Computer Networks (IT-502) - Unit 3 Short Notes

UNIT 3: Computer Networks - Comprehensive Short Notes

I. Network Models and Architecture

A. ISO-OSI Reference Model

A 7-layer architectural framework for network communication, promoting interoperability through standardized layer functions.

Service Primitives: Four operations a layer provides to the layer above it:

1. REQUEST: Upper layer requests a service.

2. INDICATION: Lower layer notifies upper layer of an event (e.g., incoming data).

3. RESPONSE: Upper layer replies to an indication.

4. CONFIRM: Lower layer confirms completion of a requested service.

Peer-to-Peer Communication: Each layer on the source machine communicates with its peer layer on the destination machine using a Protocol Data Unit (PDU). The PDU name changes per layer (Data → Segment → Packet → Frame → Bit).

[!TIP] Exam Focus: OSI layers, their primary functions, and associated devices are frequently asked. Remember: Switches operate at Layer 2, Routers at Layer 3.

B. TCP/IP Reference Model

A practical, 4-layer model that underpins the modern Internet.

Comparison: OSI vs. TCP/IP

Advantages of TCP/IP:

• Interoperability: De facto global standard.

• Scalability: Proven in massive networks (Internet).

• Flexibility: Can run over any physical network.

Disadvantages:

• Lacks Generality: Tied to specific protocols.

• Layered Boundaries: Blurred (e.g., HTTP uses TCP/IP directly).

• No Strong encapsulation: Original model didn't strictly enforce layer independence.

C. Connection-Oriented vs. Connectionless Services

D. Network Types and Topologies

Network Types (by Scale):

• PAN (Personal Area Network): < 10m (e.g., Bluetooth, USB).

• LAN (Local Area Network): Single building/campus (e.g., Ethernet).

• WAN (Wide Area Network): Country/continent (e.g., Internet backbone).

• Internetwork: Interconnection of multiple networks (e.g., the Internet).

Physical vs. Logical Topology:

• Physical: Actual wire/cable layout.

• Logical: Path of signal flow (may differ from physical).

E. Network Devices

[!TIP] Common Pitfall: Hubs are repeaters (Layer 1), not "dumb switches." Switches build MAC tables dynamically.

II. Physical and Data Link Layer

A. Transmission Media

Guided Media (Wired):

• Twisted Pair (UTP/STP): 8 wires in pairs; categories (Cat5e, Cat6). Used in Ethernet, telephone.

• Coaxial Cable: Central conductor, insulator, shield, jacket. Used in cable TV, older Ethernet.

• Fiber Optic: Glass/plastic fibers; light pulses. Single-mode (long distance), Multi-mode (short distance). High bandwidth, immune to EMI.

Unguided Media (Wireless):

• Radio: Omnidirectional; used in Wi-Fi, Bluetooth.

• Microwave: Directional; terrestrial & satellite.

• Infrared: Line-of-sight; short range (remote controls).

• Satellite: Broad coverage, high latency.

B. Physical Layer Concepts

1. Channel Capacity Theorems:

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

$$ \text{Max Bit Rate} = 2B \log_2 M \ \text{bps} $$

*Example:* For B=3kHz, M=4 (2 bits/signal), Max Rate = 2*3000*log₂(4) = **12,000 bps**.

• Shannon Theorem (Noisy): Maximum data rate for a channel of bandwidth B Hz with SNR:

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

where SNR is linear (not dB). **This is the absolute theoretical limit.**

2. Line Coding (Mapping bits to signals):

• NRZ (Non-Return to Zero): High=1, Low=0. DC component issue.

• Manchester: Mid-bit transition. Clock recovery. Used in Ethernet (10BASE-T).

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

• 4B/5B: 4 data bits encoded as 5-bit code (only 16 of 32 codes used). Ensures enough transitions for clocking. Used in Fast Ethernet (100BASE-TX).

3. Multiplexing:

• FDM (Frequency Division): Each signal gets a frequency band. Used in radio, TV, DSL.

• TDM (Time Division): Each signal gets a time slot (time slot = sample of all channels). Synchronous TDM: Fixed slots, even if channel idle. Statistical TDM: Dynamic slot allocation based on demand.

• WDM (Wavelength Division): Fiber version of FDM; multiple light wavelengths on same fiber.

C. Data Link Layer Functions & Design Issues

1. Framing: Delimiting data units (frames).

• Character-Oriented: Use special characters (e.g., STX/ETX in BISYNC).

