Computer Communication Networks

Chapter 4: The Medium Access Control Sublayer

Comprehensive slide-by-slide explanation with original diagrams and full conceptual coverage.
Slide 1 The Medium Access Control (MAC) Sublayer — Chapter 4

Explanation:
This chapter focuses on the MAC (Medium Access Control) sublayer, which is part of the Data Link
Layer in the OSI model. The MAC sublayer is responsible for controlling how devices on a shared
network medium (like a cable or wireless channel) gain access to transmit data. Without proper MAC
protocols, multiple devices trying to send at the same time would cause chaos — their signals would
collide and corrupt each other. The chapter covers static and dynamic allocation, ALOHA, CSMA/CD,
collision-free protocols, Ethernet, 802.11 Wi-Fi, and VLANs.
Slide 2 Static Channel Allocation

Explanation:
Static channel allocation divides a shared medium into fixed portions (using FDM or TDM) and assigns
each user a dedicated slice. The service delay formula T = 1/(µC - λ) applies to a single channel with
capacity C bps, arrival rate λ frames/sec, and mean frame length 1/µ bits. When the channel is split into
N sub-channels, the delay becomes T_N = N/(µC - λ), which is N times worse. This shows that static
allocation is highly inefficient: if a station has nothing to send, its slot goes to waste. Dynamic allocation
schemes were developed to avoid this problem by letting any station use the full channel whenever it
needs it.
Slide 3 Assumptions for Dynamic Channel Allocation

Explanation:
Before studying dynamic MAC protocols, five key assumptions must be understood: (1) Independent
Traffic — each of N stations generates frames independently according to a Poisson process. (2) Single
Channel — all communication uses one shared channel. (3) Observable Collisions — every station can
detect when two frames overlap and collide. (4) Continuous or Slotted Time — time can flow
continuously, or be divided into discrete slots; slotted systems perform better because all transmissions
are aligned. (5) Carrier Sense or No Carrier Sense — some protocols let stations listen before
transmitting (carrier sense), greatly reducing collisions.
Slide 4 Multiple Access Protocols — Overview

Explanation:
This slide introduces the five major categories of multiple access protocols covered in the chapter: (1)
ALOHA — simplest random-access scheme where stations transmit whenever they want. (2) CSMA
(Carrier Sense Multiple Access) — stations listen before transmitting to avoid unnecessary collisions.
(3) Collision-Free Protocols — ensure no collisions ever occur, using techniques like token passing or
bit-map reservation. (4) Limited-Contention Protocols — combine low-delay contention at light load with
efficient collision-free access at heavy load. (5) Wireless LAN Protocols — special techniques needed
because wireless nodes can't always hear each other (hidden/exposed terminal problems).
Slide 5 ALOHA (1) — Pure ALOHA

Explanation:
Pure ALOHA, developed at the University of Hawaii in the 1970s, is the simplest random access
protocol. Any station that has a frame to send transmits it immediately at any arbitrary time, without
listening to the channel first. If two frames overlap in time at all, a collision occurs and both are
destroyed. The diagram shows five users (A–E) transmitting at different times. User A's first frame
collides with User B; User A's second frame collides with User C. After a collision, each station waits a
random amount of time before retransmitting to avoid repeatedly colliding with each other. The
maximum efficiency of pure ALOHA is only 18.4% (1/2e), meaning much of the channel bandwidth is
wasted on collisions.
Slide 6 ALOHA (2) — Throughput Analysis

Explanation:
For throughput analysis, if G is the mean number of transmission attempts per frame time, the
vulnerable period for any frame is 2t (where t is the frame transmission time) because a collision can be
caused by a frame starting up to one full frame time before or after. During a 2-frame interval, the
average transmissions attempted = 2G. Using the Poisson distribution, the probability of zero other
frames during the vulnerable period is e^(-2G). Therefore, throughput S = G * e^(-2G). This is
maximized at G=0.5 giving S_max = 1/(2e) ≈ 0.184, meaning pure ALOHA can use at most 18.4% of
channel capacity. Slotted ALOHA doubles this to 36.8% by restricting transmissions to slot boundaries.
Slide 7 ALOHA (3) — Retransmission Probability

Explanation:
When a frame collides, it must be retransmitted. The probability that a specific transmission attempt
succeeds is P0 = e^(-2G) (probability no other station transmits during the vulnerable period). The
probability that a frame requires exactly k transmission attempts is P(k) = (1 - P0)^(k-1) * P0, which is a
geometric distribution. The expected number of transmissions needed E[k] = 1/P0 = e^(2G). At high
load (large G), this grows exponentially — frames need many retransmissions, causing congestion
collapse. This instability is a major weakness of pure ALOHA.
Slide 8 Persistent and Non-persistent CSMA

