>
HomeLTE 4GFoundationsSpectrum & Bands
◍ FoundationsBeginner

Spectrum: E-UTRA Bands, FDD/TDD & Bandwidths in LTE 4G

E-UTRA operating bands, FDD vs TDD duplexing, and the 1.4–20 MHz channel bandwidths.

📚 3GPP-basedTS 36.101TS 36.104

Radio is delivered on assigned slices of spectrum, and how a network uses those slices is decided long before any call is set up. LTE splits the world into numbered operating bands, chooses either paired (FDD) or unpaired (TDD) duplexing, and runs a channel of a fixed width inside a band. Every carrier is then named by a single integer, the EARFCN. This page is grounded in TS 36.101 (UE radio transmission and reception), TS 36.104 (base-station radio transmission and reception) and TS 36.211 (physical channels and modulation).

Introduction

Spectrum is the raw material of a mobile network, and LTE imposes a strict grammar on it. A deployment is defined by three orthogonal choices: which operating band (a numbered block of frequencies with a fixed duplex mode), which duplexing scheme (FDD or TDD, a property of the band, not a free choice), and which channel bandwidth (one of six fixed widths from 1.4 to 20 MHz). Fix those three and a fourth value — the EARFCN integer — pins the exact centre frequency down to the 100 kHz raster.

These decisions happen at network planning time, long before a UE ever camps. They determine the coverage footprint (low bands reach far, high bands carry more), the uplink/downlink balance (rigid for FDD, tunable for TDD), and the peak throughput (a single carrier caps at 20 MHz / 100 RB, and Carrier Aggregation is how LTE-Advanced breaks that ceiling). A UE only participates once it has been told, through broadcast system information, which band, bandwidth and EARFCN the cell is using.

Why it matters: nearly every RF-planning, roaming and carrier-aggregation question comes back to these fundamentals. If you can read a band number and immediately know its duplex mode, rough frequency and likely role (coverage vs capacity), and translate an EARFCN into megahertz, the rest of LTE spectrum engineering is bookkeeping.

Why spectrum is organised this way

💡

In plain words: spectrum is like a crowded radio dial with strict city zoning. Regulators carve the dial into numbered lots (bands), each zoned for a duplex mode you cannot change. Inside a lot you may build a house of one of six standard widths (the channel bandwidth), and every house needs an exact street address so callers find it — that address is the EARFCN. Nobody gets to pick "any frequency I like"; you pick a lot, a house size and an address from the approved catalogue.

The concrete purpose of all this structure is global interoperability and interference control. If every operator picked arbitrary frequencies and duplex arrangements, a phone certified in one country could not roam to another, and adjacent networks would jam each other. By enumerating bands (TS 36.101), fixing the duplex mode per band, standardising six channel widths, and defining a 100 kHz raster of legal centre frequencies, 3GPP guarantees that a UE's RF front-end and a base station's transmitter agree exactly on where a carrier sits — so cell search, roaming and carrier aggregation all work from one shared catalogue.

What

A fixed catalogue: numbered operating bands, a per-band duplex mode (FDD/TDD), six channel bandwidths, and an integer EARFCN that maps one-to-one to a centre frequency.

Why

Enumerated, standardised spectrum is what makes global roaming, predictable UE hardware and interference-free adjacent channels possible.

How

The UE reads the band and bandwidth from broadcast system information, computes each candidate carrier's frequency from its EARFCN via per-band constants and the 100 kHz raster, and searches for sync signals in the centre 6 RB.

Duplexing — FDD vs TDD

Every mobile link has to carry traffic in two directions: downlink (DL, network to UE) and uplink (UL, UE to network). Duplexing is simply how those two directions are kept apart so they don't collide. LTE supports two schemes, and which one applies is a fixed property of the band — you cannot run a TDD band as FDD or vice versa.

What

FDD (Frequency Division Duplex) uses paired spectrum: one carrier for DL and a separate carrier for UL, offset by a fixed duplex spacing. Both directions transmit continuously and simultaneously.

