UNIT 2: COMPUTER NETWORKS - EXAM-FOCUSED SHORT NOTES

Based on rigorous analysis of RGPV past papers (2022-2025). Topics marked (Very High Frequency) are mandatory for 7-mark questions.

I. NETWORK FOUNDATIONS & REFERENCE MODELS

A. Network Criteria & Performance Metrics

• Throughput: Actual data delivery rate (bps).

• Delay (Latency): Sum of:

  • Transmission Delay: $$\displaystyle T_{tx} = \frac{\text{Packet Size (bits)}}{\text{Bandwidth (bps)}} $$

  • Propagation Delay: $$\displaystyle T_{prop} = \frac{\text{Distance (m)}}{\text{Propagation Speed (m/s)}} $$

  • Queuing Delay: Time waiting in buffer.

  • Processing Delay: Time to examine packet header.

• Bandwidth: Maximum theoretical data rate (Hz or bps).

• Efficiency/Utilization: $$\displaystyle \frac{\text{Useful Data}}{\text{Total Data Sent}} $$ or $$\displaystyle \frac{T_{prop}}{T_{prop} + T_{tx}} $$ for Stop-and-Wait.

B. Reference Models

1. ISO-OSI Seven-Layer Model (Very High Frequency)

• Peer-to-Peer Communication: Each layer at source provides service to the layer above and uses service of the layer below. Communication is between corresponding layers on different systems.

2. TCP/IP Protocol Suite / Five-Layer Model (Very High Frequency)

• Comparison: OSI vs. TCP/IP

  • Design: OSI is theoretical, protocol-independent. TCP/IP is practical, protocol-centric.

  • Layers: TCP/IP combines OSI's Session/Presentation/Application into one Application layer.

  • Implementation: TCP/IP protocols (IP, TCP) came first; OSI model was created later to standardize.

  • Advantage of TCP/IP: Ubiquitous, robust, scalable. Disadvantage: Less rigid layering, harder to replace protocols.

C. Network Types & Topologies

• Types: PAN (<10m), LAN (building/campus), MAN (city), WAN (country/globe), Internetwork (global).

• Physical vs. Logical Topology: Physical = actual wire layout. Logical = data flow path (e.g., Ethernet logical bus, physical star).

• Common Topologies:

  • Bus: Simple, single cable. Disadv: Single point of failure, difficult troubleshooting.

  • Star: Central hub/switch. Adv: Easy to manage, single link failure doesn't bring down network. Disadv: Hub/switch failure is critical.

  • Ring: Token passing. Disadv: Single failure breaks ring (unless dual ring like FDDI).

  • Mesh: Full/partial connectivity. Adv: High reliability, redundancy. Disadv: Expensive, complex.

  • Tree/Hybrid: Combination of topologies.

II. PHYSICAL LAYER & TRANSMISSION MEDIA

A. Guided Transmission Media

B. Unguided Transmission Media

• Radio: Omnidirectional, through walls, WiFi, cellular.

• Microwave: Directional, line-of-sight, point-to-point backbones.

• Infrared: Directional, short-range, line-of-sight, remote controls.

• Satellite: Long delay, wide coverage, broadcast.

C. Data Transmission Fundamentals

1. Nyquist Theorem & Shannon's Capacity Formula (High Frequency)

• Nyquist (Noiseless): Maximum bit rate for a noiseless channel of bandwidth B Hz using M signal levels.

$$R_{max} = 2B \log_2 M \text{ (bps)}$$

• Shannon (Noisy): Maximum theoretical capacity C (bps) for a channel of bandwidth B Hz with signal-to-noise ratio SNR.

$$C = B \log_2 (1 + SNR)$$

> [!TIP] **Key Insight:** Nyquist gives achievable rate with *M* levels. Shannon gives *ultimate limit* due to noise. To approach Shannon limit, need complex coding.

2. Line Coding Techniques (Medium Frequency)

• Unipolar: All positive (e.g., NRZ-L). Has DC component.

• Polar: Positive/Negative (e.g., NRZ-I, NRZ-L). Better than unipolar.

• Bipolar (AMI): 0=zero, 1=alternating +V/-V. No DC, easy error detection.

• Manchester: Transition in middle of bit period. 1=high-to-low, 0=low-to-high. Self-clocking.

