How Mathematics Improves The World | Letting Many Wireless Systems Share the Same Air
Your phone is shouting into the same universe as everybody else’s phone.
So is the Wi-Fi router.
The aircraft radar.
The weather radar.
The navigation satellite.
The communications satellite.
The radio telescope trying to hear an unimaginably weak signal from deep space.
They are all using electromagnetic waves.
Yet your phone call usually does not become weather radar.
Your neighbour’s Wi-Fi usually does not erase your GPS.
A radio telescope does not simply surrender because millions of transmitters exist elsewhere on Earth.
This coexistence is not automatic.
It is engineered.
And once many systems want the same part of the spectrum, the engineering becomes deeply mathematical.
Quick Read
The electromagnetic spectrum contains a vast range of frequencies, but useful spectrum for particular wireless applications is scarce. Different frequency bands have different propagation, antenna, atmospheric and hardware properties. Regulators allocate bands and define technical rules so services can coexist. Engineers then design systems to share frequency, time, space and code resources while keeping interference within acceptable limits.
Mathematics helps through channel allocation, graph colouring, optimisation, probability, information theory, signal processing, antenna arrays, game theory and control. Two nearby transmitters may need different channels. Two far-apart transmitters can reuse the same channel. A base station can reduce power to limit interference. An antenna array can steer energy toward one user and nulls toward another. Scheduling can separate users in time. Spread-spectrum and coding methods can separate signals in code or statistical structure.
Sharing becomes especially difficult when one user is passive. A radio astronomy receiver or Earth-observation sensor may be listening for signals far weaker than ordinary communications. It does not transmit anything that lets other systems know it is present. NSF’s recent spectrum programmes explicitly focus on coexistence among commercial communications, radar, satellites, Earth sensing and passive scientific uses, including online interference prediction and computation of operating constraints that protect vulnerable users.
One-sentence answer: Mathematics improves the world by converting the electromagnetic spectrum from a chaotic commons into a coordinated shared resource, deciding which transmitters may use which frequencies, times, powers and directions so many wireless systems can coexist without overwhelming one another.
Frequency Is an Address in the Spectrum
An electromagnetic wave oscillates.
Frequency measures how many cycles occur each second.
One hertz means one cycle per second.
One megahertz means one million.
One gigahertz means one billion.
Different wireless systems occupy different frequency bands.
Frequency therefore acts somewhat like an address.
A receiver can tune to one band and reject much of the energy elsewhere.
But filters are not perfect.
Transmitters leak some energy outside their intended channels.
Receivers have finite selectivity.
Strong nearby signals can overwhelm weak desired ones.
So frequency separation reduces interference.
It does not make interference impossible.
Bandwidth: A Signal Occupies More Than One Exact Frequency
A useful communication signal is not usually a single infinitely narrow frequency.
Information modulates a carrier and spreads energy over a band of frequencies.
The width of that band is bandwidth.
More bandwidth can support higher data rates under suitable conditions.
But bandwidth is shared.
Give one service 100 MHz and those frequencies are unavailable for an incompatible nearby service unless the systems can coexist.
Spectrum management is therefore partly an allocation problem:
who gets which interval of frequency, where, when and under what technical constraints?
Why “Spectrum Is Scarce” Needs a Careful Meaning
The electromagnetic spectrum is enormous.
So how can it be scarce?
Because not every frequency is equally useful for every application.
Lower frequencies often propagate farther and penetrate obstacles better but require larger antennas for similar electrical size.
Higher frequencies can offer large bandwidths and small antennas but may suffer greater blockage, atmospheric absorption or shorter practical coverage depending on the band.
Hardware has limits.
International allocations exist.
Existing systems have decades of installed equipment.
Passive scientific observations need protected frequencies because nature itself emits at particular spectral lines.
The scarcity is not “we ran out of numbers”.
It is “many valuable uses want frequency regions with favourable physical and institutional properties”.
Interference: When the Receiver Hears Too Much
A wireless receiver tries to estimate a desired signal buried among noise and interference.
A simple measure is signal-to-interference-plus-noise ratio:
SINR = desired signal power ÷ (interference power + noise power)
High SINR makes decoding easier.