Why

TDD (Time Division Duplex) uses a single unpaired carrier and lets DL and UL take turns in time. It needs only one block of spectrum and can be re-balanced toward DL-heavy traffic.

How

FDD needs a duplex filter to isolate the always-on UL and DL carriers. TDD needs tight network-wide time synchronisation and a guard period so a distant UE's uplink doesn't overrun the switch back to downlink.

The practical trade-off runs deep. FDD gives symmetric, low-latency, continuous links and is the workhorse of the low bands, but it burns two chunks of spectrum and wastes uplink capacity that most consumers never use. TDD fits into a single chunk, can bias the split heavily toward the downlink (which is where the vast majority of mobile traffic flows), and — because DL and UL share exactly the same frequency — enables channel reciprocity, which massive-MIMO beamforming relies on. The costs are a small added latency, the loss of spectrum to guard periods, and the operational burden of keeping every base station's clock aligned so that neighbouring cells switch between DL and UL at the same instant. If they drift, one cell's downlink blasts straight into another cell's uplink reception.

LTE also defines a niche third mode, HD-FDD (half-duplex FDD), which uses paired spectrum but forbids the UE from transmitting and receiving at the same instant. This drops the duplex filter requirement and simplifies cheap IoT devices, at the cost of throughput. The overwhelming majority of deployments are full-duplex FDD or TDD.

🎯

One-line test: two carriers separated in frequency → FDD. One carrier split in time → TDD. The band number tells you which, and it is not negotiable.

TDD UL/DL Configurations and the Special Subframe

Because a TDD carrier shares one frequency between DL and UL, the network must publish a rhythm that says which subframes go which way. An LTE radio frame is 10 ms long and holds ten 1 ms subframes numbered 0–9. TS 36.211 defines seven fixed patterns, the TDD UL/DL configurations numbered 0 through 6, and the cell signals its choice in subframeAssignment inside SIB1 (specifically the TDD-Config IE).

In every configuration, subframe 0 and subframe 5 are always downlink (they carry the synchronisation signals and broadcast information), and subframe 2 is always uplink. Subframe 1 is always a special subframe (S), the bridge that lets the cell turn around from DL to UL. The remaining subframes are what distinguish the configurations, ranging from UL-heavy config 0 to almost-entirely-DL config 5.

ConfigSF0SF1SF2SF3SF4SF5SF6SF7SF8SF9DL:ULSwitch period
0DSUUUDSUUU2:6 (UL heavy)5 ms
1DSUUDDSUUD4:4 (balanced)5 ms
2DSUDDDSUDD6:2 (DL heavy)5 ms
3DSUUUDDDDD6:310 ms
4DSUUDDDDDD7:210 ms
5DSUDDDDDDD8:1 (most DL)10 ms
6DSUUUDSUUD3:55 ms

The switch-point periodicity tells you how often the direction turns around. Configurations 0, 1, 2 and 6 switch every 5 ms (two special subframes per frame, at SF1 and SF6), which keeps uplink latency low. Configurations 3, 4 and 5 switch only once per frame (one special subframe at SF1), maximising downlink capacity at the expense of a longer wait for an uplink opportunity. Most commercial TDD deployments use configuration 1 (balanced) or configuration 2 (DL-heavy), because real mobile traffic is downlink-dominated.

The special subframe itself is not one thing — it is split into three fields. DwPTS (Downlink Pilot Time Slot) carries downlink data and control at the start; GP (Guard Period) is dead air in the middle that gives the UE time to switch its transceiver and absorbs the round-trip propagation delay, so it sets the maximum usable cell radius; and UpPTS (Uplink Pilot Time Slot) at the end carries sounding reference signals or a short PRACH. TS 36.211 defines ten special-subframe configurations (0–9) that trade DwPTS length against GP length — a longer guard period supports a larger cell but sacrifices downlink symbols.

