Build identical LSDB (Link State Database).

  3. Run Dijkstra on LSDB.

• Algorithm (Shortest Path First - SPF):

  1. Set $N'$ = {source node}, cost to self = 0.

  2. Find node $w$ not in $N'$ with smallest tentative cost.

  3. Add $w$ to $N'$, update costs to neighbors via $w$.

  4. Repeat until all nodes in $N'$.

• Advantages: Fast convergence, no count-to-infinity, supports complex metrics.

• Disadvantages: High memory/computation (O($$\displaystyle n^2 $$)), flooding overhead.

Comparison:

VII. TRANSPORT LAYER

A. Services & Functions

• Segmentation & reassembly.

• Connection management (establishment, termination).

• Flow control (receiver-driven).

• Error control (retransmission).

• Multiplexing/demultiplexing (ports).

B. Connection-Oriented vs Connectionless

C. Transmission Control Protocol (TCP)

Header Format (20-60 bytes):

[Source Port 16][Dest Port 16][Seq Num 32][Ack Num 32][Data Offset 4][Reserved 6][Flags 6][Window 16][Checksum 16][Urgent Ptr 16][Options 0-40][Padding]

• Flags: URG, ACK, PSH, RST, SYN, FIN.

• Window: Receiver's advertised window size (flow control).

• Checksum: Covers header+data, pseudo-header (src IP, dst IP, protocol, length).

Connection Establishment (3-Way Handshake):

1. SYN: Client → Server, SEQ=x, SYN=1.

2. SYN-ACK: Server → Client, SEQ=y, ACK=x+1, SYN=1, ACK=1.

3. ACK: Client → Server, SEQ=x+1, ACK=y+1, ACK=1.

Connection Termination (4-Way Handshake - Graceful):

1. FIN: A → B, FIN=1, SEQ=u.

2. ACK: B → A, ACK=u+1, ACK=1.

3. FIN: B → A, FIN=1, SEQ=v.

4. ACK: A → B, ACK=v+1, ACK=1.

Why Graceful? Ensures all data delivered, resources released cleanly, prevents data loss.

Flow Control: Receiver advertises window size in ACK. Sender's send window = min( congestion window, receiver window ).

Congestion Control:

• Slow Start: Start with $$\displaystyle cwnd=1 $$ MSS, double each RTT until ssthresh.

• Congestion Avoidance: After ssthresh, increase $cwnd$ by 1 MSS per RTT (linear).

• Fast Retransmit: 3 duplicate ACKs → retransmit missing segment immediately.

• Fast Recovery: After fast retransmit, set ssthresh = cwnd/2, cwnd = ssthresh + 3, then linear increase.

Goal: Probe for available bandwidth, avoid network collapse.

D. User Datagram Protocol (UDP)

Header Format (8 bytes):

[Source Port 16][Dest Port 16][Length 16][Checksum 16]

• Length: Header + data (bytes).

• Checksum: Optional in IPv4, mandatory in IPv6. Covers pseudo-header.

• Characteristics: Connectionless, no reliability, no ordering, no congestion control.

• Applications: DNS (queries), VoIP, streaming, DHCP, SNMP.

E. TCP vs UDP Summary

VIII. APPLICATION LAYER

A. Domain Name System (DNS)

• Purpose: Hierarchical, distributed database mapping domain names → IP addresses.

• Hierarchy:

  • Root Servers (13 sets): Top-level.

  • TLD Servers: .com, .org, .in, country-code TLDs.

  • Authoritative Servers: Organizations' own servers (e.g., google.com).

  • Local DNS Resolver: ISP/organization's server.

• Resolution:

  • Recursive: Resolver asks server, server returns final answer (or error).

  • Iterative: Server returns referral to lower-level server if not authoritative.

• Caching: Each server caches responses with TTL (Time-To-Live). Reduces latency & traffic.

• Resource Records (RR):

  • A: IPv4 address.

  • AAAA: IPv6 address.

  • CNAME: Canonical name (alias).

  • MX: Mail exchange server.

  • NS: Name server.

  • PTR: Pointer (reverse lookup).

B. Electronic Mail (Email)

Architecture:

1. User Agent (UA): Email client (Outlook, Thunderbird).

2. Message Transfer Agent (MTA): Server that transfers email (sendmail, Postfix).

3. Message Delivery Agent (MDA): Delivers to mailbox (procmail).

Protocols:

• SMTP (Simple Mail Transfer Protocol) - Port 25:

  • Push protocol (client → server).

  • Commands: HELO, MAIL FROM, RCPT TO, DATA, QUIT.

  • ESMTP: Extended with EHLO, supports extensions (size, authentication, TLS).

