EECS 3213 (Priority Deck)

What is circuit switching?

Establishing a dedicated end-to-end circuit before transmitting.

What is packet switching?

Splitting data into packets and transmitting them over shared network links without reserving a dedicated path.

What defines connectionless packet-switched service?

Packets can be transmitted without establishing a connection between the sender and receiver beforehand.

Circuit vs packet switching: key trade-off?

Circuits reserve capacity and can offer QoS; packets use capacity efficiently but may incur delay or loss.

Large vs small packets: main trade-off?

Large packets reduce header overhead (i.e. require fewer headers for the same amount of data transmitted) but make a retransmission costlier if lost or corrupted.

How does a bus topology operate?

Devices share a backbone; frames propagate along it. Simultaneous sends can collide.

How does a ring topology regulate access?

Each device connects to two neighbours; a circulating token grants permission to transmit.

Star topology: hub vs switch?

Hub broadcasts to all connected stations; switch forwards known frames only toward the destination.

What happens if a link fails in mesh, star, bus, or ring?

Mesh: alternate links remain. Star: one device affected. Bus backbone: whole LAN affected. Ring: may break ring.

What does IP best-effort delivery mean?

IP tries to forward packets but does not guarantee delivery, correct order, or freedom from corruption.

How is an IP address divided conceptually?

Net ID identifies the destination network; Host ID identifies a host within that network.

Why can routers use Net ID instead of a route for every host?

Hosts on the same network share a Net ID, so routers can use one routing entry for the entire network instead of one per host.

Why use network layering?

Abstract away underlying technologies so applications can use common services across different networks.

What is a networking protocol?

Rules and conventions governing communication between peer entities at the same layer.

What is a service in a layered architecture?

A function provided by one layer to the layer immediately above it.

What are peer processes?

Corresponding entities at the same network layer on different machines.

SDU vs PDU?

SDU: data passed down by the upper layer. PDU: that data plus the current layer’s protocol information.

What is encapsulation?

Each downward layer adds its own header or control information around the data.

Name the seven OSI layers, top to bottom.

Application, Presentation, Session, Transport, Network, Data Link, Physical.

Name the five practical TCP/IP layers, top to bottom.

Application, Transport, Network, Data Link, Physical.

What is the Physical Layer responsible for?

Transmitting bits as physical signals, including signal levels, bit duration, and data rate.

What is the Data Link Layer responsible for?

Hop-to-hop frame delivery over a link, including framing, MAC addressing, error control, flow control, and access control.

What does Medium Access Control (MAC) do?

Decides which station may transmit when several share the same medium.

Data Link vs Network Layer: delivery scope?

Data Link moves frames across one link/hop; Network routes packets across multiple links/networks.

What functions belong to the Network Layer?

Logical addressing, routing, congestion control, and fragmentation/reassembly.

Network vs Transport: where is delivery targeted?

Network gets packets to the destination host; Transport gets data to the destination process.

Which addressing mechanism reaches the correct application on a host?

Transport-layer port addressing (the destination port number).

MAC vs IP vs port address: what does each identify?

MAC: local network interface; IP: host/network across the Internet; port: process/application on that host.

Why are MAC addresses unsuitable for scalable WAN routing?

They do not identify a network’s location/hierarchy, so routers cannot aggregate routes efficiently.

Data Link vs Transport flow/error control?

Data Link operates across one hop; Transport performs control end-to-end between processes.

What are the data-unit names from Transport down to Physical?

Transport: segment; Network: packet; Data Link: frame; Physical: bit.

Which stack layers does a router normally implement?

Physical, Data Link, and Network (three layers in the 5-layer model).

Which addresses do TCP, IP, and Data Link headers carry?

TCP: ports; IP: source/destination IP addresses; Data Link: source/destination MAC addresses.

Do hosts on different LAN technologies need to understand each other’s Data Link protocols?

No. Each uses its local link technology; IP connects the dissimilar networks.

ipconfig vs ipconfig /all?

ipconfig: IP settings, mask, gateway. /all additionally includes MAC address, DNS and DHCP details.

What do ping and tracert test?

ping: reachability and round-trip delay. tracert: hops along the route and their delays.

Analog vs digital DATA?

Analog data has continuously varying values; digital data has discrete values.

Analog vs digital SIGNAL?

Analog signal: continuous amplitudes. Digital signal: limited discrete signal levels.

What are the three sine-wave parameters?

Peak amplitude A, frequency f, and phase φ (offset relative to time zero).

What phase puts a sine wave at its positive maximum at t = 0?

+90° or +π/2 radians.

What does a periodic vs nonperiodic composite spectrum look like?

Periodic: discrete frequency spikes. Nonperiodic: a continuous range of frequencies.

What are the axes of time-domain vs frequency-domain plots?

Time domain: amplitude vs time. Frequency domain: amplitude vs frequency.

In a frequency-domain spike plot, what do location and height mean?

Horizontal position = component frequency; spike height = peak amplitude.

How is signal bandwidth defined?