🔑

Guard period sizing: the GP in the special subframe must exceed the round-trip propagation time to the farthest UE. A longer GP means a bigger cell but fewer usable symbols, which is why rural TDD macro cells and dense urban small cells often pick different special-subframe configurations.

E-UTRA Operating Bands

3GPP does not let operators pick arbitrary frequencies. TS 36.101 enumerates E-UTRA operating bands — numbered blocks of spectrum, each with a defined UL frequency range, DL frequency range and duplex mode. A UE is certified for a specific list of bands, and that list decides where in the world it can attach and roam.

The numbering scheme is worth internalising. Bands roughly 1 through 32 are predominantly FDD paired bands, while bands 33 onward (through the mid-40s) are the TDD unpaired bands. People name a band by its rough centre frequency ("Band 3 is 1800", "Band 7 is 2600") even though the exact edges are what the spec defines. A single physical frequency range can appear under several band numbers in different regions — the 700 MHz range alone spawned B12, B13, B17 and B28 for different regulatory plans.

Common FDD bands

BandNicknameUL (MHz)DL (MHz)Duplex spacingRegion / use
B121001920–19802110–2170190 MHzEurope/Asia core capacity
B21900 PCS1850–19101930–199080 MHzAmericas
B31800 DCS1710–17851805–188095 MHzMost-deployed LTE band globally
B4AWS-11710–17552110–2155400 MHzNorth America
B5850824–849869–89445 MHzAmericas coverage
B726002500–25702620–2690120 MHzUrban capacity
B8900 GSM880–915925–96045 MHzWide coverage (refarmed GSM)
B12700 lower SMH699–716729–74630 MHzUS coverage
B13700 upper C777–787746–756−31 MHz (reverse)US (Verizon)
B17700 b704–716734–74630 MHzUS (subset of B12)
B20800 (dig. dividend)832–862791–821−41 MHz (reverse)Europe coverage floor
B28700 APT703–748758–80355 MHzAPAC/Europe coverage

Notice a subtlety: bands B13 and B20 use a reverse duplex arrangement, where the uplink sits above the downlink in frequency — the opposite of the usual DL-above-UL layout. That is why their duplex spacing is written as negative. This kind of detail is exactly why the definitive source is always the TS 36.101 table, not memory.

Common TDD bands

BandNicknameFrequency (MHz)BandwidthRegion / use
B3420002010–202515 MHzChina/Asia
B3826002570–262050 MHzEurope capacity (centre gap of B7)
B3919001880–192040 MHzChina
B4023002300–2400100 MHzAsia/Europe capacity
B4125002496–2690194 MHzWide TDD capacity (US, China)
B4235003400–3600200 MHzC-band capacity
B4337003600–3800200 MHzC-band capacity
B48CBRS 3.53550–3700150 MHzUS shared/private (CBRS)
🔑

Coverage vs capacity: low bands like B20 (800) and B8 (900) travel far and penetrate walls but carry little bandwidth; high bands like B41 (2500) or B42 (3500) carry huge bandwidth but don't reach as far. Operators layer both — low band for the coverage floor, high band for capacity on top. B38 is a neat example of engineering economy: it is exactly the unpaired centre gap sitting between the B7 uplink and downlink.

Channel Bandwidths, Resource Blocks and Guard Bands

Inside a chosen band, the carrier occupies a channel bandwidth. LTE defines exactly six: 1.4, 3, 5, 10, 15 and 20 MHz. Each maps to a fixed number of resource blocks (N_RB) — the basic unit of scheduling in the frequency domain. A resource block is 12 subcarriers wide, and with LTE's fixed subcarrier spacing of 15 kHz, one RB spans 12 × 15 kHz = 180 kHz.