• Bit-Oriented: Use bit patterns (e.g., 01111110 flag in HDLC).

2. Error Detection & Correction:

• Parity Bits: Single bit for odd/even parity. Detects single-bit errors only.

• CRC (Cyclic Redundancy Check):

  • Polynomial Representation: Data D(x) and Generator G(x) (e.g., 10011 for CRC-CCITT).

  • Computation:

    1. Append r zeros to D(x), where r = degree(G(x)).

    2. Divide augmented D(x) by G(x) using modulo-2 division (XOR, no carries).

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

    4. Transmit: D(x) * x^r + R(x).

    5. Check: Receiver divides received polynomial by G(x). Remainder 0 = no error.

  • Example: Given in past papers (e.g., 1101011011 and 10011).

• Hamming Codes (Error Correction): Adds k parity bits to m data bits to correct single-bit errors. Distance d_min = 3. Positions 1,2,4,... are parity bits.

3. Bit/Byte Stuffing:

• Bit Stuffing (Bit-Oriented): In HDLC, after 5 consecutive 1s in data, insert a 0. Receiver removes it after detecting 01111110.

• Byte Stuffing (Byte-Oriented): In PPP, if flag byte 0x7E or escape byte 0x7D appears in data, insert escape byte before it and XOR the data byte.

D. Flow Control

1. Stop-and-Wait:

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

• Efficiency (U): $$\displaystyle U = \frac{1}{1 + 2a} $$ where $$\displaystyle a = \frac{\text{Propagation Delay}}{\text{Frame Transmission Time}} $$

• For U ≥ 50%: $ 1 + 2a \leq 2 \Rightarrow a \leq 0.5 $. So, Propagation Delay ≤ Frame Transmission Time.

2. Sliding Window:

• Go-Back-N (GBN):

  • Sender window size = N, Receiver window size = 1.

  • If ACK not received: Sender re-transmits frame n and all subsequent frames (n+1, n+2...).

  • Link Utilization: $$\displaystyle \approx \frac{N}{1+2a} $$ (for large N).

• Selective Repeat (SR):

  • Sender & Receiver window size = N (max sequence number = 2N).

  • If ACK not received: Sender re-transmits only frame n.

  • Requires individual ACKs and buffering out-of-order frames at receiver.

  • More efficient than GBN on lossy links, but more complex.

Piggybacking: Putting ACK on outgoing data frames instead of sending separate ACK frames. Improves efficiency in full-duplex links.

E. Multiple Access Protocols (MAC Sublayer)

1. Channel Allocation:

• Static: FDM, TDM, WDMA. Fixed assignment; inefficient for bursty traffic.

• Dynamic: Random access, reservation, polling, token passing.

2. Random Access (Contention-Based):

• Pure ALOHA: Transmit anytime. Max Efficiency = 18.4%. Throughput S = G * e^{-2G}.

• Slotted ALOHA: Time synchronized; transmit only at slot start. Max Efficiency = 36.8%. Throughput S = G * e^{-G}.

• CSMA (Carrier Sense Multiple Access):

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

  • Non-Persistent: If busy, wait random time, then sense again.

  • p-Persistent (Slotted): If idle, transmit with probability p, else defer to next slot.

• CSMA/CD (Collision Detection - Ethernet):

  • Procedure: Sense → Transmit → Detect collision? → If yes, jam signal → Binary Exponential Backoff → Retransmit.

  • Minimum Frame Time: Must be ≥ 2 * Max Propagation Delay (for collision detection).

• CSMA/CA (Collision Avoidance - Wi-Fi 802.11):

  • DCF (Distributed Coordination Function): Uses RTS/CTS (optional) to avoid hidden terminal problem. Inter-frame spaces (SIFS, DIFS) for priority.

  • PCF (Point Coordination Function): Centralized, contention-free (optional).

3. Binary Exponential Backoff (BEB):

After k-th collision (k ≤ 10), choose random slot from 0 to (2^k - 1). After 10 collisions, k frozen at 10. Reduces collision probability after repeated failures.

F. LAN Standards (IEEE 802)

G. Switching Techniques

H. Bridges and Switching

1. Transparent Bridging (Local Bridge - Ethernet):

• Learning: Builds MAC address table by recording source MAC & incoming port.