• Differential Manchester: Transition at start of bit period. 0=transition, 1=no transition. More robust.

3. Multiplexing Techniques

• FDM: Different frequencies for different signals (Radio, TV).

• TDM: Different time slots. Synchronous TDM: Fixed slots, unused slots idle. Statistical TDM: Dynamic slot allocation.

  • Frame Size in TDM: Sum of bits from all input sources per cycle.

• WDM: Optical version of FDM. Different wavelengths on same fiber.

III. DATA LINK LAYER

A. Functions & Services

• Framing: Delineating packet boundaries (Byte/bit stuffing).

• Physical Addressing: MAC addresses in frame header.

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

• Flow Control: Matching sender/receiver speeds (Stop-and-Wait, Sliding Window).

• Link Management: Establishing, maintaining, releasing link.

B. Framing & Bit/Byte Stuffing (High Frequency)

• Byte-Oriented (Character Stuffing): Uses special flag byte (e.g., 0x7E). Insert escape byte (0x7D) before flag/escape in data.

• Bit-Oriented (Bit Stuffing): Uses flag pattern 01111110. After 5 consecutive 1s in data, insert a 0. Receiver removes stuffed 0 after 5 ones.

  Example: Data 01111110 → Stuffed: 011111010 (0 inserted after 5 ones). Flag added: 01111110 011111010 01111110.

C. Error Detection & Correction (Very High Frequency)

1. Error Detection Codes

• VRC/LRC: Single parity bit per char (VRC) or per block (LRC). Can detect single-bit errors.

• Checksum: Sum of data words (1's complement). Used in IP/TCP/UDP.

• CRC (Cyclic Redundancy Check) – CALCULATION MANDATORY.

  • Polynomials: Data $D(x)$, Generator $G(x)$ (e.g., 10011 for CRC-CCITT).

  • Process:

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

    2. Perform modulo-2 division (XOR) of augmented data by $G(x)$.

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

    4. Transmitted Codeword = Original Data + $R(x)$.

  • At Receiver: Divide received codeword by $G(x)$. If remainder = 0 → no error (assumed).

2. Error Correction Codes

• Hamming Code: Adds $k$ parity bits to $m$ data bits to correct single-bit errors. $$\displaystyle 2^k \ge m + k + 1 $$.

• Problem with Simple Parity: Can only detect odd number of errors, cannot correct. Hamming solves this.

D. Data Link Protocols & Flow Control (Very High Frequency)

1. Stop-and-Wait

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

• Efficiency (Link Utilization): $$\displaystyle U = \frac{1}{1 + 2a} $$ where $$\displaystyle a = \frac{T_{prop}}{T_{tx}} $$.

• For ≥50% efficiency: $$\displaystyle 1 + 2a \le 2 \Rightarrow a \le 0.5 \Rightarrow T_{prop} \le 0.5 T_{tx} $$.

  \boxed{T_{prop} \le 0.5 \times T_{tx}}

2. Sliding Window Protocols (Very High Frequency)

• Piggybacking: Attaching ACK to data frame going in reverse direction (used in full-duplex).

E. Multiple Access Protocols (MAC Sublayer) (Very High Frequency)

1. Static Allocation: FDMA, TDMA, CDMA (fixed assignment). 2. Random Access (Contention-Based):

• Pure ALOHA: Transmit anytime. Throughput $$\displaystyle S = G e^{-2G} $$. Max $$\displaystyle S_{max} = 0.184 $$ at $$\displaystyle G=0.5 $$.

• Slotted ALOHA: Transmit only at slot start. $$\displaystyle S = G e^{-G} $$. Max $$\displaystyle S_{max} = 0.368 $$ at $$\displaystyle G=1 $$.

  \boxed{S_{max}(\text{Slotted}) = \frac{1}{e} \approx 0.368}

• CSMA & Variants:

  • 1-persistent: Sense channel, if idle transmit immediately; if busy, sense continuously.

  • Non-persistent: Sense, if idle transmit; if busy, wait random time, retry.

  • p-persistent (slotted): Sense at slot start, if idle transmit with prob $p$, else defer to next slot.

  • Comparison: 1-persistent → high collision, high delay. Non-persistent → lower collision, higher delay. p-persistent → balances.