• POP3 (Post Office Protocol v3) - Port 110:

  • Pull protocol. Downloads & deletes (or keeps) from server.

  • Simple, no folder support.

• IMAP (Internet Message Access Protocol) - Port 143/993 (SSL):

  • Pull protocol. Keeps mail on server, supports folders, partial fetch.

Message Format (RFC 5322):

Header:

From: [email protected]

To: [email protected]

Subject: Hello

Date: ...

Body:

This is the message.

C. World Wide Web (WWW)

• HTTP (Hypertext Transfer Protocol):

  • Request: Method SP Request-URI SP HTTP/1.1\r\n + headers.

  • Methods: GET (retrieve), POST (submit), PUT, DELETE.

  • Response: HTTP/1.1 Status-Code Reason-Phrase\r\n + headers + body.

  • Status Codes: 200 OK, 301 Moved Permanently, 404 Not Found, 500 Internal Server Error.

• HTTPS: HTTP over TLS/SSL (port 443). Provides encryption, server authentication.

• URL (Uniform Resource Locator): scheme://host:port/path?query#fragment

• Cookies: Key-value pairs stored by browser, sent with requests (session management).

• Sessions: Maintain state via cookies, URL rewriting, hidden fields.

D. File Transfer Protocol (FTP)

• Two Connections:

  1. Control Connection: Port 21, persistent, commands/responses.

  2. Data Connection: Port 20 (active mode) or dynamic port (passive mode), per file transfer.

• Active Mode: Server connects back to client's port 20.

• Passive Mode: Client initiates data connection to server's dynamic port (firewall-friendly).

• Commands: USER, PASS, LIST, RETR, STOR, QUIT.

E. Simple Network Management Protocol (SNMP)

• Components:

  • Manager: Central monitoring station.

  • Agent: Software on managed device (router, switch).

  • MIB (Management Information Base): Database of manageable objects (variables).

  • Management Station: Runs manager software.

• Operations:

  • get, get-next, set (from manager to agent).

  • trap (asynchronous alert from agent to manager).

• Versions:

  • SNMPv1: Community-based (no encryption).

  • SNMPv2c: Enhanced operations, still community-based.

  • SNMPv3: User-based, authentication & encryption (secure).

IX. MISCELLANEOUS TOPICS

A. Network Topologies

Star vs Mesh: Star is centralized, cost-effective for LANs. Mesh is fully decentralized, used in WAN backbones for reliability.

B. Virtual LANs (VLANs)

• Concept: Logical segmentation of a physical LAN into multiple broadcast domains.

• Benefits:

  • Security: Isolate sensitive groups.

  • Broadcast Reduction: Smaller broadcast domains.

  • Flexibility: Users can be grouped by function, not location.

• Tagging (IEEE 802.1Q): Inserts 4-byte VLAN tag in Ethernet frame header (between source MAC and EtherType). Tag includes VLAN ID (12 bits, 4094 VLANs max).

• Implementation: Switch ports configured as access (single VLAN, untagged) or trunk (multiple VLANs, tagged).

C. Network Design & Performance Metrics

• Throughput: Actual achieved data rate (bps).

• Bandwidth: Maximum theoretical capacity (bps).

• Delay (Latency): Time for a bit to travel from source to destination.

  • $$\displaystyle \text{Total Delay} = \text{Transmission Delay} + \text{Propagation Delay} + \text{Queuing Delay} + \text{Processing Delay} $$

• Reliability: Measured by MTBF (Mean Time Between Failures), error rates.

D. Service Primitives (OSI)

Four primitives for layer-to-layer communication:

1. Request: Service user → service provider (e.g., "connect").

2. Indication: Service provider → service user (e.g., "incoming call").

3. Response: Service user → service provider (e.g., "accept call").

4. Confirm: Service provider → service user (e.g., "connection confirmed").

Used in connection-oriented services (e.g., transport layer).

X. CALCULATION-BASED PROBLEMS (EXAM FOCUS)

1. Nyquist Capacity

$$C = 2B \log_2 M$$

Example: $$\displaystyle B=3 $$ kHz, $$\displaystyle M=4 $$ → $$\displaystyle C=12 $$ kbps. If $$\displaystyle M=8 $$ → $$\displaystyle C=18 $$ kbps.

2. Shannon Capacity

$$C = B \log_2 (1 + \text{SNR})$$

Example: $$\displaystyle B=4 $$ kHz, SNR=1000 → $C \approx 39.88$ kbps.

3. ALOHA Throughput

• Pure: $$\displaystyle S = G e^{-2G} $$, $$\displaystyle S_{max}=0.184 $$ at $$\displaystyle G=0.5 $$.