Explanation:
CSMA (Carrier Sense Multiple Access) protocols listen to the channel before transmitting, significantly
reducing collisions. The graph compares channel utilization vs. load for different variants. 1-persistent
CSMA: if the channel is busy, wait and transmit immediately when it goes idle — this can cause
collisions when multiple stations all waited. Non-persistent CSMA: if busy, wait a random time then
sense again — less greedy, less collision probability, but higher delay when channel is lightly loaded.
p-persistent CSMA (slotted): when idle, transmit with probability p — balances delay and collision. The
graph shows that CSMA dramatically outperforms ALOHA at higher loads.
Slide 9 CSMA with Collision Detection (CSMA/CD)

Explanation:
CSMA/CD improves efficiency by having stations not only listen before sending, but also monitor the
channel while transmitting. If a collision is detected (the received signal differs from what was sent), the
station immediately aborts transmission and sends a brief jam signal to alert all others. This saves the
rest of the frame transmission time. A station must wait 2T (the round-trip propagation time) to
determine if it successfully seized the channel — for a 1 km cable, T ≈ 5 µsec so 2T ≈ 10 µsec.
CSMA/CD operates in three states: Contention (competing for the channel), Transmission (successfully
sending), Idle (channel is free). CSMA/CD is the basis of classic Ethernet (IEEE 802.3).
Slide 10 Collision-Free Protocols — Bitmap Protocol

Explanation:
The bit-map (reservation) protocol guarantees zero collisions. In each contention period of N slots,
station i announces its intention to send by setting bit i in slot i. After all N stations have had a chance to
reserve, data frames are sent in order of station number. Low-numbered stations must wait an average
1.5N slots (they must wait for the full N-bit reservation period plus all higher-priority stations ahead of
them), while high-numbered stations wait only 0.5N slots on average. The mean wait is N slots. Channel
efficiency = d/(d + N), where d is the frame length in bits. As d grows large, efficiency approaches 100%.
The overhead is the N-bit reservation period.
Slide 11 Collision-Free Protocols — Token Ring (IEEE 802.5/802.17)

Explanation:
In token ring, stations are connected in a logical ring. A special 3-byte token frame circulates
continuously around the ring. Only the station holding the token is permitted to transmit. After sending a
frame (or after a timer expires), the station releases the token to the next station downstream. The
diagram shows the ring topology with arrows indicating the direction of token circulation. IEEE 802.5
was the original token ring standard; IEEE 802.17 is the Resilient Packet Ring standard. The efficiency
equals that of the bit-map protocol. Token ring is deterministic — every station gets equal, fair access —
unlike CSMA/CD where stations compete randomly.
Slide 12 Collision-Free Protocols — Binary Countdown

Explanation:
Binary countdown is a collision-free protocol where stations use their binary addresses as priorities. All
stations wanting to transmit simultaneously broadcast their address bits from the most significant bit
downward. Bit-by-bit, stations that have a 0 where any station has a 1 drop out of contention (because
OR logic on the shared medium means the higher bit wins). After log2(N) bit slots, exactly the
highest-addressed station wins and transmits its data. A dash in the table means silence (station
dropped out). Channel efficiency = d/(d + log2 N). If the first bits of a data frame are the station's
address, overhead disappears and efficiency reaches 100%, since the arbitration happens during
address transmission.
Slide 13 Limited-Contention Protocols

Explanation:
Limited-contention protocols adaptively combine the best of both worlds: at low load, they use
contention (like CSMA) to achieve low delay; at high load, they switch to collision-free operation for high
efficiency. The probability q that exactly one station transmits in a slot (success) is maximized when p =
1/k (where k is the number of ready stations). As k→∞, q→1/e ≈ 0.37. The key idea: stations are divided
into groups, and slots are assigned to groups. At low load (few stations), let many stations share each
slot (contention). At high load (many stations), assign fewer stations per slot to avoid collisions.
Slide 14 Limited-Contention Protocols — Group Assignment

Explanation:
This slide details how stations are grouped in limited-contention protocols. Stations are divided into (not
necessarily disjoint) groups, and only group i members may contend in slot i. Two extreme cases: (1)
Each group has exactly one member — this becomes collision-free like binary countdown. (2) One
group contains all stations — this becomes pure contention like CSMA. The optimal strategy is to adjust
group sizes dynamically: large groups (many stations per slot) when load is low, giving fast response;
small groups (one station per slot) when load is high, avoiding collisions. This adaptability is the
hallmark of limited-contention protocols.
Slide 15 The Adaptive Tree Walk Protocol

