From radio waves to your pocket: how cellular and Wi-Fi actually carry data
Photo: N43 and HermesEvery tap, stream, and message rides on radio waves carved into carefully divided slices of spectrum. Here is how cellular towers and Wi-Fi routers turn those invisible waves into the data your phone uses — and why spec sheets rarely match reality.
01The shared spectrum problem
Wireless data always begins with the same physical constraint: there is only so much radio spectrum to go around. Every frequency band is a finite resource, and every phone, router, and tower broadcasting in a band is talking over everyone else in range. Regulators slice the spectrum into licensed bands, sold to mobile operators, and unlicensed bands, such as the 2.4 GHz and 5 GHz ranges that Wi-Fi uses, where any device may transmit as long as it obeys power and interference rules.
A cellular network deals with scarcity by dividing coverage over land areas called cells, each served by fixed-location transceivers. Each cell typically uses different frequencies from its neighbors to avoid interference, so the same channels can be reused again a few cells away. That reuse is the whole trick: the network's total capacity grows not by making any single transmitter faster, but by tiling the map with thousands of small, carefully separated cells.
Wi-Fi solves the same problem differently. Instead of a central planner assigning frequencies, unlicensed-band devices listen before transmitting and back off when the air is busy. It is a polite free-for-all that works well in a home or café and strains when dozens of devices crowd onto one channel.
02What each generation actually changed: 1G to 5G
Cellular generations are best understood as answers to the question of how cleverly you can encode data into a fixed slice of radio spectrum. 1G carried analog voice with no meaningful data at all. 2G digitized the signal and introduced text messaging; GSM's circuit-switched data limped along at roughly 0.064 Mbps, one modest phone call's worth of capacity. 3G, standardized around the ITU's IMT-2000 family, brought packet switching and peak targets near 2 Mbps, enough for early mobile browsing.
4G, defined against the IMT-Advanced framework, moved to an all-IP network and pushed the official peak target to about 1,000 Mbps (1 Gbps). 5G goes further still: it is built on the 3GPP 5G system and the New Radio (NR) air interface, first specified in 3GPP Release 15 and developed to meet the ITU IMT-2020 framework, with large-scale deployments beginning in 2019. Its IMT-2020 peak data-rate target sits at roughly 20,000 Mbps (20 Gbps) under ideal conditions.
The chart below shows those headline targets on a logarithmic scale, because each generation's leap is roughly an order of magnitude. Note what the numbers are: specification ceilings for ideal lab conditions, not the speeds a phone achieves on a busy street.
03How your phone finds and talks to a cell tower
Before any data flows, the phone has to introduce itself. It continuously scans the known frequency bands, measuring the strength of broadcast signals from nearby cells. Each cell announces itself on well-known channels, and the phone ranks the candidates by signal quality and network priority, then requests access through the best one. This process — cell search, selection, and attach — happens every time you switch a phone on, cross between towers, or return to coverage after a flight.
Once attached, the tower and phone negotiate what they can each do: which frequency band to use, how much bandwidth, which modulation scheme, and how the link will be scheduled. The base station acts as an air-traffic controller, granting each device short, precise time slots and frequency blocks so that hundreds of connections in one cell stay separated. As you move, the network hands your connection from cell to cell, reusing those tiled frequencies so the conversation never drops.
Uplink and downlink are deliberately asymmetric. Towers transmit with far more power than a battery-powered phone can, so download capacity always exceeds upload. Scheduling also adapts in real time: a phone at the cell edge gets a more robust, slower modulation than one standing beside the mast, which is a big part of why measured speeds swing so widely within a single cell.
04Wi-Fi: the other wireless network in your pocket
Cellular and Wi-Fi are siblings built on the same physics, raised under different rules. Wi-Fi lives in the unlicensed bands and is standardized by the IEEE as the 802.11 family, with friendly generation names assigned by the Wi-Fi Alliance. Your phone quietly decides all day whether a given task belongs on the cellular network or on whichever Wi-Fi network it trusts, and most of the world's mobile data actually travels over Wi-Fi for at least part of its journey.
