Spectrum Access and Medium Access Control

Introduction

Radio spectrum is a shared and physically constrained resource. A wireless network must determine both how spectrum is partitioned and which device may transmit at a given time. Multiple-access techniques provide separability in frequency, time, code, or orthogonal subcarriers. Medium access control (MAC) then applies contention, carrier sensing, scheduling, or combinations of these mechanisms.

These choices affect capacity, latency, energy consumption, coverage, fairness, mobility support, and interference. They also operate within regulatory arrangements ranging from exclusive licensing to unlicensed and tiered spectrum sharing.

Learning Objectives

After studying this chapter, students should be able to:

1. Compare FDMA, TDMA, CDMA, and OFDMA.
2. Analyze random-access, carrier-sensing, and scheduled MAC protocols.
3. Explain FDD, TDD, and in-band full-duplex operation.
4. Calculate throughput, SINR, processing gain, and basic wireless link budgets.
5. Evaluate scheduling algorithms in terms of throughput, fairness, and delay.
6. Explain interference-management and spectrum-sharing mechanisms.
7. Select appropriate access mechanisms for practical wireless applications.

1. Dimensions of Multiple Access

Multiple access allows several users to share a communication system. Users can be separated along one or more dimensions:

• Frequency: users occupy different frequency bands.
• Time: users transmit in different time intervals.
• Code: users employ distinguishable spreading sequences.
• Orthogonal subcarriers: users receive subsets of closely spaced subcarriers.
• Space: antennas separate users through beams or spatial streams.

The first four correspond principally to FDMA, TDMA, CDMA, and OFDMA. Modern systems often combine them with spatial division multiple access and centralized scheduling.

2. Frequency Division Multiple Access

In FDMA, the available bandwidth is divided into non-overlapping channels. A user is allocated one channel for the duration of a session or transmission.

If total usable bandwidth is B_T, each user channel has bandwidth B_c, and adjacent channels require guard bandwidth B_g, an approximate channel count is

Here, N is dimensionless, while all bandwidth quantities must use the same unit. Edge guards and unequal channel plans may reduce the actual number.

Guard bands limit adjacent-channel interference caused by imperfect filters. FDMA offers continuous transmission and modest synchronization requirements, but fixed allocation is inefficient when traffic is bursty. Traditional analog cellular and narrowband radio systems are representative applications.

3. Time Division Multiple Access

TDMA divides a carrier into repeating frames, each containing time slots. Multiple users can therefore occupy the same frequency at different times.

If a frame contains N_s equal slots and a user receives n_i slots, its ideal share of the carrier rate is

where R_i is the user’s ideal rate in bit/s and R_c is the carrier rate in bit/s. Protocol headers, guard times, retransmissions, and unused slots reduce the delivered rate.

Transmitters must be synchronized so that bursts do not overlap. A guard time accommodates timing error and propagation delay. TDMA supports discontinuous transmission, which can save terminal energy, but long frame durations may introduce latency.

Worked Example: TDMA Efficiency

A 1 Mbit/s carrier uses eight slots per frame. Each slot lasts 0.50 ms, of which 0.04 ms is guard and synchronization overhead. A user receives two slots.

The usable fraction per slot is

The user’s payload rate is

Thus, the payload rate is 230 kbit/s, assuming no additional coding or retransmission overhead.

4. Code Division Multiple Access

CDMA users transmit simultaneously over the same nominal time-frequency resource using distinct spreading sequences. A data symbol is multiplied by a higher-rate chip sequence. The receiver correlates the received waveform with the desired sequence.

The processing gain is approximately

where G_p is dimensionless, W is spread bandwidth in hertz, and R_b is information bit rate in bit/s. This approximation assumes comparable definitions of occupied spread bandwidth and chip rate.

For example, spreading a 20 kbit/s signal over 1 MHz gives

The decibel value is 10log₁₀(50) ≈ 17.0 dB. Processing gain enables the receiver to suppress some narrowband interference, but it does not eliminate multiple-access interference.

