Static Medium Access — TDMA, FDMA, CDMA

Pre-allocated time, frequency, or code slots. The schemes mobile and satellite links rely on, and why static beats stochastic for them.

Concept Intermediate
8 min read
mac tdma fdma cdma channel-partitioning

Summary#

When many stations share one medium, somebody has to decide who transmits when. The static approach — channel partitioning — slices the medium into fixed allocations and hands each station its own slice. FDMA (Frequency-Division Multiple Access) gives each station a private frequency band. TDMA (Time-Division Multiple Access) gives each station a recurring time slot. CDMA (Code-Division Multiple Access) lets every station transmit at once on the same band but with mathematically orthogonal codes the receiver can untangle. All three are static in the sense that the allocation is pre-negotiated; stations don’t gamble on contention.

Static allocation is the opposite design choice from ALOHA / CSMA. There are no collisions because no two stations are ever active on the same (time, frequency, code) cell — but unused slots stay unused. For traffic that is steady and full-rate (a voice call, a satellite uplink at link capacity), static is optimal. For bursty, irregular traffic (an HTTP request, a Slack message), static wastes most of the channel.

Why it matters#

Static MAC is the protocol family that runs every cellular network, most satellite links, and every legacy radio broadcast system. Knowing which scheme each generation of mobile uses — GSM (TDMA), IS-95 / CDMA2000 (CDMA), LTE and 5G NR (OFDMA, a hybrid) — is table stakes for any conversation about wireless networks. It is also a clean worked example of a recurring design question: when do you partition a shared resource statically vs let users contend for it?

The same trade shows up far beyond radio. CPU schedulers choose between fixed time slices (real-time OSes) and contention (general-purpose schedulers). Storage QoS systems choose between reserved IOPS (static) and best-effort (contended). Once you see the pattern in TDMA / FDMA / CDMA, you see it in every multi-tenant system.

How it works#

FDMA — Frequency-Division Multiple Access#

Carve the available spectrum into N non-overlapping frequency bands; give each station one band. Stations transmit continuously, in parallel, on disjoint frequencies. The receiver tunes its filter to one band to pick out one station.

|◄── band 1 ──►|◄── band 2 ──►|◄── band 3 ──►|◄── band 4 ──►|
┌──────────────┬──────────────┬──────────────┬──────────────┐
│ Station A │ Station B │ Station C │ Station D │ Frequency
│ (continuous) │ (continuous) │ (continuous) │ (continuous) │
└──────────────┴──────────────┴──────────────┴──────────────┘
(time runs forever)

FDMA is the oldest scheme — AM and FM radio broadcast, classical analogue cellular (1G AMPS), and most satellite C-band uplinks use it. The hardware requirement is a sharp filter at the receiver; the design constraint is guard bands between channels to prevent inter-channel interference, which waste a few percent of the spectrum.

TDMA — Time-Division Multiple Access#

All stations share the full bandwidth but only one transmits at a time. The shared timeline is sliced into a recurring frame, and each frame is divided into slots; each station owns one slot per frame.

┌───┬───┬───┬───┬───┬───┬───┬───┬───┬───┬───┬───┐
│ A │ B │ C │ D │ A │ B │ C │ D │ A │ B │ C │ D │ Time
└───┴───┴───┴───┴───┴───┴───┴───┴───┴───┴───┴───┘
|◄── frame ──►|◄── frame ──►|◄── frame ──►|

TDMA is the scheme 2G GSM used: each 200 kHz GSM channel carried 8 time slots, so 8 voice calls multiplexed onto one frequency. The hardware requirement is precise time synchronisation — stations must transmit only inside their slot, which means GPS or network-supplied clocks. Guard times between slots cover propagation delay variance, which (like FDMA’s guard bands) waste a few percent.

CDMA — Code-Division Multiple Access#

All stations transmit at once on the same frequency at the same time — but each station’s signal is spread across the band by multiplying it with a unique pseudorandom code (a “chipping sequence”). The codes are chosen so they are nearly orthogonal: the receiver multiplies the received signal by the desired station’s code and the wanted signal pops out while the others average to noise.