Low SINR increases error probability.
The system can improve SINR by:
- increasing desired transmit power;
- reducing interfering power;
- changing frequency;
- changing time slot;
- using directional antennas;
- moving the transmitter or receiver;
- using better coding and modulation;
- coordinating schedules.
Each lever has costs.
Mathematics decides how to trade them.
The Near–Far Problem: Loud Neighbours Can Drown Distant Friends
Imagine one transmitter one metre from a receiver and another one kilometre away.
If both transmit at similar power on the same frequency, the nearby signal can be enormously stronger.
The weak distant signal disappears underneath it.
This is the near–far problem.
Wireless systems use power control, scheduling and multiuser detection to manage it.
The lesson is important:
fair transmit power is not equal transmit power.
A nearby device may need to whisper so a distant device can be heard.
Path Loss: Radio Power Spreads With Distance
As radio waves propagate, received power generally decreases with distance.
In ideal free space, power density spreads approximately with the inverse square of distance.
Real environments add reflections, diffraction, absorption and shadowing.
Engineers model path loss using logarithmic relationships and empirically fitted exponents.
This creates the possibility of frequency reuse.
If two transmitters using the same channel are far enough apart, their mutual interference may be acceptably small.
Distance becomes a spectrum resource.
Frequency Reuse: The Same Channel Can Exist in Many Places
Early cellular networks were built around a powerful idea.
Divide the coverage area into cells.
Assign frequency channels so nearby cells that would interfere do not use the same frequencies at the same time.
Reuse those frequencies farther away.
The same spectrum can serve many more users because geography separates the transmitters.
Modern cellular networks use far more sophisticated reuse and interference-management techniques, but the principle remains:
a frequency is not consumed globally merely because someone uses it locally.
Channel Allocation Becomes a Graph-Colouring Problem
Suppose ten transmitters need channels.
Some pairs are close enough to interfere.
Others are far apart and can safely reuse the same frequency.
Build a conflict graph.
Each transmitter is a vertex.
Draw an edge between two transmitters that cannot share a channel.
Now assign colours so adjacent vertices have different colours.
Each colour represents a frequency channel.
Minimise the number of colours.
You have transformed radio planning into graph colouring.
The real world adds unequal interference, variable powers, multiple channels and stochastic traffic.
But the graph abstraction captures the first structural idea.
Graph Colouring Is Easy to Explain and Hard to Solve
For a small graph, children can colour by eye.
For a large arbitrary graph, finding the minimum number of colours is computationally difficult.
Practical systems therefore use heuristics, approximations, decomposition and problem-specific structure.
This is a useful correction to the phrase “just optimise it”.
Some optimisation problems become brutally hard as they scale.
Good engineering often aims for a very good answer quickly rather than a mathematically perfect answer too late.
Time Division: Share the Same Frequency at Different Moments
If two users interfere on the same frequency, they do not necessarily need different frequencies.
They can take turns.
User A transmits now.
User B transmits in the next slot.
This is time-division multiple access in broad spirit.
The scheduling problem becomes:
- who transmits next?
- for how long?
- with what priority?
- how do we meet latency deadlines?
- how do we remain fair?
Frequency scarcity becomes a scheduling problem.
Frequency Division: Give Each User a Different Lane
Another strategy is to divide bandwidth into subchannels.
Different users occupy different frequency intervals simultaneously.
Orthogonal frequency-division multiplexing goes further, using many closely spaced orthogonal subcarriers.
Orthogonality allows overlapping spectral shapes without mutual interference at ideal sampling points.
This is a particularly elegant mathematical idea:
signals can overlap physically and remain separable mathematically.
Code Division: Signals Can Share Time and Frequency Too
Spread-spectrum systems allow multiple users to occupy the same broad frequency range at the same time while using different codes or spreading sequences.
A receiver correlates the incoming signal with the desired user’s code.
Other users appear more like interference.
Good sequence design and power control become essential.
The larger lesson is that sharing has several dimensions:
- frequency;
- time;
- space;
- code;
- power;
- direction.
Mathematics keeps finding new axes along which users can separate.