• Forwarding: If destination MAC in table → forward to that port; else flood to all except incoming.

• Filtering: If source & destination on same port → discard frame.

• Loop Prevention: Uses Spanning Tree Protocol (STP) to block redundant paths.

2. Remote Bridges:

Connect LANs over WAN links. Challenges:

• Latency: Long propagation delays break CSMA/CD assumptions.

• Scalability: STP may block needed paths.

• Loop Prevention: More complex over WAN.

3. Switches (Multiport Bridges):

• MAC Table Building: Same learning process as bridges.

• Switching Methods:

  • Store-and-Forward: Error-checked; default for most switches.

  • Cut-Through: Low latency; used in high-frequency trading.

I. Data Link Layer Protocols

1. HDLC (High-Level Data Link Control):

• Frame Format:

  
  [Flag: 01111110] [Address] [Control] [Info] [FCS] [Flag: 01111110]
  
  

• Control Field: Defines frame type:

  • I-Frame (Info): 00 (seq #, ack #) – data transfer.

  • S-Frame (Supervisory): 10 (seq #, ack #) – flow control (RR, RNR, REJ).

  • U-Frame (Unnumbered): 11 (modifier) – link management (SNRM, DISC, UA).

• Modes: NRM (Normal Response), ABM (Asynchronous Balanced), ARM (Asynchronous Response).

2. PPP (Point-to-Point Protocol):

• Frame Format:

  
  [Flag: 0x7E] [Address: 0xFF] [Control: 0x03] [Protocol] [Data] [FCS] [Flag: 0x7E]
  
  

• Phases:

  1. Link Establishment (LCP): Configure link parameters.

  2. Authentication (Optional): PAP (plaintext) or CHAP (challenge-handshake).

  3. Network Layer Protocol (NCP): Enable network protocols (e.g., IPCP for IP).

  4. Link Termination.

• Applications: Dial-up, DSL, direct router-to-router links.

3. SLIP (Serial Line IP):

• Limitations: No error detection, no authentication, no protocol type field (only IP), no compression. Obsolete, replaced by PPP.

4. Frame Relay:

• Architecture: DTE (user device) ↔ DCE (Frame Relay switch).

• Frame Format: No MAC addresses; uses DLCI (Data Link Connection Identifier) to identify virtual circuit.

• Virtual Circuits: PVC (Permanent) pre-configured; SVC (Switched) dynamic.

• Congestion Management: DE (Discard Eligibility) bit; FECN/BECN bits to signal congestion to sender/receiver.

III. Network Layer

A. Routing Algorithms

1. Distance Vector Routing (DVR - Bellman-Ford):

• Principle: Each router knows distance (cost) to destination and next-hop. Shares its entire table with neighbors periodically.

• Equation: $$\displaystyle D_x(y) = \min_v \{ c(x,v) + D_v(y) \} $$ where x is current router, v is neighbor, y is destination.

• Example: Given a graph with costs, each router iteratively updates its table based on neighbor's vectors.

• Problems: Count-to-infinity, slow convergence, routing loops.

• RIP (Routing Information Protocol): Uses DVR, hop count metric (max 15 hops). Solutions: Split horizon (don't advertise route back to source), Poison reverse (advertise infinite metric back).

2. Link State Routing (LSR - Dijkstra's SPF):

• Principle: Each router has complete map of network (link state database). Runs Dijkstra's algorithm to compute shortest path tree.

• Steps:

  1. Flooding: LSAs (Link State Advertisements) flooded to all routers.

  2. Database Sync: All routers have identical LSDB.

  3. SPF Computation: Dijkstra's on LSDB from self as root.

• Example: Given graph with link costs, compute shortest path tree.

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

• Disadvantages: High memory/CPU overhead on large networks.

DVR vs LSR Comparison:

B. Routing Metrics & Path Determination

• Metrics: Hop count, bandwidth, delay, cost, reliability, load.

• Best Path: Based on metric (e.g., shortest path = min sum of costs).

• Multi-Path Routing: ECMP (Equal-Cost Multi-Path) uses multiple equal-cost paths.

C. Special Routing

• Broadcast Routing: Send packet to all nodes.

  • Flooding: Forward on all interfaces except incoming (needs TTL, sequence # to prevent loops).

  • Reverse Path Forwarding (RPF): Forward only if incoming interface is on shortest path back to source.

• Multicast Routing: Send packet to group of interested nodes.