• Binary Exponential Backoff (BEB): After $$\displaystyle k^{th} $$ collision, choose random slot from $0$ to $$\displaystyle (2^k - 1) $$. $k$ capped (e.g., 10 for Ethernet).

3. Controlled Access: Reservation, Polling, Token Passing (brief).

F. LAN Standards & Protocols (IEEE 802) (Very High Frequency)

• Ethernet Minimum Frame Size Calculation:

  To detect collision, frame must be transmitted before signal reaches farthest end.

$$T_{tx} \ge 2 \times T_{prop} \Rightarrow \text{Min Frame Size} \ge 2 \times \text{Propagation Distance} \times \text{Bandwidth}$$

> **Example:** For 2.5 km, speed $$\displaystyle 2 \times 10^8 $$ m/s, 1 Gbps: $$\displaystyle T_{prop} = 12.5 \mu s $$, Min Frame = $$\displaystyle 2 \times 12.5 \mu s \times 10^9 $$ bps = **25,000 bits ≈ 3125 bytes**.

G. Data Link Layer Devices & Protocols

1. Bridges & Switches (Very High Frequency)

• Layer 2 devices. Forward based on MAC address.

• Transparent Bridging (Switch Operation):

  1. Learning: Build MAC table from source MAC of incoming frames.

  2. Forwarding: If dest MAC in table, forward out specific port; else flood.

  3. Filtering: If src & dest on same port, discard.

• Spanning Tree Protocol (STP): Prevents loops in redundant topologies by blocking redundant paths.

• Comparison: Hub (Physical layer, broadcasts to all). Bridge/Switch (Data Link, intelligent forwarding, reduces collisions).

2. Virtual LANs (VLANs) (Medium Frequency)

• Concept: Logically group users/devices into separate broadcast domains on same physical switch.

• Benefits: Security, segmentation, reduced broadcast traffic.

• Tagging (IEEE 802.1Q): Inserts 4-byte VLAN tag in Ethernet frame header.

3. Point-to-Point Protocols

• HDLC (High-Level Data Link Control): Bit-oriented, synchronous. Frame: Flag 01111110, Address, Control, Info, FCS, Flag.

• PPP (Point-to-Point Protocol): For serial links (dial-up, DSL).

  • Phases: Link Establishment (LCP), Authentication (PAP/CHAP), Network Layer Protocol (NCP).

  • Frame: Flag, Address (0xFF), Control (0x03), Protocol (e.g., 0x0021 for IP), Data, FCS, Flag.

• SLIP: Older, simple, no error detection, no multiplexing.

H. Data Link Layer Issues

• Error Control vs. Flow Control: Error control ensures data integrity (retransmission). Flow control prevents overwhelming receiver (window size).

• Framing Efficiency: Overhead from headers/trailers. Bit stuffing adds minimal overhead.

IV. NETWORK LAYER

A. Functions & Design Issues

• Routing: Determining path (algorithms: DVR, LSR).

• Forwarding: Moving packet from input to output link (using forwarding table).

• Congestion Control: Preventing network overload.

• Internetworking: Connecting heterogeneous networks (routers).

• Addressing: Logical (IP) addresses.

B. IP Addressing (Very High Frequency)

1. IPv4

• Classful Addressing (Legacy):

  | Class | First Bit | Range (First Octet) | Default Mask | Networks | Hosts/Net | | :--- | :--- | :--- | :--- | :--- | :--- | | A | 0 | 1-126 | 255.0.0.0 | 128 | 16M | | B | 10 | 128-191 | 255.255.0.0 | 16K | 64K | | C | 110 | 192-223 | 255.255.255.0 | 2M | 254 | | D | 1110 | 224-239 | - | Multicast | - | | E | 1111 | 240-255 | - | Experimental | - |

• Limitations: Wasted addresses, no subnetting, routing table explosion.

• Classless Inter-Domain Routing (CIDR) & Subnetting (Very High Frequency)

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

  • Subnet Mask: 255.255.255.0 for /24. Convert prefix to dotted decimal.