Explanation:
The adaptive tree walk protocol organizes stations as leaves of a binary tree. Transmission slots are
assigned by walking the tree. If q is the average number of ready stations per slot, the search should
begin at level i = log2(q) in the tree, where the expected number of ready stations below that level is
approximately 1 (2^(-i) * q = 1 when i = log2 q). Level 0 is the root (all N stations), level 1 has two
subtrees of N/2 each, level i has 2^i subtrees each covering 2^(-i) of all stations. If a collision occurs at a
node, the algorithm descends into the left subtree first, then the right. If a slot is idle, it backtracks up.
This balances efficiency at all load levels.
Slide 16 Wireless LAN Protocols — Hidden and Exposed Terminals

Explanation:
Wireless networks have unique problems that wired networks don't. The hidden terminal problem:
Station A transmits to B. Station C (out of A's range) wants to transmit to B too. C cannot hear A, so C
thinks the channel is free and transmits, causing a collision at B. C is 'hidden' from A. The exposed
terminal problem: B transmits to A. Station C wants to transmit to D (outside B's range). C senses B's
signal and waits, thinking the channel is busy — but C transmitting to D would NOT cause a collision at
A. C is 'exposed' to B unnecessarily. Both problems show that simple carrier sensing is insufficient in
wireless networks — a station's ability to hear does not tell it the full picture of who is interfering at the
receiver.
Slide 17 Wireless LAN Protocols — MACA

Explanation:
MACA (Multiple Access with Collision Avoidance) addresses hidden/exposed terminals using short
control frames before data transmission. Step 1: Sender A broadcasts an RTS (Request To Send, 30
bytes) to B, containing the intended data length. Step 2: Receiver B responds with a CTS (Clear To
Send) if it is free, echoing the data length. All stations overhearing the CTS know to stay silent for the
duration of the upcoming data transmission (they know how long from the length field). Stations hearing
only the RTS but not the CTS know they are far from B and won't interfere, so they may transmit. This
mechanism solves the hidden terminal problem. MACA is the conceptual basis for 802.11's RTS/CTS
optional mechanism.
Slide 18 Ethernet (IEEE 802.3) — Overview

Explanation:
Ethernet, standardized as IEEE 802.3, is the dominant wired LAN technology. It has evolved through
several generations: Classic Ethernet (10 Mbps): shared medium using thick or thin coaxial cable. Thick
Ethernet (10BASE5) supports up to 500 m segments with 100 users; Thin Ethernet (10BASE2) supports
185 m with 30 users. Segments can be extended up to 2.5 km total using up to 4 repeaters. Switched
Ethernet: replaced hubs with switches for full-duplex, collision-free operation. Fast Ethernet (IEEE
802.3u): 100 Mbps. Gigabit Ethernet (IEEE 802.3z/ab): 1 Gbps. 10 Gigabit Ethernet (IEEE 802.3ae): 10
Gbps and beyond.
Slide 19 Classic Ethernet Physical Layer

Explanation:
The classic Ethernet physical layer uses Manchester encoding to represent bits on the coaxial cable.
Manchester encoding encodes a 1 as a high-to-low transition and a 0 as a low-to-high transition in the
middle of each bit period. This ensures continuous transitions on the wire, allowing clock recovery at the
receiver. The transceiver (connected to the thick coax via a vampire tap) handles signal detection and
collision detection. The 10 Mbps rate is achieved with Manchester encoding at 20 Mbaud. The diagram
shows the multi-layer structure from the thick coaxial cable up through the transceiver cable to the NIC
in the computer.
Slide 20 Classic Ethernet MAC Sublayer — Frame Format

Explanation:
The Ethernet MAC frame structure consists of: Preamble (7 bytes): 10101010... — a 10 MHz square
wave for 6.4 µsec that allows receivers to synchronize their clocks. Start of Frame (1 byte): 10101011
— marks the start of the actual frame. Destination Address (6 bytes): if the first bit is 0, unicast (single
destination); if 1, multicast (group); if all 1s, broadcast (everyone). Source Address (6 bytes): globally
unique 48-bit MAC address. First 3 bytes are the OUI (Organizationally Unique Identifier) assigned by
IEEE to the manufacturer; last 3 bytes are assigned by the manufacturer to each NIC. Length/Type (2
bytes), Data, and Checksum (CRC-32) follow.
Slide 21 MAC Sublayer Protocol — Minimum Frame Size

Explanation:
CSMA/CD requires that a frame's transmission time be at least 2 × propagation delay (round-trip time,
RTT), so that the sender is still transmitting when a collision signal returns. If transmission ends before
the collision signal arrives, the sender can't detect the collision. For a 2500 m maximum cable: RTT = 50
µsec. At 10 Mbps, 50 µsec × 10 Mbps = 500 bits. Rounded up to a byte boundary gives the minimum
frame size of 512 bits = 64 bytes. Frames shorter than 64 bytes must be padded with dummy data. This
is why the Ethernet minimum frame size is 64 bytes — it's not arbitrary, it's derived from physics.
Slide 22 CSMA/CD with Binary Exponential Backoff