Near–Far Problem and Power Control

A nearby mobile can arrive at the base station with much greater power than a distant mobile, overwhelming correlation-based separation. Fast uplink power control is therefore important. Imperfect code orthogonality, multipath, and power-control errors make CDMA capacity interference limited rather than defined by a fixed number of channels.

Receivers may use matched filters, RAKE combining, interference cancellation, or multiuser detection. CDMA also supports soft handover, in which more than one base station participates in a connection.

5. OFDM and OFDMA

Orthogonal frequency division multiplexing (OFDM) divides a broadband channel into many narrowband subcarriers. Over the useful symbol duration T_u, adjacent subcarrier spacing is commonly selected as

where Δf is in hertz and T_u is in seconds. The subcarriers overlap spectrally but remain orthogonal because the integral of any two distinct subcarriers over T_u is zero under ideal synchronization.

OFDMA extends OFDM by assigning different groups of subcarriers and OFDM symbols to different users. A scheduler can exploit variations in channel quality across users, frequency, and time.

Figure: An OFDMA scheduler allocates two-dimensional time-frequency resource blocks to different users.

5.1 Cyclic Prefix

A cyclic prefix copies the end of an OFDM symbol and places it before the symbol. Let the sampled channel impulse response contain L_h significant taps. To prevent intersymbol interference from entering the retained useful interval and to preserve the circular-convolution model, the cyclic-prefix length must satisfy

Equivalently, in continuous-time engineering terms, the cyclic-prefix duration should be at least as long as the relevant maximum channel excess delay, with allowance for timing uncertainty. The cyclic prefix must be sufficiently long relative to the channel impulse response; it does not make an arbitrarily long channel harmless.

When this condition and synchronization assumptions hold, the channel’s linear convolution appears as circular convolution over the retained useful samples. The discrete Fourier transform then converts it into per-subcarrier multiplication, permitting simple frequency-domain equalization.

A longer prefix improves tolerance to delay spread but reduces efficiency. If T_CP is prefix duration, the symbol-time efficiency is

Worked Example: Cyclic-Prefix Overhead

Let Δf = 15 kHz, so

With T_CP = 4.76 μs,

The cyclic-prefix overhead is approximately 6.7%. If the relevant maximum excess delay is 3.2 μs and timing is properly controlled, this prefix is sufficiently long.

5.2 Practical OFDMA Limitations

OFDMA requires frequency and timing synchronization. Carrier-frequency offset destroys orthogonality and creates intercarrier interference. High peak-to-average power ratio also reduces power-amplifier efficiency. These disadvantages must be balanced against scalable bandwidth, frequency-selective scheduling, and straightforward equalization.

6. Random Access

Random-access protocols let devices contend for transmission opportunities. They are attractive when traffic is sporadic because resources need not be reserved continuously.

6.1 ALOHA

In pure ALOHA, a station transmits whenever a frame is ready. Under the classical assumptions of equal frame durations, Poisson attempts, independent users, and destructive collisions, normalized throughput is

where G is the mean offered traffic in attempts per frame time and S is successful frames per frame time. Its maximum is 1/(2e) ≈ 0.184 at G = 0.5.

Slotted ALOHA restricts attempts to slot boundaries:

Its maximum is 1/e ≈ 0.368 at G = 1. These results are theoretical benchmarks, not universal performance formulas.

6.2 Random Access in Cellular Systems

A cellular terminal may initially lack a scheduled uplink resource. It transmits a random-access preamble, after which the base station can provide timing adjustment and scheduled resources. Colliding or undetected preambles cause backoff and retransmission. Access parameters must accommodate both ordinary arrivals and bursts such as emergency events or mass machine reconnection.

7. Carrier-Sensing Protocols

Carrier-sense multiple access (CSMA) requires a station to listen before transmitting. If the channel is busy, it defers according to the protocol.