  • Subnet Design Problem (Step-by-Step):

    1. Given network X.Y.Z.0/N and need S subnets.

    2. Borrow bits: $$\displaystyle 2^b \ge S $$ → $b$ = bits to borrow from host part.

    3. New Subnet Mask: Original mask + $b$ ones in host part.

    4. Subnet Increment: $$\displaystyle 2^{(32-N-b)} $$.

    5. Subnets: X.Y.Z.0, X.Y.Z.Increment, X.Y.Z.2*Increment...

    6. Each Subnet:

       • Network Address: First address (all host bits 0).

       • Usable Range: Network+1 to Broadcast-1.

       • Broadcast Address: Last address (all host bits 1).

  • Example: 192.168.10.0/24 → 4 subnets.

    • Borrow $$\displaystyle b=2 $$ bits (2^2=4). New mask: /26 (255.255.255.192).

    • Increment = $$\displaystyle 2^{(32-24-2)} = 64 $$.

    • Subnets: 192.168.10.0/26, 192.168.10.64/26, 192.168.10.128/26, 192.168.10.192/26.

    • First subnet usable: 192.168.10.1 - 192.168.10.62, Broadcast: 192.168.10.63.

  • VLSM (Variable Length Subnet Mask): Allocate different size subnets from same block (e.g., /26, /27, /28).

2. IPv6

• Motivation: Address exhaustion, simpler header (no checksum, fixed 40-byte), built-in QoS (flow label), security (IPsec mandatory), autoconfiguration.

• Format: 128-bit, hexadecimal, colon-separated (2001:0db8:85a3::8a2e:0370:7334). :: compresses consecutive zeros.

• Header: Version, Traffic Class, Flow Label, Payload Length, Next Header, Hop Limit, Source/Dest Address (128-bit each). No options (use extension headers).

• Comparison IPv4 vs IPv6:

  | Feature | IPv4 | IPv6 | | :--- | :--- | :--- | | Address Size | 32-bit | 128-bit | | Header | Variable (20-60 bytes) | Fixed 40 bytes | | Checksum | Yes (header) | No | | Fragmentation | Router & Source | Source only | | Options | Header options | Extension headers | | Address Notation | Dotted decimal | Hexadecimal colon-separated | | Autoconfig | Manual/DHCP | Stateless autoconfig (SLAAC) |

C. Address Resolution & Mapping

1. ARP (Address Resolution Protocol) (Very High Frequency)

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

• Operation:

  1. Host checks ARP cache. If miss, broadcasts ARP Request: "Who has IP X.X.X.X? Tell Y.Y.Y.Y (my IP)".

  2. Host with IP X.X.X.X unicasts ARP Reply: "X.X.X.X is at MAC AA:BB:CC:DD:EE:FF".

  3. Both update ARP caches.

• Proxy ARP: Router answers ARP for another host's IP (to hide subnet structure).

2. RARP (Reverse ARP) (High Frequency)

• Purpose: Map MAC address → IP address (for diskless workstations).

• Operation: Diskless station broadcasts its MAC, RARP server replies with assigned IP.

D. Routing Algorithms (Very High Frequency)

1. Distance Vector Routing (DVR) – Bellman-Ford (e.g., RIP)

• Principle: Each router knows distance (cost) to neighbors. Shares its entire routing table with neighbors periodically.

• Bellman-Ford Equation: $$\displaystyle D_x(y) = \min_{v \in \text{neighbors}} \left\{ c(x,v) + D_v(y) \right\} $$

  • $$\displaystyle D_x(y) $$ = cost from $x$ to $y$.

  • $c(x,v)$ = cost from $x$ to neighbor $v$.

• Algorithm Steps:

  1. Initialize: $$\displaystyle D_x(y) = c(x,y) $$ if neighbor, else $\infty$; $$\displaystyle D_x(x)=0 $$.

  2. Iterate: For each destination $y$, update $$\displaystyle D_x(y) = \min \left[ D_x(y), c(x,v) + D_v(y) \right] $$ for all neighbors $v$.

  3. Repeat until no changes (convergence).

• Problems:

  • Count-to-Infinity: Slow convergence for a failed link. Solved by Split Horizon (don't advertise route back to source) and Poisoned Reverse (advertise infinite cost back).

  • Slow Convergence.

  • Routing Loops.