Spatial Reuse: Antennas Can Point
An old radio tower radiates broadly.
Modern antenna arrays can shape beams.
By controlling the phase and amplitude across many antenna elements, the transmitted waves add constructively in some directions and destructively in others.
Beamforming concentrates energy toward a user.
Null steering reduces energy toward an interfered receiver.
Two users can sometimes share the same time and frequency if their spatial channels are sufficiently distinct.
Space becomes another orthogonal resource.
MIMO: More Antennas Can Create More Channels
Multiple-input multiple-output systems use several transmitting and receiving antennas.
Multipath, once treated only as a nuisance, can become useful.
Different propagation paths create a channel matrix.
Linear algebra reveals how many independent spatial streams the channel can support.
Singular-value decomposition turns the channel into approximately independent modes under suitable processing.
This is an astonishing example of Mathematics changing the interpretation of interference.
Multiple reflected copies can become extra capacity.
Shannon Capacity Returns
In the previous article on error-correcting codes, we met Shannon’s channel capacity.
For an ideal additive white Gaussian noise channel, the celebrated formula is:
C = B log₂(1 + S/N)
where C is capacity, B is bandwidth and S/N is signal-to-noise ratio.
Wireless sharing complicates the picture because interference from other users enters the denominator.
Give one user more power and their rate rises.
But their interference can reduce somebody else’s rate.
Now capacity is no longer an individual property.
It becomes a coupled network problem.
Power Control: Everyone Shouting Is Not Optimal
Suppose every phone transmits at maximum power.
Each signal becomes stronger.
So does the interference.
Battery drains faster.
Cell-edge users may still suffer because everyone else got louder too.
Power control tries to meet performance targets with as little unnecessary power as possible.
In some models, power-control algorithms iterate:
measure interference;
adjust power;
measure again;
converge toward a feasible operating point.
The network becomes a conversation among transmitters about how loudly each one needs to speak.
Fairness: Fast Users Can Consume the Whole Channel
Suppose a scheduler always serves whichever user currently has the best radio channel.
That maximises throughput.
Users near the base station receive excellent service.
Users behind a building may receive almost nothing.
A fair scheduler sacrifices some total throughput to ensure weaker users receive service.
Proportional-fair scheduling is one mathematical compromise used in wireless systems.
It rewards high instantaneous rate relative to a user’s own recent average, balancing opportunism and fairness.
The same question appears everywhere in this series.
Maximum total output is not the only legitimate objective.
Passive Spectrum Users: The Listener Who Never Talks Back
Wireless sharing becomes especially difficult when one user is passive.
A radio astronomy telescope listens.
It does not transmit a beacon saying:
Please stop. I am observing hydrogen from another galaxy.
Earth-observation satellites also measure weak natural emissions in protected bands.
These signals can be orders of magnitude weaker than ordinary communication signals.
NSF’s SWIFT and NewSpectrum programmes explicitly highlight coexistence with passive scientific uses because conventional sharing assumptions break down when the protected receiver cannot coordinate actively.
The mathematics has to protect a user who may be invisible to the transmitting network.
Radio Astronomy: Hearing the Universe Beneath Civilisation
Radio telescopes detect extraordinarily weak electromagnetic emissions from stars, galaxies, pulsars, molecular clouds and the early universe.
A nearby transmitter can be billions of times stronger at the receiver.
This is why radio quiet zones exist around major observatories and why certain bands receive special protection internationally.
But satellites complicate geographic isolation.
A constellation passes over every quiet zone.
Aggregate interference from thousands of transmitters can matter even when each individual transmitter meets limits.
The interference problem becomes a sum over an enormous moving network.
Aggregate Interference: Many Harmless Signals Can Add Up
Suppose one transmitter produces interference one hundred times below a harmful threshold.
Safe.
Now suppose one thousand similar transmitters are visible simultaneously.
Their powers add statistically and geometrically.
The aggregate may no longer be safe.
Regulation therefore cannot always examine devices one at a time.
Constellations and dense networks need system-level interference analysis.
This is the same lesson seen in traffic and power grids:
many individually acceptable actions can produce a collectively unacceptable state.