Common policies include:

• 1-persistent: transmit immediately when the channel becomes idle.
• Non-persistent: wait for a random interval before sensing again.
• p-persistent: in slotted operation, transmit with probability p when idle.

Collision detection is difficult in wireless devices because a transmitter’s own signal can be many orders of magnitude stronger than a received signal. Wireless LANs therefore use collision avoidance, acknowledgements, random backoff, and optionally request-to-send/clear-to-send exchanges.

Hidden and Exposed Terminals

A hidden terminal cannot sense another transmitter but can interfere at a common receiver. An exposed terminal unnecessarily defers after hearing a transmission even though its own transmission might not harm the relevant receiver. Carrier sensing therefore gives only a local estimate of interference conditions.

Contention performs well under light and moderate bursty load, but collisions and backoff delays grow under high load. It also provides weaker deterministic latency guarantees than controlled scheduling.

8. Centralized Scheduling

In a scheduled network, a base station or access point selects users, resource blocks, rates, and possibly transmit powers. Scheduling can respond to channel quality, queue length, delay, service class, and fairness requirements.

8.1 Representative Scheduling Policies

Round robin rotates service among users. It is simple and fair in access opportunities but does not exploit channel variations.

Maximum-rate scheduling chooses the user with the highest instantaneous supported rate. It maximizes short-term throughput but may starve disadvantaged users.

Proportional-fair scheduling commonly selects

where r_i(t) is user i’s achievable rate in the current interval and R̄_i(t) is its historical average delivered rate.

A typical exponential update is

where 0 < α ≤ 1 controls averaging memory and x_i(t) is the rate actually delivered to user i during interval t. For an unselected user, x_i(t)=0, so its average normally decays. An equivalent implementation may apply this decay separately. Updating only the selected user without equivalent decay would change the intended metric.

Worked Example: Proportional-Fair Decision

Three users have achievable rates of 12, 8, and 4 Mbit/s. Their historical averages are 6, 2, and 2 Mbit/s. Their metrics are respectively 2, 4, and 2. User 2 is selected even though User 1 has the highest instantaneous rate.

For α=0.1, if User 2 receives 8 Mbit/s while the others receive zero,

Scheduling design must also consider packets rather than only physical-layer rates. Delay-sensitive traffic may require deadline-aware or queue-aware policies.

9. Duplexing

Duplexing separates uplink and downlink transmissions.

Frequency-Division Duplexing

FDD uses separate frequency bands for uplink and downlink. It supports simultaneous transmission and reception with duplex filters. It is well suited to wide-area coverage and relatively stable bidirectional traffic but requires paired spectrum.

Time-Division Duplexing

TDD uses one band and alternates between uplink and downlink time intervals. The traffic ratio can be adapted, although neighboring cells should coordinate switching patterns to limit base-station-to-base-station and terminal-to-terminal interference. TDD also enables channel reciprocity for beamforming when radio-frequency hardware is calibrated and the channel remains sufficiently stable.

In-Band Full Duplex

In-band full duplex attempts simultaneous transmission and reception on the same time-frequency resource. Its central challenge is self-interference. Passive isolation, analog cancellation, and digital cancellation must suppress leakage before it saturates the receiver. Residual self-interference and cross-link interference mean that doubling the link rate is not generally achievable.

10. Link Budget and Interference Analysis

A basic received-power link budget in decibels is

where P_R and P_T are received and transmit powers in dBm; G_T and G_R are antenna gains in dBi; L_P is propagation loss in dB; and L_M represents other losses or margins in dB. Gains and losses can be added in decibel form because the underlying linear power ratios multiply.

Receiver thermal-noise power is

in dBm, using the approximate room-temperature noise density of −174 dBm/Hz. Here, B is receiver noise bandwidth in hertz and F is receiver noise figure in decibels.

With multiple interferers, linear powers must be added:

All quantities in this expression are linear powers in consistent units, not dBm.