Channel BW (MHz)Resource blocks (N_RB)Occupied subcarriers (12 × RB)Transmission BW (MHz)Guard band (MHz)Occupancy
1.46721.080.3277%
3151802.70.3090%
5253004.50.5090%
10506009.01.090%
157590013.51.590%
20100120018.02.090%

Notice the gap between the nominal channel bandwidth and the actual occupied bandwidth. A 20 MHz channel carries 100 RB = 1200 subcarriers = 18 MHz of active signal; the remaining 2 MHz is guard band split across the two channel edges so adjacent channels don't interfere. For the five larger bandwidths the ratio is a clean 90% — this occupied bandwidth (the band that contains 99% of the transmitted power) is a defining LTE design choice. The 1.4 MHz channel is the exception: to keep the RB count practical it only reaches about 77% occupancy, spending proportionally more on guard band.

one RB → 12 subcarriers × 15 kHz = 180 kHz   |   20 MHz channel → 100 RB = 1200 subcarriers = 18 MHz occupied (90%)

The centre of an LTE carrier also has a special property: the middle 6 RB (72 subcarriers, exactly the 1.4 MHz worth) always host the PSS, SSS and PBCH, regardless of the channel's total width. That is what lets a UE find and read any LTE cell using only a narrow 1.08 MHz window before it even learns the full bandwidth from SIB2. The downlink channel bandwidth is broadcast as dl-Bandwidth in the MasterInformationBlock (MIB), enumerated in resource blocks (n6, n15, n25, n50, n75, n100).

🎯

Contrast with 5G NR: LTE fixes the subcarrier spacing at 15 kHz for every channel — one numerology, full stop. NR introduces flexible numerology (15, 30, 60, 120, 240 kHz), so an NR RB can be 180 kHz, 360 kHz, 720 kHz and so on, and NR channels reach 100 MHz (FR1) or 400 MHz (FR2) in a single carrier. LTE has one gear; NR has a gearbox.

EARFCN — Naming the Carrier Frequency

Once a band and bandwidth are chosen, the carrier still has to sit at a specific centre frequency, and both network and UE must agree on it exactly. Rather than exchange raw megahertz values, LTE uses an integer index: the EARFCN (E-UTRA Absolute Radio Frequency Channel Number), which runs from 0 to 65535 across all bands.

What

An EARFCN is an integer that maps one-to-one to a carrier centre frequency. There is a separate DL EARFCN (N_DL) and UL EARFCN (N_UL); each band owns a contiguous range of EARFCN values defined in TS 36.101.

Why

A single integer is compact to signal and unambiguous. Because each band owns a distinct EARFCN range, the number alone tells the UE the band, the duplex mode and the frequency raster.

How

The DL frequency is F_DL = F_DL_low + 0.1 × (N_DL − N_Offs-DL), where F_DL_low and N_Offs-DL are per-band constants from the spec. The UL frequency uses the matching UL constants.

F_DL = F_DL_low + 0.1 × ( N_DL − N_Offs-DL )   |   F_UL = F_UL_low + 0.1 × ( N_UL − N_Offs-UL )

The factor 0.1 is the 100 kHz channel raster — every valid LTE centre frequency lands on a multiple of 100 kHz. A worked example makes it concrete. For B1, the spec gives F_DL_low = 2110 MHz and N_Offs-DL = 0, so DL EARFCN 300 resolves to 2110 + 0.1 × (300 − 0) = 2140 MHz. The matching UL uses F_UL_low = 1920 MHz and N_Offs-UL = 18000, so UL EARFCN 18300 gives 1920 + 0.1 × (18300 − 18000) = 1950 MHz — exactly 190 MHz below the downlink, which is B1's duplex spacing.

For an FDD band there are always two EARFCNs in play — a DL EARFCN and a UL EARFCN — separated by the band's duplex spacing. For a TDD band a single EARFCN describes the one shared carrier used for both directions. When a UE performs cell search it scans candidate EARFCNs on the defined raster until it finds a cell's synchronisation signals in that central 6-RB window. In a handover the target carrier is signalled in RRCConnectionReconfiguration via the dl-CarrierFreq (EARFCN) inside the mobilityControlInfo, and neighbour carriers to measure are listed by EARFCN in the measurement objects of SIB5.