Weather Radar: The Receiver Protects Everyone Else
Weather radar transmits and receives carefully timed radio pulses to infer precipitation and atmospheric motion.
Interference can create false echoes or obscure real ones.
A spectrum-sharing decision that slightly improves broadband capacity but degrades severe-weather detection may have a terrible social trade-off.
This is why spectrum value cannot be measured only in commercial throughput.
Passive and safety-critical services may have enormous public value despite transmitting little data.
Mathematics can model interference.
Public policy decides what interference risk society accepts.
Adjacent-Channel Interference: Your Neighbouring Lane Still Matters
Two systems can be assigned different frequency channels and still interfere.
Transmitters emit some energy outside their allocated band.
Receivers admit some energy outside the desired band.
A very strong adjacent signal can desensitise the receiver.
Guard bands provide empty frequency space between services.
Better filters can reduce the required guard band.
This creates an engineering trade-off:
use more spectrum as empty protection, or spend more on hardware selectivity.
Mathematics turns the trade-off into measurable spectral masks and interference budgets.
Interference Temperature and Risk Budgets
Instead of protecting a receiver by giving it an enormous exclusive band everywhere, sharing frameworks can define an interference threshold.
Multiple transmitters may operate as long as the total interference at the protected receiver remains below the limit with sufficient probability.
This turns spectrum sharing into a risk-budget problem.
How much interference can each user contribute?
How uncertain is propagation?
What probability of exceeding the threshold is acceptable?
The answer depends on the service.
A streaming video can tolerate occasional packet loss.
A protected scientific measurement may not.
Propagation Models: Predict Where the Radio Energy Goes
To manage interference, engineers need path-loss models.
Free-space equations work for unobstructed ideal paths.
Urban environments require buildings, terrain and clutter.
Atmospheric ducts can carry signals unexpectedly far.
Rain attenuates some high-frequency bands.
Vegetation changes propagation.
Satellite paths vary with elevation angle.
A spectrum-sharing database therefore cannot simply draw a fixed circle around a transmitter.
Protection regions depend on frequency, power, antenna pattern, terrain, weather statistics and receiver sensitivity.
Geography becomes probability.
Dynamic Spectrum Access: The Channel Assignment Can Change
Traditional spectrum allocation is relatively static.
One service receives a band.
Another service uses another.
But many bands are underused at particular times or places.
Dynamic spectrum access allows secondary users to transmit when they can do so without harmful interference to protected incumbents.
The control loop may be:
observe → predict interference → compute allowed power/frequency → transmit → monitor → update
This is spectrum management becoming real-time optimisation.
Databases: Sometimes the Radio Asks Permission First
One sharing architecture uses a central database.
A device reports its location and technical parameters.
The database knows where protected incumbents operate.
A propagation model predicts interference.
The system returns which channels and powers are allowed.
This architecture converts regulation into a computational query.
But the database can be wrong.
Location can be wrong.
Propagation models can be wrong.
The incumbent may move.
So the system needs conservative margins and monitoring.
Spectrum Sensing: Listen Before You Talk
Another approach is sensing.
Before transmitting, a device listens to determine whether the channel appears occupied.
Wi-Fi uses carrier-sense principles at the medium-access layer.
Cognitive radio research explores more sophisticated spectrum sensing.
The difficulty is the hidden-node problem.
A transmitter may not hear an incumbent that the intended receiver can hear.
Absence of detected energy is not proof that transmitting is safe.
This is another epistemic boundary:
“I did not detect it” is weaker than “it is not there”.
Cooperative Sensing: Several Ears Are Better Than One
Multiple devices can share sensing reports.
One receiver sits in a shadow behind a building.
Another has clear line of sight.
Combine their observations and detection improves.
But cooperation creates new problems.
Which sensor is trustworthy?
How much bandwidth is consumed by sensing reports?
Could a malicious device report fake interference?
Distributed information improves observability and creates an authentication problem simultaneously.
Game Theory: Every Wireless Network Has Selfish Users
Suppose every transmitter chooses its own power to maximise its own data rate.
The individually rational strategy may be maximum power.
If everyone does that, interference explodes.