2. Link State Routing (LSR) – Dijkstra's SPF (e.g., OSPF)

• Principle: Each router has complete map (LSDB) of network topology (link costs).

• Steps:

  1. Discover Neighbors & Link Cost: Hello packets.

  2. Build Link State Packets (LSP): Each router creates packet with its links and costs.

  3. Flood LSPs: Distribute LSPs to all other routers (reliable flooding).

  4. Build LSDB: Each router collects all LSPs → complete topology map.

  5. Compute Shortest Path: Run Dijkstra's algorithm on LSDB.

• Dijkstra's Algorithm (SPF):

  1. Set $$\displaystyle N' = \{\text{source}\} $$, $$\displaystyle C(\text{source})=0 $$, $$\displaystyle C(\text{others})=\infty $$.

  2. Find node $w$ not in $N'$ with smallest $C(w)$.

  3. Add $w$ to $N'$. Update costs for neighbors $v$: $$\displaystyle C(v) = \min[C(v), C(w) + \text{cost}(w,v)] $$.

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

• Advantages: Fast convergence, no count-to-infinity, supports VLSM/CIDR. Disadv: More memory/CPU, flooding overhead.

3. Comparison: DVR vs. LSR

E. Congestion Control in Network Layer

• Causes: Too many sources sending, slow processors, low bandwidth buffers.

• General Principles & Techniques:

  1. Traffic-Aware Routing: Avoid congested paths.

  2. Admission Control: New connections denied if network congested.

  3. Traffic Shaping: Smooth bursty traffic.

     • Leaky Bucket: Constant output rate, input burst buffered/dropped.

     • Token Bucket: Tokens arrive at rate $r$. Bursts up to $b$ tokens allowed. More flexible.

       Comparison: Leaky Bucket = rigid, smooth output. Token Bucket = allows bursts, better for bursty sources.

  4. Load Shedding: Discard packets (e.g., based on priority).

  5. Congestion Signaling: Implicit (packet loss/delay) or Explicit (choke packets, ECN bits in IP header).

F. Internet Control & Support Protocols

1. ICMP (Internet Control Message Protocol) (Very High Frequency)

• Purpose: Network layer protocol for error reporting & diagnostics.

• Common Messages:

  • Destination Unreachable (Port, Protocol, Network).

  • Time Exceeded (TTL expired – traceroute).

  • Echo Request/Reply (ping).

  • Redirect (better next-hop).

• Encapsulation: Inside IP datagram (Protocol number 1).

2. IPv4 Header Format (Medium Frequency)

0                   1                   2                   3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1

+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

|Version|  IHL  |Type of Service|          Total Length         |

+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

|        Identification         |Flags|      Fragment Offset    |

+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

|  Time to Live |    Protocol   |         Header Checksum       |

+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

|                       Source Address                          |

+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

|                    Destination Address                        |

+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

|                    Options                    |    Padding    |

+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

• Key Fields: Version (4), IHL (header length in 32-bit words), Total Length, Identification/Flags/Fragment Offset (fragmentation), TTL, Protocol (6=TCP, 17=UDP), Header Checksum, Source/Dest IP.

G. Network Layer Devices

• Router: Layer 3. Forwards packets based on IP address using routing table. Connects different networks.

• Gateway: General term for device that connects networks using different protocols (often application-layer gateway, e.g., email gateway).

V. TRANSPORT LAYER

A. Services & Primitives

• Process-to-Process Delivery: Uses port numbers (16-bit) to identify application processes.

• Multiplexing: Multiple application processes → single network connection.

• Demultiplexing: Incoming segments delivered to correct socket (IP + Port).

• Connection-Oriented (TCP): Handshake, reliable, ordered.

• Connectionless (UDP): No setup, best-effort.

B. User Datagram Protocol (UDP) (High Frequency)

• Characteristics: Unreliable, connectionless, no flow/congestion control, no ordering, minimal overhead.

• Header (8 bytes):

  
  0      7 8     15 16    23 24    31
  
  +--------+--------+--------+--------+
  
  |     Source Port    |   Dest Port   |
  
  +--------+--------+--------+--------+
  
  |      Length       |    Checksum    |
  
  +--------+--------+--------+--------+
  
  |          Data (optional)          |
  
  +-----------------------------------+
  
  

• Applications: DNS, VoIP, streaming, DHCP, SNMP (where speed > reliability).