Explanation:
After each collision, Ethernet uses binary exponential backoff to determine how long to wait before
retransmitting. After the i-th consecutive collision, each colliding station randomly chooses a wait time
between 0 and 2^i − 1 time slots (each slot = 51.2 µsec = 512 bit times at 10 Mbps). 1st collision: wait 0
or 1 slots. 2nd collision: wait 0–3 slots. 10th collision (and beyond): wait 0–1023 slots (frozen at
maximum). After 16 consecutive collisions, the controller gives up and reports failure to the host. The
exponential growth of the wait window ensures that as more stations compete, they spread out in time,
reducing collision probability. The random choice prevents synchronized retransmissions.
Slide 23 Ethernet Performance Analysis

Explanation:
After a frame transmission, a contention period follows where stations compete for the channel. If k
stations are ready, the probability A that some station acquires the channel in a slot is maximized when
p = 1/k, giving A → 1/e ≈ 0.37 as k → ∞. The probability the contention interval has exactly j slots is (1 -
A)^(j-1) × A. Mean contention slots = 1/A. Mean contention interval w = 2τ/A (since each slot = 2τ, the
propagation delay). Since w ≤ 2τe ≈ 5.4τ, longer cables increase contention overhead. Mean frame
transmission P = frame_size / bandwidth. Efficiency = P/(P + w). This shows that short frames or long
cables reduce efficiency.
Slide 24 Efficiency of Ethernet at 10 Mbps with 512-bit Slots

Explanation:
This slide shows a graph of Ethernet channel efficiency as a function of the number of competing
stations and frame length. Key observations: Longer frames achieve much higher efficiency — a
1024-byte frame at few stations achieves near-100% efficiency, while a 64-byte minimum frame drops
to around 50% with many stations. More competing stations (higher k) always reduce efficiency
because contention periods become more frequent. The trade-off: short frames reduce latency for
interactive traffic but waste bandwidth on contention overhead. This analysis justifies the Ethernet
design choice of a 64-byte minimum and 1500-byte maximum frame size.
Slide 25 Switched Ethernet (1) — Hub vs. Switch

Explanation:
Classic Ethernet uses a hub, which is simply a multi-port repeater: any signal entering one port is
broadcast to all other ports, and all stations share one collision domain. A switch, by contrast, is an
intelligent device that reads MAC addresses and forwards frames only to the specific port connected to
the destination. Hub: all ports in the same collision domain; bandwidth shared; half-duplex; collisions
occur. Switch: each port is its own collision domain; full-duplex between switch and station possible; no
collisions; dedicated bandwidth per port. Switches dramatically improved Ethernet performance and
enabled modern gigabit networking.
Slide 26 Switched Ethernet (2) — Switch Architecture

Explanation:
The diagram shows a typical switched Ethernet setup with multiple computers connected individually to
switch ports via twisted-pair cables, and a hub connected to one switch port (creating a legacy shared
segment). The switch maintains a MAC address table mapping addresses to ports. When a frame
arrives, the switch looks up the destination MAC; if known, it forwards to that port only. If unknown, it
floods to all ports except the source. Each switch-to-station link operates independently and
simultaneously — N full- duplex links provide N times the bandwidth of a shared hub. This is why
modern networks use switches everywhere.
Slide 27 Fast Ethernet (100 Mbps) — Physical Variants

Explanation:
Fast Ethernet (IEEE 802.3u) increased speed from 10 to 100 Mbps with three physical layer options:
100BASE-T4: uses 4 pairs of UTP wiring at 25 MHz bandwidth. One pair transmits to the switch, one
from the switch; the other two are bidirectional. Three voltage levels for symbols (8B6T encoding).
100BASE-TX: uses 2 pairs of Cat 5 UTP or STP. 4B/5B encoding (every 4 data bits encoded as 5 bits
for synchronization) runs at 125 MHz clock, delivering 100 Mbps data rate. 100BASE-FX: uses fiber
optic cable, but maximum cable length is limited to under 250 m when using hubs (for collision detection
to work). With switches, this limit doesn't apply.
Slide 28 Gigabit Ethernet (1 Gbps)

Explanation:
Gigabit Ethernet (IEEE 802.3z for fiber, 802.3ab for copper) scales Ethernet to 1 Gbps. Two techniques
extend the minimum cable length to 200 m when using hubs: (1) Carrier Extension: the hardware pads
short frames out to 512 bytes (4096 bits at 1 Gbps = 4 µsec), long enough for collision detection at 200
m. (2) Frame Bursting: allows a station to send multiple frames back-to-back after seizing the channel,
improving efficiency. 8B/10B encoding (8 data bits → 10 line bits) is used for clock synchronization. For
UTP copper (1000BASE-T), all 4 wire pairs carry data simultaneously in full-duplex mode using 5
voltage levels (PAM-5). Pause frames allow flow control between switch and station.
Slide 29 10 Gigabit Ethernet