💡

Mental model: the EARFCN is a channel number, like a TV channel. "Channel 300" means one exact frequency; the UE and network both look it up in the same TS 36.101 table and land on the same MHz down to 100 kHz.

Picturing FDD vs TDD

The clearest way to internalise duplexing is to draw it. FDD spends two frequency blocks and runs both directions all the time; TDD spends one frequency block and slices it in time, using the special subframe as the turnaround.

FDD — paired spectrum frequency time DL carrier (continuous) UL carrier (continuous) duplex spacing TDD — unpaired spectrum frequency time DL S UL DL S UL one carrier, shared over time (S = special subframe) FDD: DL and UL happen at the same time on different frequencies TDD: DL and UL happen on the same frequency at different times low bands (B5 850, B8 900, B3 1800, B20 800) are usually FDD · high bands (B40 2300, B41 2500, B42 3500) often TDD the special subframe (DwPTS · GP · UpPTS) is the guard that lets TDD turn around
Figure 1. FDD splits directions across two paired carriers separated by the duplex spacing; TDD packs both directions onto one carrier and alternates them in time, using the special subframe as the DL-to-UL turnaround.

Carrier Aggregation and Bandwidth Classes

A single LTE carrier tops out at 20 MHz (100 RB), which caps the peak rate. LTE-Advanced breaks that ceiling with Carrier Aggregation (CA): the network bonds several carriers — called component carriers (CC) — so the UE effectively transmits and receives across their combined bandwidth. Five aggregated 20 MHz carriers give 100 MHz and 500 RB, roughly five times the throughput. One component carrier is the primary cell (PCell) that anchors the RRC connection; the others are added as secondary cells (SCell) purely for extra capacity.

Component carriers combine in three arrangements. Intra-band contiguous stacks adjacent blocks in one band. Intra-band non-contiguous uses two separated blocks within the same band. Inter-band combines carriers in different bands entirely, and can even mix FDD and TDD — for example pairing an FDD coverage layer with a TDD capacity layer. A specific combination is written like CA_3C (band 3, contiguous, two CCs), CA_1A-3A (one CC each on bands 1 and 3) or CA_3A-7A-20A (three-band inter-band). The trailing letter is the bandwidth class.

TS 36.101 defines the CA bandwidth classes by the aggregated number of resource blocks and the number of contiguously aggregated carriers:

CA bandwidth classAggregated RB (N_RB,agg)Max contiguous CCsRough aggregated BW
A≤ 1001up to 20 MHz
B≤ 1002up to 20 MHz (two narrow CCs)
C100 – 200220 – 40 MHz
D200 – 300340 – 60 MHz
E300 – 400460 – 80 MHz
F400 – 500580 – 100 MHz

So CA_3C means up to 200 RB of contiguous spectrum in band 3 — two 20 MHz carriers bonded into a 40 MHz pipe. The letter grows as more contiguous bandwidth is aggregated. Class B is a rare case (two carriers that together still stay within 100 RB). A UE's data sheet lists exactly which band combinations and classes it supports, because supporting a combination requires the right RF front-end and enough baseband to process every CC at once. This is why the band, duplex mode, bandwidth and EARFCN concepts on this page are the foundation: CA is simply the disciplined assembly of several such carriers into one logical connection.

🔀

LTE ↔ NR: NR keeps carrier aggregation but leans much less on it, because a single NR carrier is already wide (up to 100 MHz in FR1, 400 MHz in FR2) thanks to flexible numerology. Where LTE bonds five 20 MHz carriers to reach 100 MHz, one NR carrier does it alone. NR also adds bandwidth parts (BWP) — the UE operates in a configurable slice of the carrier rather than the whole thing — and reuses the EARFCN idea as the NR-ARFCN with a finer global raster (ΔFglobal of 5/15/60 kHz). EN-DC dual connectivity then aggregates an LTE PCell with an NR SCG across the two radios.