┌───────────────────────────────────────────────────────────┐
│ A·codeA + B·codeB + C·codeC + D·codeD (all simultaneous) │ full band, full time
└───────────────────────────────────────────────────────────┘
receiver × codeA → A signal recovered
receiver × codeB → B signal recovered
(codes nearly orthogonal — others appear as noise)

CDMA was Qualcomm’s IS-95 (cdmaOne) and its 3G successors CDMA2000 and WCDMA. The hardware requirement is sophisticated DSP at the receiver to despread the signal and tight power control across stations (the near-far problem — a strong nearby transmitter drowns out a weak distant one even with orthogonal codes). The trade is that each user appears as low-power background noise to every other user, so capacity is soft — adding one more user slightly raises everyone’s noise floor rather than triggering a hard cutoff.

Where 4G and 5G fit#

4G LTE and 5G NR use OFDMA (Orthogonal Frequency-Division Multiple Access) — a hybrid. The spectrum is divided into many narrow sub-carriers (FDMA-style) but assignments rotate at fine time granularity (TDMA-style), so the scheduler at the base station can hand each user a different (sub-carrier, time-slot) mix every millisecond. This combines FDMA’s freedom from collisions with TDMA’s bursty-traffic efficiency — and is why modern cellular can carry both voice (small steady) and video streaming (large bursty) without separate physical channels.

Variants and trade-offs#

Static partitioning (FDMA / TDMA / CDMA / OFDMA). Guaranteed throughput per station. No collisions, no retransmissions from contention. Predictable latency — your slot recurs every frame. Best for steady, full-rate traffic (voice, real-time video, satellite uplinks). Wastes capacity when users are idle.
Stochastic / contention-based (ALOHA, CSMA, CSMA-CA). Statistical multiplexing — bursty users share the medium dynamically. High utilisation under low load. Collisions and back-offs cap throughput once load rises (~18% for pure ALOHA, ~37% slotted, ~90% switched Ethernet because there is no contention at all).

Comparing the three static schemes directly:

  • FDMA. Simplest hardware (just analog filters), no synchronisation needed, but rigid — once a band is assigned, no other station can use it even if the owner is silent. Guard bands waste spectrum.
  • TDMA. More efficient than FDMA when a station’s traffic is bursty, because unused slots within the frame can be reassigned. Requires tight time synchronisation across all stations. Guard times waste time.
  • CDMA. Most flexible — capacity degrades smoothly with users rather than hitting a hard limit, all users see the full bandwidth. Requires sophisticated DSP, power control, and longer code sequences for more users. Patent-encumbered through the 2010s, which slowed adoption.
Why does cellular use static MAC but Wi-Fi uses CSMA?

Different traffic models. A cellular tower coordinates a known set of registered users with predictable QoS requirements (voice calls need a guaranteed slot). A Wi-Fi access point serves an unknown set of devices with wildly bursty traffic (a phone idle 99% of the time, a laptop downloading at line rate, a doorbell sending one message a minute). Static allocation would waste most of Wi-Fi’s air time on idle clients; cellular’s base-station scheduling can re-allocate slots fast enough (per-millisecond in LTE/5G) to recover the efficiency. The two systems converge in OFDMA, which 5G uses for cellular and Wi-Fi 6 uses for Wi-Fi — same scheduling math, different deployment.

When this is asked in interviews#

Static MAC questions usually appear in two places: warm-ups on networking screens (“what are the three channel-partitioning schemes?”) and deeper systems-design conversations where the candidate is expected to reason about resource allocation. Strong answers tie the protocol to where it shipped: GSM is TDMA, IS-95 is CDMA, AM/FM is FDMA, LTE/5G is OFDMA.

Common follow-ups:

  • “Why was CDMA seen as superior to TDMA in the 2G era?” — Soft capacity, no hard call-blocking, easier frequency reuse across cells (the same band can be used in adjacent cells because codes orthogonalise them), and better resilience to multipath fading.
  • “What goes wrong if you use TDMA without tight time sync?” — Slots collide at their boundaries; the receiver gets garbled bits from two stations overlapping. GPS or backhaul-supplied clocks are mandatory.
  • “What is the near-far problem and how does CDMA handle it?” — A strong nearby transmitter saturates the receiver’s dynamic range, drowning weaker distant signals even with orthogonal codes. Cellular CDMA does aggressive closed-loop power control — every few milliseconds, the base station tells each handset to turn up or down to equalise received power.
  • “Where does OFDMA fit in the family?” — A hybrid: many narrow FDMA-style sub-carriers, with TDMA-style scheduling assigning them per-millisecond. Combines FDMA’s no-collision property with TDMA’s burst efficiency.
Search ESC

Keyboard shortcuts

Shortcuts are disabled while typing in inputs.