Explanation:
10 Gigabit Ethernet (IEEE 802.3ae) extends Ethernet to 10 Gbps, primarily for data centers and
backbone links. Three physical options: Fiber (10GBASE-SR/LR/ER): uses 64B/66B coding (64 data
bits encoded as 66 bits — very low overhead of 3%). CX (twinaxial copper, 10GBASE-CX4): uses
8B/10B coding at 3.125 Gsymbol/sec on each of 4 pairs. T (Cat 6a/7 copper, 10GBASE-T): requires 16
voltage levels (PAM-16) and uses LDPC (Low Density Parity Check) error correction codes to maintain
signal integrity. 40 Gbps and 100 Gbps standards are also defined, used in modern cloud data centers.
Slide 30 Ethernet Benefits

Explanation:
Ethernet has become the universal LAN standard due to five key advantages: Reliable: switching
eliminated shared collision domains; full-duplex links with flow control give near-zero packet loss under
normal conditions. Cheap: twisted pair copper cables cost pennies per meter; commodity NICs cost a
few dollars; switches are mass-produced. Easy to maintain: plug-and-play with no software
configuration required; auto-negotiation handles speed and duplex automatically. Well integrated with
IP: both Ethernet and IP are connectionless, stateless protocols — no circuits to set up, no connection
state to manage. Flexible: evolved from 10 Mbps to 100 Gbps+ with minimal changes to the frame
format or network architecture.
Slide 31 Wireless LANs — 802.11 Topics Overview

Explanation:
IEEE 802.11 (Wi-Fi) is the dominant wireless LAN standard. This section covers five areas: (1)
Architecture and protocol stack — how 802.11 fits into the OSI model and its two operating modes
(infrastructure and ad-hoc). (2) Physical layer — modulation schemes, frequency bands, and data rates
for 802.11b/a/g/n. (3) MAC sublayer protocol — CSMA/CA (Collision Avoidance) instead of CD,
RTS/CTS, NAV. (4) Frame structure — the 802.11 data frame format with four address fields. (5)
Services — the services 802.11 provides including distribution, integration, and security (WPA/WPA2).
Slide 32 802.11 Architecture — Infrastructure Mode

Explanation:
In infrastructure mode (the most common Wi-Fi setup), wireless clients communicate through an
Access Point (AP). The AP is connected to the wired network (shown as 'To Network' in the diagram).
All communication between clients goes through the AP — even if two clients are within range of each
other, they do not communicate directly. The set of all clients associated with one AP forms a BSS
(Basic Service Set). Multiple APs connected by a wired backbone (the Distribution System) form an
ESS (Extended Service Set), allowing seamless roaming. This is equivalent to how cellular networks
work with base stations.
Slide 33 802.11 Architecture — Ad-hoc Mode

Explanation:
In ad-hoc mode (also called IBSS — Independent Basic Service Set), wireless stations communicate
directly with each other without any access point or wired infrastructure. Stations within range of each
other form a peer-to-peer network spontaneously. This mode is useful for temporary networks (e.g.,
sharing files between two laptops when no AP is present, emergency first-responder networks). Ad-hoc
mode has limitations: stations far apart may not be directly reachable (no multi-hop routing at Layer 2),
and there is no centralized management of channel access. Ad-hoc mode is less commonly used than
infrastructure mode.
Slide 34 802.11 Protocol Stack

Explanation:
The 802.11 standard has evolved through multiple generations at the physical layer while keeping the
same MAC sublayer: 802.11 (legacy, 1997): 1–2 Mbps in 2.4 GHz. 802.11b (1999): up to 11 Mbps in
2.4 GHz — the first widely adopted version. 802.11a (1999): up to 54 Mbps in 5 GHz band. 802.11g
(2003): up to 54 Mbps in 2.4 GHz — backward compatible with 802.11b. 802.11n (2009): up to 600
Mbps using MIMO and wider 40 MHz channels. All versions share the same LLC (Logical Link Control)
sublayer above and present a uniform interface to the network layer. The physical layer differences are
hidden from higher layers.
Slide 35 802.11b Physical Layer