C. Transmission Control Protocol (TCP) (Very High Frequency)

1. TCP Header Format (High Frequency)

 0                   1                   2                   3

 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1

+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

|          Source Port          |       Destination Port        |

+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

|                        Sequence Number                        |

+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

|                    Acknowledgment Number                      |

+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

|  Data |           |U|A|P|R|S|F|                               |
| Offset| Reserved  |R|C|S|S|Y|I|            Window             |
|       |           |G|K|H|T|N|N|                               |

+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

|            Checksum           |         Urgent Pointer        |

+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

|                    Options                    |    Padding    |

+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

|                             data                              |

+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

• Key Fields:

  • Seq/Ack Number: Byte-stream numbering (not packet number).

  • Flags:

    • URG (Urgent pointer valid)

    • ACK (Ack field valid – set after 3-way handshake)

    • PSH (Push function – deliver to app immediately)

    • RST (Reset connection)

    • SYN (Synchronize sequence numbers – connection setup)

    • FIN (Terminate connection)

  • Window Size: Receiver's advertised window (bytes).

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

2. Connection Management

• Three-Way Handshake (Establishment):

  1. Client → Server: SYN=1, Seq=x

  2. Server → Client: SYN=1, ACK=1, Seq=y, Ack=x+1

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

  Why 3-way? Prevents old duplicate SYN segments from causing invalid connections.

• Four-Way Handshake (Termination – Graceful):

  1. Client → Server: FIN=1, Seq=u

  2. Server → Client: ACK=1, Ack=u+1 (may still send data)

  3. Server → Client: FIN=1, Seq=v, Ack=u+1

  4. Client → Server: ACK=1, Seq=u+1, Ack=v+1

  Why Graceful? Ensures all data is delivered before closing. Each side closes independently.

3. Flow Control

• Sliding Window (Receiver-Advertised): Receiver tells sender its available buffer (Window Size in header). Sender cannot send more than Window bytes beyond last ACKed byte.

• Window can shrink (zero window) if buffer full. Sender must stop and probe periodically.

4. Congestion Control (Very High Frequency)

• Need: Prevent network collapse due to overloaded routers.

• Variables:

  • cwnd (Congestion Window): Sender's limit based on network congestion (in bytes).

  • ssthresh (Slow Start Threshold): Threshold to switch from Slow Start to Congestion Avoidance.

  • rwnd (Receiver Window): Receiver's advertised limit.

  • Effective Window: min(cwnd, rwnd).

• Algorithms:

  1. Slow Start:

     • Initially cwnd = 1 MSS (Maximum Segment Size).

     • For each ACK received: cwnd += 1 MSS.

     • Exponential growth until cwnd >= ssthresh.

  2. Congestion Avoidance:

     • When cwnd >= ssthresh.

     • For each RTT: cwnd += 1 MSS (Additive Increase).

  3. Fast Retransmit & Fast Recovery:

     • On 3 duplicate ACKs (indicates a single packet loss):

       • ssthresh = cwnd / 2

       • cwnd = ssthresh + 3 MSS (inflate for packets in flight)

       • Retransmit missing segment immediately.

     • On new ACK (acknowledges new data after fast recovery):

       • cwnd = ssthresh (back to congestion avoidance).

5. TCP Timer Management (Brief)

• RTO (Retransmission Timeout): Calculated from RTT samples (Jacobson/Karels algorithm).

• Persist Timer: Prevents zero-window deadlock (probe window).

• Keepalive Timer: Detect dead connections.

D. Comparison: TCP vs. UDP (Very High Frequency)

VI. APPLICATION LAYER & APPLICATION PROTOCOLS

A. Domain Name System (DNS) (Very High Frequency)

• Purpose: Hierarchical, distributed database translating domain names → IP addresses (and vice versa).

• Components:

  • Resolvers: Client-side library/utility (nslookup, dig).

  • Name Servers: Authoritative (holds zone files), Root, TLD (.com, .org).

  • Zones: Administrative domain (e.g., example.com).