Worked Example: Interference-Aware Link Budget

A base station transmits 30 dBm through a 13 dBi antenna. The receiver antenna gain is 0 dBi, path loss is 126 dB, and other losses are 2 dB.

For a 10 MHz receiver bandwidth and 7 dB noise figure,

Suppose aggregate interference is −100 dBm. Converting and adding noise and interference gives

This corresponds to −95.24 dBm. Therefore,

The ideal Shannon upper bound is

where C is bit/s and SINR_linear is dimensionless. Here, SINR_linear = 10^10.24/10 ≈ 10.57, giving

This is an ideal bound for the assumed bandwidth and SINR. Coding gaps, pilots, cyclic prefixes, scheduling, fading, and protocol overhead lower practical throughput.

11. Interference Management

Interference management can operate in frequency, time, power, code, and space.

• Frequency reuse: increases separation between co-channel cells, sacrificing some spectral reuse.
• Power control: limits unnecessary interference while maintaining service quality.
• Fractional frequency reuse: assigns different reuse patterns to cell-center and cell-edge users.
• Inter-cell coordination: exchanges load or scheduling information to avoid harmful resource overlap.
• Beamforming: directs energy toward intended users and away from interfered receivers.
• Interference cancellation: estimates and subtracts known or decoded interference.
• Coordinated multipoint: coordinates geographically separated transmitters or receivers.
• Successive interference cancellation: decodes selected signals in sequence, provided received powers and coding permit reliable cancellation.

Aggressive interference suppression may require additional signaling, channel-state information, processing, and synchronization. The best method depends on whether interference is predictable, bursty, localized, or produced by independent networks.

12. Spectrum-Sharing Mechanisms

Spectrum governance determines who may transmit, where, at what power, and under what conditions.

12.1 Exclusive Licensing

A licensee receives defined usage rights in a band and geographic area. Exclusive licensing facilitates interference control and investment but can leave spectrum underused in some times or locations.

12.2 Unlicensed Access

Unlicensed bands permit compliant devices to operate without an individual spectrum license. Rules may specify power limits, occupied bandwidth, duty cycle, or listen-before-talk behavior. Access is flexible, but quality cannot normally be guaranteed because unrelated networks coexist.

12.3 Tiered and Dynamic Sharing

Tiered systems prioritize incumbents over secondary or general-access users. Databases, sensing, exclusion zones, and automated frequency coordination may protect higher-priority users.

Three cognitive-radio models are commonly distinguished:

• Interweave: a secondary user transmits when it detects an unused opportunity.
• Underlay: a secondary transmission is allowed while keeping interference at protected receivers below a specified constraint.
• Overlay: a secondary system exploits knowledge of primary signaling, potentially assisting or coding around it; this requires substantial information and coordination.

Spectrum sensing is characterized by probability of detection P_D and probability of false alarm P_FA. Increasing sensitivity often raises both. Missed detection can harm an incumbent, whereas false alarms waste opportunities. Hidden incumbents, fading, uncertain noise power, and malicious emulation complicate sensing.

13. Illustrative Case: Designing Access for a Campus Private Network

Background

This fictional case concerns a university deploying a 20 MHz TDD private cellular network for lecture video, laboratory sensors, autonomous carts, and emergency communications.

Problem

The network must support high downlink demand, sporadic sensor access, low-latency cart control, and users near a neighboring co-channel network. Indoor delay spread reaches 3.2 μs.

Analysis

A pure contention system would be simple for sensors but could produce unbounded backoff delay during busy periods. Fixed FDMA would waste resources because user demands vary rapidly. OFDMA permits frequency-selective scheduling and differentiated quality of service.

The selected OFDM configuration uses 15 kHz subcarrier spacing, giving a 66.67 μs useful symbol. A 4.76 μs cyclic prefix exceeds the measured relevant maximum excess delay and therefore satisfies the channel-memory requirement with limited margin, subject to timing control.