Explanation:
802.11b operates in the 2.4 GHz ISM band using DSSS (Direct Sequence Spread Spectrum) with four
data rates: 1 Mbps: 11-chip Barker code spreads each bit across 11 chips at 11 Mchips/sec, using
BPSK (Binary Phase Shift Keying) — 1 bit per 11 chips. 2 Mbps: QPSK (Quadrature PSK) — 2 bits per
11 chips, same chip rate. 5.5 Mbps: CCK (Complementary Code Keying) — 4 bits per 8 chips. 11 Mbps:
CCK — 8 bits per 8 chips. Spread spectrum resists interference and multipath fading. The chip rate of
11 Mchips/sec uses 22 MHz of bandwidth per channel. Three non-overlapping channels exist in 2.4
GHz (channels 1, 6, 11).
Slide 36 802.11a, g, n Physical Layer

Explanation:
802.11a uses the 5 GHz band with OFDM (Orthogonal Frequency Division Multiplexing). 52 subcarriers:
48 for data, 4 for synchronization pilot tones. Each symbol lasts 4 µsec and carries 1, 2, 4, or 6 bits
depending on modulation (BPSK/QPSK/16-QAM/64-QAM). Supports 8 data rates from 6 to 54 Mbps.
The 5 GHz band has 1/7 the range of 2.4 GHz (higher frequency = more attenuation). 802.11g brings
OFDM to 2.4 GHz — same rates as 802.11a but same range as 802.11b. 802.11n doubles channel
width to 40 MHz and uses MIMO (Multiple Input Multiple Output) with multiple antennas to multiply
throughput (up to 4×4 MIMO). Maximum data rate: 600 Mbps with 4 spatial streams in 40 MHz.
Slide 37 802.11 MAC — CSMA/CA Protocol

Explanation:
Because 802.11 stations cannot detect collisions while transmitting (full-duplex is not possible in
wireless — the transmitted signal swamps the receiver), 802.11 uses CSMA/CA (Collision Avoidance)
instead of CSMA/CD. Before transmitting, a station waits for the channel to be idle for DIFS duration.
Then it waits an additional random backoff time (in slots). If the channel becomes busy during backoff,
the counter freezes until the channel is idle again. After sending, the sender waits for an ACK. If no ACK
arrives within a timeout, it assumes a collision and doubles the backoff window (exponential backoff).
The diagram shows the timing sequence. ACKs are essential because the sender cannot know if its
frame was received correctly.
Slide 38 802.11 Collision Avoidance — Virtual Channel Sensing (NAV)

Explanation:
802.11 uses virtual carrier sensing via the NAV (Network Allocation Vector) to reduce collisions. Every
frame header contains a Duration field indicating how long the channel will be occupied (including the
upcoming ACK). All stations overhearing this field set their NAV timer to that value. While NAV > 0, a
station considers the channel busy and does not transmit, even if it cannot physically hear the ongoing
transmission. RTS/CTS (Request To Send / Clear To Send) is an optional mechanism: the sender first
sends a short RTS; the receiver responds with CTS. Both carry the Duration field. Stations hearing
either RTS or CTS set their NAV, preventing hidden terminal collisions. NAV is updated by overhearing
any frame.
Slide 39 CSMA/CA — Practical Details

Explanation:
In practice, the optional RTS/CTS mechanism is rarely used in 802.11 because the overhead of the two
extra frames slows down communication (especially for short data frames). It is only beneficial for large
frames where collision cost outweighs RTS/CTS overhead. The main collision avoidance mechanisms
actually used are: random backoff (exponential growth of the contention window after each failure) and
virtual carrier sensing by overhearing the NAV field in all frames. Fragmentation: large frames can be
split into smaller fragments. A burst of fragments is sent after channel acquisition; each fragment must
be acknowledged before the next is sent. This limits the damage from a frame error — only one small
fragment must be retransmitted, not the entire large frame.
Slide 40 802.11 Power Management

Explanation:
Battery life is critical for wireless devices. 802.11 defines a power save mode: A station informs its AP
that it is entering sleep mode. The AP buffers any incoming frames destined for the sleeping station.
The AP periodically sends beacon frames (typically every 100 ms) that include a TIM (Traffic Indication
Map) listing which stations have buffered frames waiting. Sleeping stations wake up at each beacon
interval, check the TIM, and request their buffered frames if indicated. If not indicated, they go back to
sleep immediately. This reduces active radio time by up to 90%, dramatically extending battery life,
while ensuring no frames are lost while the station is asleep.
Slide 41 Interframe Timing for QoS — IFS Types

Explanation:
802.11 defines multiple interframe spacing (IFS) values to implement QoS prioritization. Frames with
shorter IFS values get channel access sooner. SIFS (Short IFS): the shortest gap, used for ACKs, CTS
responses, and the second/later fragments in a burst — ensures high-priority responses happen
immediately. DIFS (Distributed Coordination Function IFS): used by normal data frames; longer than
SIFS, so ACKs always get priority. AIFS (Arbitration IFS): defined in 802.11e for QoS; different traffic
categories (voice, video, best-effort, background) use different AIFS values. EIFS (Extended IFS): used
after a frame error is detected; very long to prevent interfering with recovery.
Slide 42 802.11 Frame Structure