  • Tree-Based: Source-specific tree or shared tree (e.g., PIM - Protocol Independent Multicast).

  • DVMRP (Distance Vector Multicast Routing Protocol): Uses reverse path, tunnels for non-multicast routers.

D. IP Addressing

1. Classful Addressing (Legacy):

• Limitations: Waste of addresses (e.g., Class B too big for many orgs), rapid routing table growth.

2. CIDR (Classless Inter-Domain Routing):

• Notation: a.b.c.d / n where n = network prefix length.

• Aggregation (Supernetting): Combine multiple prefixes into one larger prefix to reduce routing table size.

• Example: 192.168.0.0/24, 192.168.1.0/24, 192.168.2.0/24 → 192.168.0.0/22.

3. Subnetting:

• Concept: Divide a larger network (/n) into smaller subnets by borrowing bits from host ID.

• Subnet Mask: 255.255.255.0 = /24. Network ID = IP AND Mask.

• VLSM (Variable Length Subnet Mask): Allocate subnets of different sizes based on need (e.g., /26, /27).

• Calculation Example (from past papers):

  • Given: 192.168.10.0/24 → Divide into 4 subnets.

  • Borrow 2 bits (2²=4). New mask: /26 = 255.255.255.192.

  • Subnets:

    1. 192.168.10.0/26 → Range: 192.168.10.1 – 192.168.10.62, BC: 192.168.10.63

    2. 192.168.10.64/26 → Range: 192.168.10.65 – 192.168.10.126, BC: 192.168.10.127

    3. 192.168.10.128/26 → Range: 192.168.10.129 – 192.168.10.190, BC: 192.168.10.191

    4. 192.168.10.192/26 → Range: 192.168.10.193 – 192.168.10.254, BC: 192.168.10.255

4. IPv4 vs IPv6:

E. Network Layer Protocols

1. ARP (Address Resolution Protocol):

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

• Operation: Broadcast ARP request: "Who has IP X? Tell Y (MAC)." Owner replies with ARP reply (unicast).

• ARP Cache: Table of IP-MAC mappings, timeout-based.

• Proxy ARP: Router answers ARP for another host, making remote host appear local.

2. RARP (Reverse ARP):

• Function: Map MAC address → IP address (diskless workstations booting).

• Operation: RARP request broadcast with MAC; RARP server replies with IP.

• Replaced by: BOOTP and DHCP.

3. ICMP (Internet Control Message Protocol):

• Role: Error reporting & diagnostics for IP (which is connectionless).

• Common Messages:

  • Echo Request/Reply: Used by ping.

  • Destination Unreachable: (Port unreachable, host unreachable).

  • Time Exceeded: Used by traceroute (TTL=0).

  • Source Quench: Deprecated (congestion signal).

  • Redirect: Suggest better next-hop.

• Tools: ping (ICMP Echo), traceroute (ICMP Time Exceeded).

F. Congestion Control in Network Layer

Causes: Buffer overflow, slow processors, low bandwidth. QoS Parameters: Throughput, delay, jitter, loss.

Techniques:

1. Load Shedding: Drop packets when overloaded.

   • Policies: Random, priority-based, round-robin.

2. Traffic Shaping: Regulate data rate.

   • Leaky Bucket: Fixed output rate; bursty input smoothed. Packets can be lost if bucket full.

   • Token Bucket: Tokens accumulate at rate r; burst up to b tokens allowed. No loss if tokens available.

3. Congestion Avoidance:

   • RED (Random Early Detection): Probabilistically drop packets before queue full, to signal senders to slow down. Prevents global sync.

G. Network Layer Services

H. Router and Gateway Functions

Router Operations:

1. Forwarding: Move packet from input to output port (performed by forwarding plane; fast, hardware-based).

2. Routing: Compute best path (performed by control plane; runs routing protocols).

3. Queuing: Manage multiple packets for same output (FIFO, priority, WFQ).

4. Switching: Internal fabric moves packet from input to output queue.

Gateway: Device that operates at Application Layer (7) to connect dissimilar networks (e.g., email gateway converts SMTP ↔ X.400, protocol translator).

IV. Transport Layer

A. Transport Services

• Connection-Oriented (TCP): Reliable, in-order, congestion-controlled, flow-controlled.

• Connectionless (UDP): Unreliable, unordered, no congestion control.