A TDD pattern favors downlink traffic but reserves periodic uplink opportunities. Neighboring systems must coordinate switching patterns to prevent cross-link interference. Proportional-fair scheduling serves ordinary users, while deadline-aware priority is applied to emergency and cart-control packets. Sensors use random access only for initial requests and then receive scheduled resources.

Edge users are protected through beamforming, power control, and avoidance of the resource blocks most heavily affected by the neighboring network. Continuous measurements monitor SINR rather than received signal strength alone.

Approach and Outcome

The illustrative design combines:

1. OFDMA for data-resource allocation.
2. Controlled random access for connection initiation.
3. Priority scheduling for safety-critical traffic.
4. Proportional fairness for elastic traffic.
5. Coordinated TDD and interference-aware allocation.

Trial measurements show that scheduled operation stabilizes delay during busy periods, while adaptive allocation improves utilization compared with fixed channel assignments. The result does not imply that OFDMA is universally superior; low-cost, low-duty-cycle devices may still favor simpler contention protocols.

Learning Points

• Multiple-access and MAC mechanisms should be combined rather than selected in isolation.
• Cyclic-prefix design must follow measured channel delay spread.
• Received power alone is insufficient where interference dominates.
• Fairness, latency, energy use, and regulatory constraints can conflict with maximum throughput.

14. Contemporary Developments and Good Practice

Machine-learning methods are being investigated for traffic prediction, channel allocation, beam selection, and contention adaptation. Their use must be constrained by reliability, training-data quality, computational cost, and explainability. A learned scheduler should not be assumed safe under traffic or propagation conditions absent from its training data.

Professional design practice should:

• measure delay spread, traffic, and interference before selecting parameters;
• distinguish theoretical capacity from delivered application throughput;
• update all users consistently in fairness algorithms;
• coordinate TDD patterns across neighboring cells;
• protect incumbent services and comply with emission limits;
• test overload, mobility, and failure conditions;
• preserve equitable access instead of optimizing only aggregate throughput;
• secure spectrum databases and control signaling against manipulation.

Chapter Summary

FDMA, TDMA, CDMA, and OFDMA separate users in frequency, time, code, and time-frequency resources respectively. FDMA is simple but inflexible; TDMA requires synchronization; CDMA depends on spreading, power control, and interference management; OFDMA enables fine-grained scheduling but requires accurate synchronization and an adequately long cyclic prefix.

Random access and carrier sensing are efficient for bursty light loads but experience collisions and variable delay under congestion. Central scheduling provides stronger control over rate, fairness, and latency. FDD, TDD, and full duplex make different trade-offs among paired spectrum, flexibility, coordination, and self-interference. Reliable network design combines link-budget calculations, SINR analysis, interference management, and an appropriate regulatory sharing model.

Key Terms

FDMA: Multiple access using separate frequency channels.

TDMA: Multiple access using distinct time slots.

CDMA: Simultaneous access distinguished by spreading codes.

OFDMA: Assignment of OFDM subcarriers and symbols to multiple users.

Orthogonality: Zero cross-correlation over a specified observation interval under stated synchronization conditions.

Cyclic prefix: Repeated end segment added before an OFDM symbol to manage channel memory.

Processing gain: Ratio of spread bandwidth to information rate.

Random access: Contention-based acquisition of transmission resources.

CSMA: Medium access based on sensing before transmission.

Hidden terminal: A transmitter unable to sense another transmitter that can interfere at a common receiver.

Proportional fairness: Scheduling that balances instantaneous opportunity against historical service.

FDD: Duplexing using separate uplink and downlink frequencies.

TDD: Duplexing using alternating uplink and downlink time intervals.

SINR: Desired signal power divided by combined interference and noise power.

Link budget: Accounting of gains and losses between transmitter and receiver.

Power control: Adjustment of transmit power to satisfy service and interference objectives.

Spectrum sensing: Detection of spectrum occupancy or protected transmissions.