Explanation:
The 802.11 data frame has a more complex structure than Ethernet because of wireless routing through
APs. Key fields: Frame Control (2 bytes): type (management/control/data), subtype, To DS / From DS
bits, WEP/privacy, and more. Duration (2 bytes): NAV value in microseconds. Four address fields (6
bytes each): Addr1 = destination; Addr2 = source; Addr3 = BSS AP address or final destination on wired
network; Addr4 = used only in mesh/WDS. This is needed because data must travel from the original
source, through the AP, to the wired network destination. Sequence control (2 bytes): for fragmentation
and duplicate detection. Body: up to 2312 bytes (larger than Ethernet's 1500). CRC (4 bytes): FCS.
Slide 43 Data Link Layer Switching — Topics

Explanation:
This section covers how bridges and switches connect multiple LAN segments at Layer 2. Topics: Uses
of bridges: connecting LANs of different speeds or technologies; filtering traffic to reduce load on each
segment. Learning bridges: how bridges dynamically build their forwarding tables without configuration.
Spanning tree bridges: how bridges eliminate loops that would cause broadcast storms. Device
comparison: repeaters, hubs, bridges, switches, routers, and gateways — which layer each operates at.
Virtual LANs (VLANs): how to logically segment a physical LAN into separate broadcast domains.
Slide 44 Using Bridges to Connect LANs

Explanation:
A bridge connects two or more LAN segments at the data link layer (Layer 2). Unlike a repeater or hub
(Layer 1), a bridge examines the MAC address in each frame and makes a forwarding decision. If the
destination is on the same segment as the source, the bridge discards the frame (filtering). If the
destination is on a different segment, the bridge forwards the frame to the appropriate port. This
reduces collision domains: each segment is its own collision domain. Bridges can also connect LANs
using different speeds or technologies (e.g., connecting a 100 Mbps and 1000 Mbps segment), acting
as a buffer between them.
Slide 45 Backward Learning — Bridge Forwarding Algorithm

Explanation:
Bridges use backward (reverse-path) learning to build forwarding tables dynamically, with no manual
configuration. Initial state: hash table is empty. When a frame arrives on port X from source address S,
the bridge records: 'station S is reachable via port X' (with the arrival timestamp). The three forwarding
rules: (1) If the destination port equals the source port — discard (frame already on correct segment).
(2) If the destination port is known and different from source — forward to that port only. (3) If the
destination is unknown — flood: forward on all ports except the source port. Old entries expire after a
few minutes to handle topology changes. Flooding ensures frames are never dropped even before the
table is populated.
Slide 46 Protocol Processing at a Bridge

Explanation:
This diagram illustrates the internal processing at a bridge when a frame arrives. The frame arrives on
an input port and passes through the physical layer. The bridge's MAC layer checks the destination
address. The bridge software consults its forwarding table (hash table). Based on the table lookup
result, the frame is either: Discarded (destination same port as source), Forwarded out the specific port
(destination known), or Flooded out all other ports (destination unknown). Notice that the bridge has
separate MAC layers for each LAN segment it connects — it is a Layer 2 device that terminates the
MAC protocol on one side and re-initiates it on the other. It does NOT look at IP addresses.
Slide 47 Loop Creation with Redundant Links

Explanation:
When multiple bridges or switches connect the same set of LANs for redundancy, loops form in the
topology. Loops cause catastrophic broadcast storms: a broadcast frame is forwarded by each bridge to
all ports, creating copies that circulate the loop forever, multiplying exponentially until the network is
paralyzed. The diagram shows two bridges connecting LAN1 and LAN2 with two paths — a loop. Any
unknown destination frame would be flooded on both paths, and both copies would be re-flooded at the
other end, creating an infinite storm. This is why the spanning tree protocol was invented — to logically
break loops while preserving physical redundancy for fault tolerance.
Slide 48 Spanning Tree Bridging

Explanation:
The IEEE 802.1D Spanning Tree Protocol (STP) solves loops by creating a logical tree topology over
the physical mesh. Steps: (1) Elect a root bridge: the bridge with the lowest ID (priority + MAC address)
becomes root. In the diagram, B1 has the lowest ID and is elected root. (2) Each bridge finds its shortest
path to the root. (3) On each LAN segment, one bridge port is designated as the forwarding port; all
others are blocked. Blocked ports don't forward data frames but still receive BPDUs (Bridge Protocol
Data Units) to detect topology changes. If a link fails, blocked ports can be unblocked to restore
connectivity. STP guarantees loop-free forwarding while maintaining redundancy.
Slide 49 Spanning Tree — Radia Perlman's Poem (1985)

