Physical Layer Overview in LTE 4G

Introduction

The physical layer (PHY, or Layer 1) is where LTE stops being a protocol and becomes electromagnetic energy in the air. Everything above it — RRC state machines, MAC scheduling, RLC segmentation — ultimately produces a transport block of bits that the PHY must carry across the radio channel and reconstruct at the far end. The PHY is the only layer that deals with turbo codes, QAM constellations, OFDM symbols and antenna ports; everything else reasons in bits and blocks.

It runs constantly, in both directions, for the entire life of a connection. From the very first moment a UE powers on it is using the PHY — correlating against the PSS/SSS to find a cell, decoding the PBCH to read the MIB, then decoding PDCCH grants and PDSCH data once connected. Every millisecond a scheduled UE has, the PHY encodes or decodes at least one transport block and reports back HARQ acknowledgements and channel measurements. It is the busiest and most latency-critical layer in the stack.

Understanding the PHY matters because almost every "why is throughput low / why won't it attach / why is coverage poor" question eventually lands here: on modulation order, code rate, reference-signal overhead, or how many resource elements actually carried the transport block. Knowing the fixed shape of the LTE PHY — one numerology, one transmit chain, a handful of transport and physical channels — lets you reason about all of that from first principles.

Why the physical layer exists

Concretely, the PHY exists to solve one hard problem the upper layers must not be burdened with: moving bits reliably across a hostile, time-varying radio channel. That means three jobs. First, protect the bits — add a CRC so errors are detectable and channel coding (turbo for data) so many of them are correctable, then let HARQ retransmit what survives neither. Second, fit the bits to the channel — choose a modulation and code rate matched to the current signal quality, and lay the coded symbols onto exactly the time-frequency resources the scheduler granted. Third, make the signal findable and measurable — transmit synchronisation and reference signals so the receiver can lock timing, estimate the channel, and report quality back up.

What the Physical Layer Delivers

Think of the physical layer (PHY, or Layer 1) as a service provider with exactly one customer above it: the MAC. The PHY offers the MAC a handful of transport channels, and everything the PHY actually does — coding, rate matching, modulation, mapping onto the air — is done to serve data flowing over those transport channels. The MAC never sees an OFDM symbol or a resource element; it just hands down a transport block once per Transmission Time Interval (TTI, 1 ms in LTE) and trusts the PHY to deliver it.

The service the PHY offers is described entirely in terms of transport channels, each with its own characteristics — whether it is shared or dedicated, broadcast to all or addressed to one, and what error-protection and multiplexing options it supports. In the downlink the workhorse is the Downlink Shared Channel (DL-SCH), which carries almost all user data and most higher-layer control. Alongside it sit the Broadcast Channel (BCH) for the fixed Master Information Block, the Paging Channel (PCH) for reaching idle UEs, and the Multicast Channel (MCH) for MBMS content delivered over a single-frequency network. In the uplink there are just two: the Uplink Shared Channel (UL-SCH) for user data and control, and the Random Access Channel (RACH), the contention channel a UE uses to first get the network's attention.

The 36.2xx Spec Map

The LTE physical layer is described across a small, well-organised family of specifications. Knowing which document holds what saves enormous time — an exam or interview question about "where is turbo coding defined?" is really a question about the 36.2xx map. The five documents divide the problem cleanly: one overview, one for what is transmitted, one for how bits are protected, one for how the PHY behaves over time, and one for what it measures.

Two Multiple-Access Schemes: OFDMA and SC-FDMA

LTE deliberately uses a different multiple-access scheme in each direction. The downlink uses OFDMA (Orthogonal Frequency Division Multiple Access): the wideband channel is split into many narrow 15 kHz subcarriers, and different UEs are served on different groups of subcarriers and symbols. OFDMA is robust against multipath — a long cyclic prefix absorbs delay spread so each subcarrier sees a flat channel — and it is easy to schedule in two dimensions, time and frequency, which is exactly what a base station with plenty of power wants.

The uplink instead uses SC-FDMA (Single-Carrier FDMA), also called DFT-spread OFDM. It is almost the same as OFDMA, with one extra step inserted before the OFDM modulator: a DFT (Discrete Fourier Transform) "spreads" the data symbols across the assigned subcarriers first. This small change makes the transmitted waveform behave much more like a single carrier, which dramatically lowers its Peak-to-Average Power Ratio (PAPR).