Everyone’s rate can fall.
This is a game-theoretic problem.
Pricing, coordination rules or distributed algorithms can align individual incentives with network performance.
The same structure appears in traffic route choice and common-resource management.
Uncoordinated rationality can produce collective stupidity.
Auctions: Sometimes Spectrum Is Allocated Through Economics
Governments often assign commercial spectrum licences through auctions.
This creates another mathematical domain.
How should bids be structured?
How do we prevent strategic gaming?
How do we allocate combinations of geographic and frequency blocks efficiently?
Combinatorial-auction theory, mechanism design and optimisation become part of radio engineering’s institutional layer.
Spectrum is physical.
Its allocation is also economic and political.
Machine Learning Can Predict Interference—But Physics Still Sets the Boundary
Modern spectrum research uses machine learning to predict channel occupancy, detect interference sources, estimate propagation and adapt resource allocation.
NSF’s spectrum-sharing programmes explicitly encourage online interference prediction and algorithmic computation of constraints.
But learned models inherit training limits.
A model trained in ordinary weather may fail during unusual atmospheric ducting.
A classifier may miss a rare incumbent signal.
A reinforcement-learning controller may exploit a simulation flaw.
Wireless safety therefore needs hybrid thinking:
learn from data, but keep hard physical and regulatory constraints where failure matters.
Robust Optimisation: Protect the Receiver Even When the Model Is Wrong
Suppose predicted path loss is 120 dB.
But the real path could vary by ±10 dB because of uncertain terrain, weather or clutter.
An optimiser using only 120 dB may choose a transmit power that violates the interference threshold in the worst case.
Robust optimisation asks for solutions that remain feasible over a defined uncertainty set.
Chance-constrained optimisation allows a small defined probability of violation.
The choice depends on the consequences.
A safety-critical receiver may justify more conservative protection than ordinary broadband traffic.
Primary Mathematics: Colour the Towers
Draw six circles on paper to represent radio towers.
Connect pairs that are close enough to interfere.
Give the child three coloured pencils representing three channels.
Adjacent circles cannot share a colour.
Can the network be coloured with three channels?
Now add one more tower.
Does the answer change?
Move one tower farther away and remove an interference edge.
Now a frequency can be reused.
The child has learned spectrum planning through graph colouring.
Secondary Mathematics: Decibels Change the Scale
Wireless engineering uses logarithmic decibels because signal powers span enormous ranges.
A power ratio of 10 corresponds to 10 dB.
A ratio of 100 corresponds to 20 dB.
A ratio of 1,000 corresponds to 30 dB.
Multiplication of linear gains becomes addition in decibels.
A link budget becomes arithmetic:
transmit power + antenna gain − path loss − other losses = received power.
Logarithms, once a school topic, become the natural language of radio.
Advanced Mathematics: The Air as a Shared Optimisation Problem
Modern wireless engineering draws on:
- Fourier analysis;
- complex numbers;
- linear algebra;
- probability;
- stochastic processes;
- information theory;
- convex optimisation;
- integer programming;
- graph theory;
- game theory;
- signal processing;
- control theory;
- machine learning.
Fourier transforms separate frequency components.
Linear algebra separates spatial streams.
Graph theory assigns channels.
Optimisation assigns power.
Probability handles fading.
Information theory defines capacity.
Game theory coordinates selfish users.
Your wireless connection is a crowded Mathematics conference happening invisibly around you.
A Classroom Thought Experiment: Everybody Talks at Once
Put six students in a room.
Each must read a different sentence to one partner across the room.
First, everyone speaks at once at normal volume.
Difficult.
Now assign three time slots.
Easier.
Now let pairs in opposite corners speak simultaneously.
Spatial reuse.
Now give each pair a different pitch range.
Frequency division.
The classroom becomes a wireless network.
A Second Thought Experiment: Protect the Quiet Scientist
One student tries to hear a whisper from the other side of the room.
Everyone else is allowed to talk.
How can the class communicate without destroying the whisper?
Options include:
- silence during listening intervals;
- distance noisy speakers from the listener;
- use different pitch bands;
- point speakers away;
- limit volume;
- coordinate schedules.