Q&A Quick Q&A

Q. How many resource blocks does a 10 MHz LTE carrier have, and how many subcarriers is that?

A. 50 RB. Each RB is 12 subcarriers, so 50 × 12 = 600 occupied subcarriers, giving about 9 MHz of transmission bandwidth inside the 10 MHz channel (the remaining 1 MHz is guard band, so 90% occupancy).

Q. How do you tell whether an LTE band is FDD or TDD from its number?

A. Roughly by range: the lower band numbers (about 132) are mostly paired FDD bands, while bands 33 onward through the mid-40s are unpaired TDD bands. For example B3 (1800) and B7 (2600) are FDD; B38, B40 and B41 are TDD. The definitive answer is always the duplex-mode column of the TS 36.101 band table.

Q. What does the EARFCN encode, and why two of them for FDD?

A. An EARFCN is an integer that maps directly to a carrier centre frequency via per-band constants (F_low, N_Offs) and the 100 kHz raster in TS 36.101. FDD needs a DL EARFCN and a UL EARFCN because it has two separate carriers offset by the duplex spacing; TDD needs only one because its single carrier is shared in time.

Q. In a TDD frame, which subframes are fixed and what is the special subframe for?

A. Subframes 0 and 5 are always downlink, subframe 2 is always uplink, and subframe 1 is always the special subframe. The special subframe carries DwPTS (downlink), a GP guard period for the transceiver turnaround and propagation delay, and UpPTS (uplink), acting as the DL-to-UL bridge. The GP length effectively limits the cell radius.

Q. Why does LTE fix the subcarrier spacing at 15 kHz when NR does not?

A. LTE targeted a single, predictable numerology, so one RB is always 180 kHz and one carrier caps at 20 MHz / 100 RB. NR added flexible numerology (15/30/60/120 kHz) to serve wider bandwidths, higher frequencies and mixed latency needs in one framework. LTE reaches beyond 20 MHz only through Carrier Aggregation, not wider single carriers.

Q. What does the CA notation CA_3C tell you?

A. Band 3, bandwidth class C. Class C means two contiguously aggregated component carriers totalling 100–200 RB, so up to about 40 MHz of contiguous spectrum in band 3.

Summary

LTE spectrum rests on four locked-together choices. The operating band (TS 36.101) fixes the frequency range and the duplex mode — FDD for paired low bands (continuous DL and UL split in frequency), TDD for unpaired high bands (one carrier shared in time, with a special subframe of DwPTS/GP/UpPTS as the turnaround and seven UL/DL configurations to tune the split). Inside a band the channel bandwidth is one of six fixed widths, each mapping to a fixed N_RB at the constant 180 kHz RB size, with roughly 90% occupancy and the rest as guard band; the centre 6 RB always carry PSS/SSS/PBCH for cell search.

The EARFCN then names the exact centre frequency as a single integer on the 100 kHz raster, with a DL/UL pair for FDD and one value for TDD. Carrier Aggregation bonds up to five 20 MHz component carriers (a PCell plus SCells) to break the single-carrier ceiling, classified by bandwidth class A–F. Read a band number for its duplex mode and coverage-vs-capacity role, translate an EARFCN into MHz, and you can reason about almost any LTE spectrum, roaming or CA question — and see clearly how NR generalises every one of these ideas with flexible numerology, wider carriers and bandwidth parts.

Where spectrum connects

Bands, duplexing and bandwidths define the canvas; the next topics show what LTE paints on it — how the RBs are organised in time and frequency, how the carrier repeats in frames, and how several carriers combine for more throughput.

Resource Grid — how RBs and subcarriers form the time/frequency mapFrame Structure — frames, subframes, slots and the TDD special subframeCarrier Aggregation — bonding component carriers for more bandwidth