• Slotted: $$\displaystyle S = G e^{-G} $$, $$\displaystyle S_{max}=0.368 $$ at $$\displaystyle G=1 $$.

Given idle probability $$\displaystyle P_{idle}=0.1 $$ (slotted): $$\displaystyle P_{idle} = e^{-G} = 0.1 $$ → $$\displaystyle G = -\ln(0.1) \approx 2.302 $$. Throughput $$\displaystyle S = G e^{-G} = 2.302 \times 0.1 = 0.2302 $$. $$\displaystyle G>1 $$ → overload (more stations than optimal).

4. TDM Frame Size

Data Rate = $$\displaystyle \frac{\text{Total bits per frame}}{\text{Frame time}} $$.

Example (Jun 2023): 190 kbps + 180 kbps channels. Frame must carry bits from both. Frame size in bits = sum of bits from each channel per frame. Data rate = frame size / frame time. If no sync bits, and each channel sends 1 bit per frame, frame size = 190+180 = 370 bits. But typically, each channel contributes multiple bits. Clarify: if multiplexing n channels each with rate $$\displaystyle R_i $$, frame size $$\displaystyle L = \sum R_i \times T_{frame} $$, data rate = $$\displaystyle L / T_{frame} = \sum R_i $$.

5. CRC Polynomial Computation

Steps: Append zeros, divide by $G(x)$, remainder is CRC.

Example (Jun 2025): $$\displaystyle D(x)=x^5+x^4+x^2+1 $$ (110101), $$\displaystyle G(x)=x^3+x+1 $$ (1011). Division yields remainder 011. Codeword: 110101011.

6. Stop-and-Wait Efficiency

$$\eta = \frac{1}{1 + 2a}, \quad a = \frac{\text{Propagation Delay}}{\text{Transmission Time}}$$

For $\eta \ge 0.5$: $1/(1+2a) \ge 0.5$ → $1 \ge 0.5 + a$ → $a \le 0.5$. So propagation delay ≤ transmission time.

7. CSMA/CD Minimum Packet Size

$$\text{Min Frame Size} = 2 \times \text{Propagation Delay} \times \text{Bandwidth}$$

Example (Nov 2022): 2 km, bandwidth $$\displaystyle 10^7 $$ bps, signal speed $$\displaystyle 2 \times 10^8 $$ m/s.

Propagation delay = distance / speed = $$\displaystyle 2000 / (2 \times 10^8) = 10^{-5} $$ s = 10 µs.

Min frame size = $$\displaystyle 2 \times 10^{-5} \times 10^7 = 200 $$ bits = 25 bytes. But Ethernet minimum is 64 bytes → must send at least 64 bytes.

8. Subnetting Problems

Example (May 2024): ISP has 190.100.0.0/16 (65,536 addresses).

• Group 1: 64 customers × 256 addresses = 16,384 addresses → Need /20 blocks ($$\displaystyle 2^{16-4}=2^{12}=4096 $$? Wait: 256 hosts → need 8 host bits → /24 per customer. 64 × 256 = 16,384 addresses = 64 /24 blocks. Starting from 190.100.0.0/24 to 190.100.63.0/24.

• Group 2: 128 customers × 128 addresses → /25 per customer (128 hosts). 128 × 128 = 16,384 addresses = 128 /25 blocks. Next available: 190.100.64.0/25 to 190.100.127.127/25 (128 blocks).

• Group 3: 128 customers × 64 addresses → /26 per customer (64 hosts). 128 × 64 = 8,192 addresses = 128 /26 blocks. Next: 190.100.128.0/26 to 190.100.191.255/26 (128 blocks).

• Used: 16,384 + 16,384 + 8,192 = 40,960 addresses.

• Remaining: 65,536 - 40,960 = 24,576 addresses (190.100.192.0/18? Actually used up to 190.100.191.255, so remaining: 190.100.192.0 - 190.100.255.255 = 16,384 addresses? Let's recalc: /16 has 256 /24 blocks. Used: 64 (group1) + 128 (group2) + 128 (group3) = 320 /24 blocks? No, group2 uses /25 (half of /24), group3 uses /26 (quarter). Better to allocate in contiguous blocks. After group1 (64 /24s), used 0.0 to 63.255. Group2: 128 /25s = 64 /24s (since 2 /25s = 1 /24). So uses 64.0 to 127.255. Group3: 128 /26s = 32 /24s (4 /26s = 1 /24). Uses 128.0 to 159.255. Total used /24 blocks: 64+64+32=160. Remaining /24 blocks: 256-160=96 → 96 × 256 = 24,576 addresses. Yes.

END OF UNIT 2 NOTES