• Multiplexing/Demultiplexing: Uses port numbers to deliver data to correct process.

• Flow Control: Receiver-driven (receiver advertises window size).

• Congestion Control: Network-aware (sender adjusts rate based on network congestion signals).

B. Transmission Control Protocol (TCP)

1. TCP Header Format (20 bytes min):

[Source Port (16)] [Dest Port (16)] [Sequence # (32)] [Ack # (32)]

[Data Offset (4)] [Reserved (6)] [Flags (6)] [Window (16)]

[Checksum (16)] [Urgent Pointer (16)] [Options (variable)]

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

• Data Offset: Header length in 32-bit words.

• Window: Receiver's available buffer size (flow control).

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

• 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, ACK=1, seq=v, ack=u+1

  4. Client → Server: ACK=1, seq=u+1, ack=v+1

• Why Graceful Termination? Ensures all data delivered, resources released properly. TIME_WAIT (2MSL) ensures last ACK received & old duplicate packets expire.

3. Reliable Data Transfer:

• Sequence Numbers: Byte-oriented, not packet-oriented.

• Acknowledgment: Cumulative ACK (acknowledges all bytes up to ack#). Selective ACK (SACK) option for out-of-order.

• Retransmission: RTO (Retransmission Timeout) estimated from RTT (Round-Trip Time) samples.

• Flow Control: Receiver advertises rwnd (receiver window) in header. Sender's effective window = min(cwnd, rwnd).

4. Congestion Control (TCP Reno):

• Variables: cwnd (congestion window), ssthresh (slow start threshold).

• Phases:

  1. Slow Start: cwnd starts at 1 MSS, doubles each RTT (exponential) until cwnd ≥ ssthresh or loss.

  2. Congestion Avoidance: cwnd increases by 1 MSS per RTT (additive increase).

  3. Fast Retransmit: 3 duplicate ACKs → retransmit missing segment without waiting for RTO.

  4. Fast Recovery: On 3 dupACKs, set ssthresh = cwnd/2, cwnd = ssthresh + 3, then additive increase.

• AIMD (Additive Increase Multiplicative Decrease): Core principle: increase slowly, cut in half on loss.

C. User Datagram Protocol (UDP)

• Header Format (8 bytes):

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

• Characteristics: Connectionless, unreliable, no congestion control, no flow control, no ordering. Low overhead, low latency.

• Applications: DNS, VoIP, streaming, DHCP, SNMP.

• Comparison with TCP:

  | Feature | TCP | UDP | | :--- | :--- | :--- | | Connection | Connection-oriented | Connectionless | | Reliability | Yes (ACK, retransmit) | No | | Ordering | Yes | No | | Congestion Control | Yes | No | | Overhead | High (20+ bytes) | Low (8 bytes) | | Use Case | File transfer, web, email | Real-time, query-response |

[!TIP] Exam Focus: TCP 3-way/4-way handshake, congestion control phases (slow start → congestion avoidance), UDP header fields.

V. Application Layer

A. Domain Name System (DNS)

• Need: Human-readable names → IP addresses.

• Hierarchy: Root servers → TLD (.com, .org, .in) → Authoritative servers → Local/recursive resolvers.

• Resolution:

  • Iterative Query: Server returns referral to next server if not authoritative.

  • Recursive Query: Server does full lookup on behalf of client (typical for resolvers).

• Caching: Each server caches results with TTL (Time to Live). Reduces latency and traffic.

• Common Records: A (IPv4), AAAA (IPv6), MX (mail exchange), CNAME (canonical name), NS (name server).

B. Electronic Mail

• Architecture:

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

  • MTA (Mail Transfer Agent): Server that transfers mail (SMTP).

  • MDA (Mail Delivery Agent): Delivers to mailbox.

• SMTP (Simple Mail Transfer Protocol):

  • Operation: TCP port 25. Push protocol (sender to receiver's MTA).

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

  • ESMTP (Extended SMTP): Supports MIME (multimedia), authentication, TLS.

• Retrieval: POP3 (download & delete), IMAP (keep on server, manipulate folders).

C. World Wide Web and HTTP

• HTTP (Hypertext Transfer Protocol):

  • Request-Response Model: Client (browser) sends request, server sends response.

  • Methods: GET (fetch), POST (submit data), HEAD, PUT, DELETE.

  • Stateless: Server doesn't retain session info between requests. Cookies used for state.