Both schemes share the same 15 kHz subcarrier spacing, the same slot/subframe timing, and the same resource grid, so a UE's receiver and transmitter reuse most of the same OFDM machinery. The DFT-spread stage is the only structural difference — which is why SC-FDMA is often just described as "OFDM with a DFT precoder in front."

Physical Channels, Physical Signals, and the Mapping

Downward, the PHY does not expose transport channels — it exposes physical channels and physical signals, and the difference matters. A physical channel carries information that originated above the PHY: a transport channel's bits, or Layer-1 control. A physical signal carries nothing from higher layers at all — it exists purely so the receiver can do its job: synchronise, estimate the channel, or measure signal quality.

The key physical signals in LTE are the Cell-specific Reference Signal (CRS), scattered across the whole downlink band so any UE can estimate the channel and measure RSRP; the Primary and Secondary Synchronisation Signals (PSS/SSS) that a UE finds first to acquire timing and the physical cell ID; and, on the uplink, the Demodulation Reference Signal (DM-RS) and Sounding Reference Signal (SRS). LTE is fundamentally a CRS-based system: the always-on, cell-wide reference signal is the backbone of downlink demodulation and measurement. The physical cell identity itself is PCI = 3 × N_ID⁽¹⁾ + N_ID⁽²⁾, where the SSS carries the cell-group N_ID⁽¹⁾ (0–167) and the PSS the sector N_ID⁽²⁾ (0–2), giving 504 distinct PCIs — and it is the PCI that seeds the CRS frequency shift and scrambling.

Each transport channel maps to one or more physical channels. This transport-to-physical mapping is the seam between the two worlds — the MAC's DL-SCH becomes bits on the PDSCH, and so on down the list.

Two physical channels have no transport channel above them because they carry Layer-1 control created inside the PHY itself: the PDCCH, which carries Downlink Control Information (DCI) — the scheduling grants that tell a UE where its PDSCH is — and the PUCCH, which carries Uplink Control Information (HARQ ACK/NACK, CQI, scheduling requests) when the UE has no PUSCH to piggyback on. A PDCCH is addressed by scrambling its CRC with a 16-bit RNTI (e.g. C-RNTI, SI-RNTI, P-RNTI, RA-RNTI), so a UE decodes only the DCI meant for it. Two further control channels round out the downlink control region: the PCFICH, which tells the UE how many symbols (1–3, or up to 4 on the narrowest bandwidths) the control region occupies, and the PHICH, which carries the uplink HARQ ACK/NACK.

The Downlink Transmit Chain, Step by Step

The best way to understand the PHY is to follow one downlink transport block from the MAC all the way to the antenna. Each stage has a single job, and the stages are pipelined in a fixed order defined across 36.212 (coding) and 36.211 (physical mapping).

Walking the pipeline: CRC attachment appends a 24-bit checksum (the gCRC24A polynomial) to the whole transport block so the receiver can detect errors and drive HARQ. Code-block segmentation then splits a large transport block into smaller code blocks — the turbo coder has a maximum input size of 6144 bits, so anything bigger is cut into pieces, each getting its own 24-bit CRC (gCRC24B) so the receiver can tell which code blocks failed; filler bits pad the first block up to a valid turbo interleaver size (one of 188 defined block lengths, 40–6144 bits). Turbo coding adds redundancy: a rate-1/3 turbo code — two 8-state constituent convolutional encoders separated by a Quadratic-Permutation-Polynomial (QPP) internal interleaver, decoded iteratively — is the LTE workhorse, producing three output streams (one systematic, two parity), while a lighter tail-biting convolutional code (constraint length 7, rate 1/3) protects small control payloads like the PBCH, PCFICH and PDCCH. Rate matching punctures or repeats the coded bits to fit the number of resource elements the scheduler allocated, using a circular buffer, and it is here that the HARQ redundancy version (RV 0–3) is selected, so each retransmission can send a different starting point in that buffer.