The generations have climbed the same way cellular has. Wi-Fi 4 (802.11n) reached a maximum physical-layer rate of 0.6 Gbps. Wi-Fi 5 (802.11ac) pushed that to 6.9 Gbps by leaning on wider channels and the cleaner 5 GHz band. Wi-Fi 6 (802.11ax) refined efficiency and scheduling to hit 9.6 Gbps, and Wi-Fi 7 extends the story further.
Wi-Fi 7 is formally IEEE 802.11be-2024, dubbed Wi-Fi 7 by the Wi-Fi Alliance and also known as Extremely High Throughput (EHT). It builds on Wi-Fi 6 in the 2.4, 5, and 6 GHz bands, adding wider 320 MHz channels, higher-order modulation, and multi-link operation that lets a device use several bands at once. As with cellular, the headline number is a sum of every best-case factor stacked together.
05Bandwidth, latency and what fast really means
Bandwidth is how much data a link can carry at once — the number every marketing page leads with. Latency is how long a single piece of data takes to get there. They are different quantities, and everyday internet experience depends on both. A speed-test result of hundreds of megabits per second still feels sluggish if each round trip takes 80 milliseconds, because page loads, video calls, and games are chains of many small round trips.
The spec-target charts above describe bandwidth ceilings. Real-world throughput is always lower: peak rates assume ideal signal strength, maximal channel width, minimal interference, and a single device monopolizing the cell. Share the cell with a stadium crowd, stand behind a concrete wall, or fall back to an older band, and the achievable fraction of that ceiling shrinks dramatically. Latency is improving more visibly in 5G than raw throughput, thanks to leaner frame structure and edge computing that moves servers physically closer to towers.
The practical takeaway is to match the metric to the task. Streaming resolution is a bandwidth question. Call quality, cloud gaming, and responsiveness are latency questions. Capacity at a crowded venue is a scheduling and cell-density question. No single number on a box answers all three.
06Keeping wireless data secure from air to core
Radio is broadcast by nature: every transmission is, in principle, receivable by anyone nearby with the right hardware. Wireless security therefore has to assume the air is hostile. Cellular systems have evolved from 2G's easily cracked encryption through 3G and 4G's stronger mutual authentication to 5G, where the subscriber identity itself is encrypted over the air and the network authenticates both directions — the phone verifies the network, not just the reverse.
Wi-Fi made the opposite journey from its early days. The original WEP encryption was broken so thoroughly that it now serves mostly as a warning label; WPA2 replaced it, and WPA3 has been the requirement for new certifications since 2020, resisting offline dictionary attacks against weak passwords. On either network, application-layer encryption — the padlock in your browser — provides a second, independent layer that protects data even end to end.
Security also spans the whole path, not just the radio hop. SIM credentials bind an account to a device, operator cores enforce policy and isolate traffic, and Wi-Fi networks segment guests from trusted devices. A realistic threat assessment treats each layer as one lock in a series rather than trusting any single one.
07What 6G research is actually exploring
6G today is a research agenda, not a product. Standardization bodies are only beginning to shape requirements, while universities and vendors explore candidate technologies. The most cited direction is the use of much higher frequency bands — sub-terahertz spectrum — where enormous contiguous bandwidths exist, at the cost of signals that travel shorter distances and pass through walls even less readily than millimeter waves.
Other threads aim at intelligence and sensing rather than raw speed. Researchers are studying networks that use AI for spectrum scheduling and beam management, cells that double as radars capable of sensing motion and shape, and integrated satellite and terrestrial coverage aimed at closing dead zones. Energy per bit is increasingly treated as a first-class design goal, since the dense small-cell geometry that raises capacity also multiplies powered equipment.
Set expectations accordingly: 5G's IMT-2020 targets took years of 3GPP releases and network investment to reach everyday phones, and whatever 6G becomes will follow the same path from paper to pocket. The useful skill is reading any generation's headline number for what it is — a shared ceiling under ideal conditions — and judging networks by the latency, coverage, and capacity you actually experience.
By N43 and Hermes for Sailor Bob News.