• Resolution Process:

  1. Resolver queries local DNS server (usually ISP's).

  2. If not cached, local server may query Root server → gets TLD server for .com.

  3. TLD server gives authoritative server for example.com.

  4. Authoritative server returns IP.

  5. Caching: Each server caches results for TTL period.

• Query Types:

  • Recursive: Server does entire lookup for client (typical from resolver to local server).

  • Iterative: Server returns best known answer (e.g., root → TLD).

B. World Wide Web & HTTP

1. HTTP (Hypertext Transfer Protocol) (High Frequency)

• Architecture: Client/Server (Browser/Web Server).

• Request Message:

  
  GET /index.html HTTP/1.1\r\n
  
  Host: www.example.com\r\n
  
  [Other headers]\r\n
  
  \r\n
  
  [Body for POST]
  
  

• Response Message:

  
  HTTP/1.1 200 OK\r\n
  
  Content-Type: text/html\r\n
  
  [Other headers]\r\n
  
  \r\n
  
  [HTML Body]
  
  

• HTTP/1.0 vs. 1.1 vs. 2:

  • 1.0: New TCP connection for each request (non-persistent).

  • 1.1: Persistent connections (keep-alive), pipelining, host header (virtual hosting).

  • 2: Binary framing, multiplexing (multiple streams over one connection), header compression.

2. WWW Architecture: URLs (scheme://host:port/path), HTML, browsers, web servers.

C. Electronic Mail

1. Email Architecture & Format

• Agents:

  • MUA (Mail User Agent): Outlook, Thunderbird (composes/reads).

  • MTA (Mail Transfer Agent): Sendmail, Postfix (transfers between servers – SMTP).

  • MDA (Mail Delivery Agent): Procmail (delivers to mailbox).

• Message Format (RFC 822):

  
  Header:
  
  From: [email protected]
  
  To: [email protected]
  
  Subject: Hello
  
  Date: ...
  
  ...
  
  [Blank line]
  
  Body:
  
  This is the message text.
  
  

2. SMTP (Simple Mail Transfer Protocol) (High Frequency)

• Operation: Push protocol (MUA→MTA, MTA→MTA). Port 25.

• Commands/Responses:

  
  S: 220 service ready
  
  C: HELO client.example.com
  
  S: 250 Hello
  
  C: MAIL FROM:<[email protected]>
  
  S: 250 OK
  
  C: RCPT TO:<[email protected]>
  
  S: 250 OK
  
  C: DATA
  
  S: 354 Start mail input; end with <CRLF>.<CRLF>
  
  C: [Message headers and body]
  
  C: .
  
  S: 250 OK: queued
  
  C: QUIT
  
  S: 221 Bye
  
  

• ESMTP (Extended SMTP): Adds EHLO command and extensions (e.g., SIZE, STARTTLS for encryption).

3. POP3 & IMAP (Pull Protocols):

• POP3: Downloads & deletes from server (simple, offline).

• IMAP: Keeps mail on server, allows folder management, multiple clients.

D. Network Management – SNMP (Medium Frequency)

• Architecture:

  • Manager: Central console (e.g., snmpwalk).

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

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

• Operations:

  • Get: Retrieve value.

  • GetNext: Retrieve next object in MIB (for table traversal).

  • Set: Change value.

  • Trap: Unsolicited notification from agent to manager (e.g., link down).

E. Other Protocols (Brief)

• FTP: Two connections – Control (TCP port 21, commands) and Data (TCP port 20, separate connection per transfer). Active vs. Passive mode.

• DHCP: Dynamic IP assignment. DORA process: Discover, Offer, Request, Acknowledge.

• P2P: Decentralized (BitTorrent – tit-for-tat, trackers) vs. Centralized (Napster).

VII. INTERNETWORKING & SWITCHING

A. Switching Techniques (High Frequency)

B. Internetworking Devices (Very High Frequency)

Key Distinction: Hub (Layer 1, broadcasts). Switch (Layer 2, MAC-based, reduces collisions). Router (Layer 3, IP-based, connects networks). Gateway (Application layer, protocol conversion).

END OF UNIT 2 NOTES
Always verify calculations (subnetting, CRC, TDM, throughput) with units. Practice routing algorithm steps (Bellman-Ford, Dijkstra) with small graphs. Understand TCP state transitions (ESTABLISHED, FIN_WAIT, TIME_WAIT).