From there the chain is physical. Scrambling multiplies the bits by a cell-and-UE-specific Gold sequence (initialised from the C-RNTI, cell ID and slot) so neighbouring cells look like mutual noise, spreading interference rather than letting it correlate. Modulation maps groups of bits onto QPSK (2 bits/symbol), 16QAM (4), 64QAM (6) or, from Rel-12, 256QAM (8) constellation symbols — higher order packs more bits per symbol but needs a cleaner channel (higher SINR). Layer mapping distributes the modulation symbols across one or more of up to 8 MIMO layers (up to two codewords), and precoding combines those layers onto the physical antenna ports — together they are how spatial multiplexing and transmit diversity happen. Resource-element mapping then places each symbol on a specific subcarrier and OFDM symbol in the resource grid, carefully stepping around the REs reserved for the CRS and other signals. Finally the OFDM modulator (an IFFT of up to 2048 points plus a cyclic prefix) turns the filled grid into the time-domain waveform sent from the antenna.

The Uplink Chain and the 1 ms TTI

The uplink chain is nearly identical to the downlink — CRC, code-block segmentation, turbo coding, rate matching, scrambling, and modulation all appear in the same order — with one structural twist and a few relabellings. The twist is the DFT-spread (transform-precoding) stage: after modulation, the symbols destined for the PUSCH pass through a DFT that spreads them across the assigned subcarriers before they reach the OFDM modulator. That single inserted transform is what turns OFDMA into the low-PAPR SC-FDMA the UE needs, as described above. The data itself is the UL-SCH riding on the PUSCH, with uplink control multiplexed in or carried separately on the PUCCH.

Timing is where LTE is deliberately rigid. The subcarrier spacing is fixed at 15 kHz, which cascades into everything: an OFDM symbol lasts about 66.7 µs (1 / 15000 s) plus its cyclic prefix, a resource block is 12 subcarriers (180 kHz) by one slot, one slot is 0.5 ms, two slots make a 1 ms subframe, and ten subframes make a 10 ms radio frame. That 1 ms subframe is the TTI — the fundamental scheduling and transmission interval. Every transport block corresponds to one TTI, HARQ timing is counted in TTIs (FDD downlink uses a fixed 8-TTI HARQ round-trip), and the whole transmit chain above runs once per TTI per transport block. Because there is only one numerology, the frame timing and resource grid are identical for every LTE deployment: simpler to build, at the cost of the flexibility NR later added.

How LTE PHY Differs from NR

If you already know 5G NR, the fastest way to fix LTE in memory is by contrast. NR kept the same overall architecture — transport channels above, physical channels below, an OFDM-based transmit chain — but changed three things that ripple through the whole PHY.

The coding change is the one interviewers love: LTE uses turbo coding for the shared data channels, while NR replaced it with LDPC (Low-Density Parity-Check) codes, which decode more efficiently at the very high throughputs and large block sizes NR targets, and polar codes for control. The reference-signal change is just as consequential: LTE is CRS-based, transmitting a cell-wide reference signal continuously whether or not there is traffic, whereas NR is deliberately "lean," relying on UE-specific DM-RS that is only sent when data is scheduled — better energy efficiency and less always-on interference. And where LTE fixes one numerology, NR offers a menu so it can serve both wide mmWave carriers and ultra-low-latency traffic. LTE's single 15 kHz choice was a good fit for the sub-6 GHz bands and delay spreads it targeted; NR simply needed more range.

Summary

The LTE physical layer is a single, fixed machine with one customer. Upward it offers the MAC six transport channels — DL-SCH, BCH, PCH, MCH in the downlink and UL-SCH, RACH in the uplink — and takes a transport block once per 1 ms TTI. Downward it exposes physical channels (PDSCH, PDCCH, PBCH, PUSCH, PUCCH, PRACH…) and physical signals (CRS, PSS/SSS, DM-RS, SRS), and everything in between — CRC, turbo coding, rate matching, scrambling, modulation, layer mapping, precoding, RE mapping, OFDM — exists to move that block reliably across the air.

Two design choices define the LTE PHY: OFDMA down, low-PAPR SC-FDMA up (protecting the UE's power amplifier), and a single 15 kHz numerology that fixes the whole frame and grid so every deployment shares one shape. The 36.2xx family documents it — 36.201 the map, 36.211 channels, 36.212 coding, 36.213 procedures, 36.214 measurements.

If you remember the transport-to-physical mapping, the order of the transmit chain, and the three things NR later made configurable (numerology, coding, reference signals), you can reconstruct almost any LTE PHY answer from first principles.

Quick Q&A