Explanation:
This slide contains the famous poem written by Radia Perlman, the inventor of the Spanning Tree
Protocol, in 1985. Perlman wrote it to explain the algorithm in a memorable way while submitting her
proposal to the IEEE committee. The poem summarizes: a spanning tree provides loop-free
connectivity (line 3-4); it spans all LANs so every station is reachable (line 5-6); the root is elected by
lowest ID (line 7-8); least-cost paths from the root are found (line 9-10); bridges compute the spanning
tree over the physical mesh (line 11-12). Perlman's invention is considered one of the most elegant
contributions to networking, and the poem is famous in computer science history.
Slide 50 Repeaters, Hubs, Bridges, Switches, Routers, and Gateways

Explanation:
Networking devices operate at different OSI layers, providing different levels of intelligence and
isolation: Repeater (Layer 1 — Physical): amplifies and regenerates electrical signals. No frame
awareness. No collision domain separation. Hub (Layer 1): multi-port repeater. Broadcasts all signals to
all ports. One collision domain. Bridge (Layer 2 — Data Link): forwards based on MAC addresses.
Separates collision domains. Switch (Layer 2): multi-port bridge. Each port is a separate collision
domain with dedicated bandwidth. Router (Layer 3 — Network): forwards based on IP addresses.
Separates broadcast domains. Connects different networks. Gateway (Layer 4–7 —
Transport/Application): protocol translator between incompatible networks (e.g., converting TCP/IP to
something else). The diagram shows which headers each device examines.
Slide 51 Virtual LANs (1) — Centralized Wiring with Hubs and Switch

Explanation:
In a traditional building with centralized wiring, all cables run to a wiring closet where hubs and a switch
are located. The problem: all stations on the same floor may be in the same broadcast domain, even if
logically they belong to different departments or security zones. A broadcast frame from any station
reaches every other station on all hubs connected to the same switch. This wastes bandwidth and is a
security risk. VLANs solve this by logically grouping ports regardless of physical location — e.g., all the
Finance department PCs on different floors can be in one VLAN, isolated from Engineering PCs, even
though they share the same physical switch.
Slide 52 Virtual LANs (2) — VLAN Operation

Explanation:
A VLAN (Virtual LAN) partitions a physical network into multiple logical broadcast domains. The diagram
shows two VLANs (gray and white) on a bridged LAN. Frames from gray VLAN nodes are forwarded
only to ports assigned to the gray VLAN, never to white VLAN ports. This provides: Isolation:
departments are logically separated; broadcast storms are contained. Security: hosts in different VLANs
cannot communicate at Layer 2; a router (with access control) is needed for inter-VLAN routing.
Flexibility: a user can move to a different floor but remain in the same VLAN by reconfiguring the switch
port, not the physical cabling. Management: easier to manage groups of users logically.
Slide 53 IEEE 802.1Q Standard (1) — VLAN-Aware Bridging

Explanation:
The IEEE 802.1Q standard defines how VLANs are implemented across multiple switches. A VLAN tag
(4 bytes) is added to the Ethernet frame header, containing a 12-bit VLAN ID (allowing up to 4094
VLANs) and priority bits. The diagram shows a bridged LAN that is partially VLAN-aware: shaded
frames carry a VLAN tag, while empty (untagged) frames are legacy frames from VLAN-unaware
stations. Trunk links between switches carry tagged frames with the VLAN ID so the receiving switch
knows which VLAN the frame belongs to. Access ports (connecting to end stations) strip the tag before
delivering the frame to the station. This allows mixed VLAN-aware and legacy equipment to coexist.
Slide 54 IEEE 802.1Q Standard (2) — Frame Format

Explanation:
The 802.1Q frame format adds a 4-byte VLAN tag between the Source Address and Length/Type fields
of the standard 802.3 Ethernet frame. The tag consists of: TPID (Tag Protocol ID, 2 bytes): always
0x8100, indicating this is an 802.1Q frame. TCI (Tag Control Information, 2 bytes): - PCP (3 bits):
Priority Code Point — 802.1p priority levels 0–7 for QoS. - DEI (1 bit): Drop Eligible Indicator — marks
frames that may be dropped under congestion. - VID (12 bits): VLAN ID — identifies which VLAN this
frame belongs to. The frame is slightly longer than legacy Ethernet (up to 1522 bytes vs 1518). Legacy
devices that don't understand the tag see a slightly larger frame with an unknown EtherType field.
Slide 55 Homework Assignment

Explanation:
This slide contains the homework assignment table listing problem numbers assigned to each student.
The problems are from the textbook chapter on the MAC sublayer (Chapter 4). Note: Problems between
26–35 are to be omitted as stated on the slide. Each student is assigned three specific problem
numbers to solve. This is administrative/assessment content and does not contain technical networking
concepts.