Highest frequency minus lowest frequency present: B = f_high − f_low.

What is the frequency bandwidth of a pure sine wave?

0 Hz: it contains just one frequency, regardless of the frequency value.

What is the theoretical bandwidth of an ideal digital signal?

Infinite: abrupt changes require an unlimited set of frequency components.

Signal bandwidth vs channel bandwidth?

Signal: frequencies the signal contains. Channel: frequencies the medium passes satisfactorily.

Low-pass channel vs band-pass channel?

Low-pass passes from 0 Hz to a cutoff; band-pass passes a range starting above 0 Hz.

Three main signal impairments?

Attenuation (loss of strength), delay distortion (shape changes), and noise (unwanted signals).

What does the sign of a power change in dB mean?

Negative dB = power loss; positive dB = gain; 0 dB = no change.

dB vs dBm vs dBW?

dB: power ratio/change. dBm: absolute power referenced to 1 mW. dBW: absolute power referenced to 1 W.

What is delay distortion?

Different frequencies arrive with different delays, changing signal shape and causing intersymbol interference.

What does signal-to-noise ratio (SNR) measure?

Useful signal power divided by noise power; higher SNR generally means clearer transmission.

What transmission uses low-pass vs band-pass channels?

Baseband digital transmission: low-pass. Broadband/carrier-based transmission: band-pass.

How are digital data and analog data sent as digital signals?

Digital data → line coding; analog data → PCM (digitization).

What methods support digital transmission of digital data?

Line coding maps bits to digital signals; block coding adds redundant bits before line coding.

What is line coding?

Mapping a stream of binary data bits onto discrete digital signal levels.

What are the two big design concerns for line coding?

Avoid an unwanted DC component and provide clock synchronization.

What is unipolar line coding, and its weakness?

Uses zero and one nonzero voltage; has a DC component and poor synchronization on repeated bits.

NRZ-L vs NRZ-I: what encodes bits?

NRZ-L: absolute voltage level. NRZ-I: transition = 1; no transition = 0.

Which bit runs cause synchronization problems in NRZ-L and NRZ-I?

NRZ-L: long runs of either bit. NRZ-I: long runs of 0s.

How does RZ encode a bit, and what is its trade-off?

Uses positive/negative level then returns to zero mid-bit: good timing, greater bandwidth.

Manchester encoding: what do mid-bit transitions represent?

0: positive-to-negative; 1: negative-to-positive. Every bit has a mid-bit timing transition.

Differential Manchester: what represents 0 vs 1?

Always transitions mid-bit for timing; at bit start, transition = 0 and no transition = 1.

How does AMI (bipolar) coding encode bits?

0 = zero voltage; successive 1s alternate positive and negative pulses.

What is AMI’s main synchronization weakness?

A long run of 0s has no transitions, so the receiver may lose bit timing.

What does B8ZS scrambling do?

Replaces eight consecutive AMI zeros with 000VB0VB to introduce timing transitions.

What is 2B1Q line coding and its trade-off?

Encodes 2 bits in one of 4 voltage levels: fewer signal elements but harder level detection/synchronization.

Line coding vs block coding?

Line coding encodes a bit stream as signals; block coding maps groups of bits to redundant codewords.

How does 4B/5B block coding help NRZ-I?

Converts 4 bits into 5 chosen bits so no more than 3 consecutive 0s occur; improves synchronization.

Data rate vs signal (baud) rate?

Data rate = bits per second; baud rate = signal elements per second. One element may encode several bits.

Which two methods digitize analog data for digital transmission?

Pulse Code Modulation (PCM) and Delta Modulation (DM).

What is PCM used for?

Converting analog data into digital bits through sampling, quantization, and binary encoding.

What are the three stages of Pulse Code Modulation (PCM)?

Sampling in time → quantization of amplitude → binary encoding.

What is the Nyquist sampling rule?

To reproduce a band-limited analog signal, sample at least twice its highest frequency: f_s ≥ 2f_max.

Why is bandwidth alone insufficient to sample a band-pass signal?

Sampling depends on f_max, not just bandwidth; you must know where the band ends.

What happens when an analog signal is undersampled?

Aliasing: its reconstructed waveform can appear to have the wrong frequency.

What is quantization in PCM?

Rounding each sampled amplitude to its nearest allowed discrete level.

How many quantization levels are possible with m bits per sample?

M = 2^m levels; equivalently m = log₂(M) bits per sample.

What is quantization error/noise?

Difference between the original sample and the rounded quantized value; prevents exact recovery.

How is PCM bit rate determined?

Samples per second × bits per sample.

PCM vs Delta Modulation (DM)?

PCM encodes absolute sample amplitudes; DM encodes whether a stepwise approximation moves up or down.

How does Delta Modulation represent each sample?

One bit: 1 moves the reference staircase up by a fixed step; 0 moves it down.

What two errors can affect Delta Modulation?

Slope-overload distortion if steps cannot follow fast changes; quantization noise from fixed-step approximation.