This is passive-spectrum protection in human form.
The quiet user does not need more power.
They need everyone else to respect an interference budget.
Why This Improves the World
1. It lets spectrum be reused
Geographic, temporal and spatial separation allows the same frequencies to support many more users.
2. It protects weak receivers from strong neighbours
Interference budgets, filters, power control and coordination keep communication systems from overwhelming one another.
3. It creates more capacity without creating more spectrum
MIMO, beamforming, frequency reuse and scheduling extract more independent communication opportunities from the same physical bands.
4. It allows active and passive systems to coexist
Protection calculations help commercial communications share the wider spectrum environment with radar, Earth sensing and radio astronomy.
5. It makes dynamic sharing possible
Databases, sensing and real-time optimisation can assign channels and powers according to current location and interference conditions.
6. It turns a public commons into governable infrastructure
Mathematical models give regulators and engineers a common language for technical limits, risk and coexistence.
What Mathematics Does Not Do
Mathematics does not decide which social use deserves spectrum most.
It does not make harmful interference morally acceptable because total throughput increased.
It does not make propagation models exact.
It does not guarantee a machine-learning controller will behave safely outside training conditions.
It does not eliminate hardware imperfections.
It does not protect scientific receivers if policy permits everyone else to overwhelm them.
And it cannot create infinite wireless capacity from finite bandwidth, power and noise.
Mathematics gives civilisation the trade-off surface.
Governance chooses where on that surface society wants to operate.
Frequently Asked Questions
What is radio spectrum?
Radio spectrum is the portion of the electromagnetic spectrum used for radio communication, sensing, navigation and scientific observation. Different services use different frequency bands according to physical properties and regulatory allocations.
What is interference?
Interference is unwanted electromagnetic energy that degrades a receiver’s ability to detect or decode its desired signal. It can come from co-channel users, adjacent channels, spurious emissions or natural noise.
What is frequency reuse?
Frequency reuse means using the same frequency channel in multiple places far enough apart, or sufficiently isolated by antennas and power control, that mutual interference stays acceptable.
How does graph colouring relate to spectrum?
A conflict graph can represent transmitters as vertices and harmful-interference relationships as edges. Assigning different channels to conflicting transmitters becomes analogous to colouring adjacent vertices with different colours.
Why are passive spectrum users difficult to protect?
Passive users such as radio telescopes and Earth-observation sensors only receive; they do not transmit signals that sharing systems can easily detect or coordinate with. Their desired signals can also be extremely weak, requiring very low interference thresholds.
Can AI solve spectrum sharing?
AI can help predict occupancy, detect interference and allocate resources, but it does not eliminate physical, regulatory or safety constraints. Robust spectrum management still requires validated propagation models, uncertainty handling, monitoring and explicit protection criteria.
Sources and Further Reading
- U.S. National Science Foundation, VINES — Spectrum Sharing-Native Design, describing spectrum sharing as increasingly crucial for future network generations.
- U.S. National Science Foundation, SWIFT spectrum-sharing research programme, including coexistence, online interference prediction and protection of passive scientific users.
- U.S. National Science Foundation, Next Era of Wireless and Spectrum, on cooperative and dynamic spectrum-sharing research.
- International Telecommunication Union, ITU Radiocommunication Sector, which coordinates international radio-spectrum allocations and technical recommendations.
Continue Through eduKateSG
Continue with How Mathematics Works. Batch 003 also includes Making Traffic Lights Cooperate Instead of Compete, Finding Useful Pages in a Web Too Large to Read, and Designing an Evacuation Before the Emergency.
Final Thought: The Air Is Quiet Only Because the Systems Are Coordinated
Stand in the middle of Singapore.
The air feels empty.
It is full of radio.
Thousands of cellular connections.
Wi-Fi.
Broadcasting.
Navigation.
Satellites.
Radar.
Scientific sensing.
They coexist because civilisation has learned not merely to transmit.
It has learned to take turns.
Separate frequencies.
Reuse space.
Control power.
Point antennas.
Protect weak listeners.
And compute what happens before everyone speaks at once.
The air is shared.
Mathematics is part of the etiquette.