  • Versions:

    • HTTP/1.0: New TCP connection per request (non-persistent).

    • HTTP/1.1: Persistent connections (keep-alive), pipelining, chunked encoding.

    • HTTP/2: Multiplexing over single TCP, header compression, server push.

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

D. Network Management (SNMP)

• SNMP (Simple Network Management Protocol):

  • Architecture:

    • Manager: Central monitoring system.

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

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

  • Operations:

    • GET: Manager reads variable.

    • SET: Manager changes variable.

    • TRAP: Agent sends unsolicited alert (e.g., link down).

E. File Transfer (FTP)

• FTP (File Transfer Protocol):

  • Two Connections:

    1. Control Connection (TCP port 21): Commands/responses (persistent).

    2. Data Connection (TCP port 20 or dynamic): Actual file data (opened/closed per transfer).

  • Active Mode: Server opens data connection to client (client specifies port via PORT).

  • Passive Mode: Client opens data connection to server (server specifies port via PASV). Firewall-friendly.

F. Other Application Layer Protocols (Brief)

• Telnet: Remote login (unencrypted, port 23). Replaced by SSH.

• SSH (Secure Shell): Encrypted remote login (port 22).

• DHCP (Dynamic Host Configuration Protocol): Dynamically assign IP, mask, gateway, DNS to clients. DORA process: Discover, Offer, Request, Acknowledgment.

VI. Additional and Cross-Cutting Topics

A. Network Topologies & Design (Recap)

See Section I.D for detailed table. Design Considerations: Cost, scalability, fault tolerance, performance, ease of installation.

B. Network Devices Deep Dive (Recap)

See Section I.E. Key Decision Point: Layer 2 (MAC) vs Layer 3 (IP) forwarding:

• Switch/Bridge: Uses MAC address table; forwards within same network.

• Router: Uses routing table; forwards between different networks.

C. Multiplexing Techniques (Recap)

See Section II.B.3. FDM (frequency bands), TDM (time slots), WDM (wavelengths on fiber).

D. Cryptography and Security (Basics)

• Symmetric Encryption: Same key for encrypt/decrypt (AES, DES). Fast, key distribution problem.

• Asymmetric Encryption: Public/private key pair (RSA). Slower, solves key distribution.

• Digital Signature: Hash of message encrypted with sender's private key. Provides authentication, non-repudiation.

• Digital Certificate: Binds public key to entity, signed by CA (Certificate Authority).

• Firewalls:

  • Packet Filtering: Stateless; examines packet headers.

  • Stateful Inspection: Tracks connection state.

  • Application-Level Gateway (Proxy): Deep packet inspection at App layer.

E. Service Primitives (OSI Model - Recap)

See Section I.A. Example in Transport Layer (TCP):

1. Transport Layer (4) requests from Application (5): TRANSPORT_CONNECT.request(dest_port)

2. Transport Layer (4) indicates to Application (5): TRANSPORT_DATA.indication(data)

3. Application (5) responds: TRANSPORT_DATA.response(ack)

4. Transport Layer (4) confirms: TRANSPORT_CONNECT.confirm()

F. Quality of Service (QoS) in Depth

• Parameters:

  • Throughput: Data delivered per unit time.

  • Delay: Time from send to receive.

  • Jitter: Variation in delay (critical for real-time).

  • Loss: Packet loss rate.

• Mechanisms:

  • Priority Queuing: Classes with priorities.

  • Weighted Fair Queuing (WFQ): Each flow gets weighted share.

  • IntServ (Integrated Services): Per-flow resource reservation (RSVP). Not scalable.

  • DiffServ (Differentiated Services): Per-hop behavior (PHB) based on DSCP bits in IP header. Scalable.

G. Performance Evaluation

• Throughput: Actual achieved data rate.

• Efficiency: Throughput / Channel capacity.

• Delay Components:

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

*   **Transmission Delay:** $$\displaystyle \frac{\text{Frame Size (bits)}}{\text{Bandwidth (bps)}} $$

*   **Propagation Delay:** $$\displaystyle \frac{\text{Distance (m)}}{\text{Propagation Speed (m/s)}} $$

• Example (Stop-and-Wait Efficiency): As in Section II.D.1: $$\displaystyle U = \frac{1}{1 + 2a} $$ where $$\displaystyle a = \frac{T_{prop}}{T_{trans}} $$.

END OF UNIT 3 NOTES