Underlay sharing: Secondary access constrained by interference imposed on protected receivers.

Interweave sharing: Secondary use of detected temporal or spectral opportunities.

Review Questions

Short Answer Questions

1. Why are guard bands required in FDMA?
2. What condition must a cyclic prefix satisfy relative to channel memory?
3. Distinguish a hidden terminal from an exposed terminal.
4. Why is uplink power control important in CDMA?
5. What is the difference between FDD and TDD?

Long Answer Questions

1. Compare FDMA, TDMA, CDMA, and OFDMA in terms of allocation, synchronization, interference, flexibility, and implementation complexity.
2. Explain pure ALOHA, slotted ALOHA, and CSMA, including their principal assumptions and limitations.
3. Discuss round-robin, maximum-rate, and proportional-fair scheduling.
4. Explain major interference-management methods used in cellular networks.
5. Compare exclusive licensing, unlicensed access, and tiered dynamic spectrum sharing.

Analytical / Application Questions

1. A 5 MHz receiver has a 6 dB noise figure. Calculate its thermal-noise power and determine SNR if received signal power is −88 dBm and interference is negligible.
2. Design a MAC strategy for a network carrying periodic industrial control traffic and sporadic battery-powered sensor reports. Justify any hybrid approach.
3. An OFDM system has a useful symbol duration of 50 μs and a cyclic prefix of 4 μs. Calculate prefix efficiency and assess suitability for a channel whose relevant maximum excess delay is 5 μs.

Multiple Choice Questions with Answers

Question 1

Which multiple-access method assigns users subsets of orthogonal subcarriers and time intervals?

A. FDMA

B. TDMA

C. CDMA

D. OFDMA

Correct Answer: D. OFDMA

Question 2

An OFDM channel has a relevant maximum excess delay of 3.2 μs. Assuming proper timing, which cyclic-prefix duration is sufficient?

A. 2.0 μs

B. 3.0 μs

C. 4.0 μs

D. The cyclic prefix must be shorter than every channel path

Correct Answer: C. 4.0 μs

Question 3

Under the classical slotted ALOHA model, the maximum normalized throughput is approximately:

A. 0.184

B. 0.368

C. 0.500

D. 1.000

Correct Answer: B. 0.368

Question 4

In a proportional-fair averaging update, what delivered rate is normally used for an unselected user during that interval?

A. Its maximum achievable rate

B. Its previous average rate

C. Zero, or an equivalent decay operation

D. The selected user's delivered rate

Correct Answer: C. Zero, or an equivalent decay operation

Question 5

A transmitter emits 20 dBm, with 10 dBi transmit gain, 0 dBi receive gain, 110 dB path loss, and 2 dB additional loss. What is received power?

A. −82 dBm

B. −78 dBm

C. −102 dBm

D. 118 dBm

Correct Answer: A. −82 dBm

Question 6

Which duplexing method alternates uplink and downlink within the same frequency band?

A. FDD

B. TDD

C. FDMA

D. CDMA

Correct Answer: B. TDD

Question 7

A station that cannot sense another transmitter, although both can interfere at the same receiver, is experiencing which problem?

A. Exposed-terminal problem

B. Near–far problem

C. Hidden-terminal problem

D. Self-interference cancellation

Correct Answer: C. Hidden-terminal problem

Question 8

Which spectrum-sharing model allows secondary transmission while constraining interference at a protected receiver?

A. Underlay

B. Interweave

C. Exclusive licensing

D. Fixed TDMA

Correct Answer: A. Underlay

Question 9

Signal power is 10 mW, interference is 2 mW, and noise is 0.5 mW. What is the linear SINR?

A. 2

B. 4

C. 5

D. 20

Correct Answer: B. 4

Question 10

A 40 kbit/s signal is spread over 2 MHz. What is its approximate processing gain as a linear ratio?

A. 20

B. 40

C. 50

D. 80

Correct Answer: C. 50