LTE Physical Layer Processing

The LTE physical layer (PHY) is a highly complex and optimized signal processing chain designed to reliably transmit data over a challenging wireless channel. Its primary function is to transform data packets from the MAC layer, known as Transport Blocks (TBs), into a precisely structured radio signal for transmission, and perform the reverse operation for reception.

Downlink (DL):

• Transmission from eNodeB to UE.
• Uses OFDMA.
• Each user is allocated a set of orthogonal subcarriers.
• Advantage: flexible frequency-domain multiplexing, robust against multipath fading.

Uplink (UL):

• Transmission from UE to eNodeB.
• Uses Single SC-FDMA.
• Adds a DFT-spreading step to lower the PAPR, saving UE battery and improving PA efficiency.

Major Processing Steps:

• Channel Coding → ensures error correction capability.
• Modulation → maps coded bits to constellation symbols.
• Resource Allocation → assigns REs/RBs in time–frequency grid.
• IFFT/FFT → OFDM modulation and demodulation.
• RF Conversion → analog signal transmission and reception.

Input: Transport Blocks (TBs)

The input to the PHY processing chain is one or more Transport Blocks (TBs), delivered from the MAC layer every TTI, which is fixed at 1 ms (equivalent to one subframe).

A TB is the payload for a given user in a TTI. It typically contains data from the Packet Data Convergence Protocol (PDCP) and Radio Link Control (RLC) layers, which are encapsulated by the MAC layer.

Size Determination: The size of the TB is not arbitrary. It is determined by the eNodeB based on the channel quality reported by the UE. The eNodeB selects a Modulation and Coding Scheme (MCS) and a few Resource Blocks (RBs) to allocate. This combination points to a specific TBS in a standardized table (3GPP TS 36.213, Table 7.1.7.1-1). This ensures both the UE and eNodeB have an unambiguous understanding of the payload size. The maximum theoretical TBS can reach 75,376 bits in a 20 MHz channel with 2×2 MIMO, and even higher with LTE-Advanced features.

Downlink Physical Layer Processing Steps

The following steps detail the sequential processing of a Transport Block for downlink transmission on the PDSCH

CRC Attachment

A 24-bit Cyclic Redundancy Check (CRC) generated using a standard polynomial is calculated from the entire TB and appended to it. This provides a powerful error detection mechanism at the receiver. After decoding, the UE re-computes the CRC on the received data and compares it with the transmitted CRC. A mismatch indicates a decoding failure, triggering a Negative Acknowledgement (NACK) for a Hybrid ARQ (HARQ) retransmission.

• A Cyclic Redundancy Check (CRC) is appended to each TB.
• 24-bit CRC ensures the eNodeB/UE can verify correctness.
• For two transport blocks (MIMO with spatial multiplexing): each TB has its own CRC.
• Example: If TB = 30,000 bits → PHY appends 24 bits → 30,024 bits total.

Code Block Segmentation

The turbo encoder in LTE has a maximum input size of Kmax​=6144 bits. If a TB (plus its 24-bit CRC) exceeds this size, it must be segmented. The TB is divided into multiple Code Blocks (CBs). If segmentation occurs, an additional 24-bit CRC is calculated and appended to each code block. This allows the receiver to verify the integrity of each CB independently, which is crucial for incremental redundancy in HARQ retransmissions.

• Turbo encoder has a maximum input size of 6144 bits.
• If TB > 6144 bits → it is segmented into multiple Code Blocks (CBs).
• Each CB gets its own 24-bit CRC.
• Example: TB = 18,000 bits → segmented into 3 CBs (each ≤ 6144 bits).

Channel Coding (Turbo Coding)

The core of LTE’s forward error correction (FEC) is a Parallel Concatenated Convolutional Code (PCCC), commonly known as a Turbo Code. It consists of two identical Recursive Systematic Convolutional (RSC) encoders running in parallel, with an internal interleaver (the Quadratic Permutation Polynomial (QPP) interleaver) permuting the input bits to the second encoder. Rate 1/3: For each input bit (the systematic bit, dk​), the encoder produces two parity bits, one from each RSC encoder. This results in a native code rate of 1/3. The output is three streams: one systematic stream and two parity streams. Tail bits are also added to properly terminate the trellis in each encoder.

• Turbo Coding = powerful FEC (Forward Error Correction).
• Encoder structure:
• One systematic bit stream (original bits).
• Two parity bit streams from convolutional encoders with interleaving.
• Code rate:
• Nominal 1/3 → three output bits per input bit.
• Adjusted using rate matching to meet desired efficiency.

• Purpose: Correct transmission errors due to fading/noise.
• Example: Input 6144 bits → Turbo encoder outputs ~18,432 bits.

Rate Matching & Interleaving

Rate matching is the process that adapts the number of coded bits to perfectly match the physical capacity of the allocated radio resources. The three bitstreams (systematic and two parity) from the turbo encoder are written into a virtual circular buffer. Bits are then read out from this buffer to generate the final bit sequence. If the allocated physical resources are small relative to the number of coded bits, some parity bits are skipped (puncturing), increasing the effective code rate. If the resources are large, some bits are repeated (repetition), decreasing the effective code rate. This circular buffer is fundamental to HARQ. Different retransmissions (Redundancy Versions – RVs) are generated by simply starting the read-out process from different predefined starting points in the buffer. This ensures that each retransmission provides new, unique parity information to the receiver, a process called incremental redundancy.

• Goal: Match coded bit output to the number of available Resource Elements (REs).
• If more bits than REs → puncturing (dropping some bits).
• If fewer bits → repetition.
• Interleaving → distributes bits to combat burst errors (e.g., deep fades).

Code Block Concatenation

If the original TB was segmented, the rate-matched bits from each individual code block are concatenated sequentially to form a single codeword ready for transmission.

• After encoding, rate matching, and interleaving, all CBs are concatenated to form one stream again.
• This ensures that the PHY layer transmits a continuous sequence per TB.

Scrambling

The codeword is scrambled by performing a bitwise XOR operation with a pseudo-random sequence.

• XOR the bitstream with a cell-specific pseudo-random sequence.
• Prevents long runs of identical bits → improves spectrum characteristics.
• Also provides a basic UE separation.
• Example: UE A and UE B with same modulation would still see different scrambled sequences.

Modulation

The scrambled bitstream is mapped into complex-valued modulation symbols. The choice of modulation scheme is determined by the MCS.

• QPSK (Quadrature Phase Shift Keying): 2 bits per symbol. Robust, used in poor channel conditions.
• 16QAM (Quadrature Amplitude Modulation): 4 bits per symbol.
• 64QAM (Quadrature Amplitude Modulation): 6 bits per symbol.
• 256QAM (in LTE-Advanced Pro): 8 bits per symbol. Requires excellent channel conditions (high SNR).
• Scrambled bits → mapped onto constellation symbols.
• Supported schemes:
  • QPSK (2 bits/symbol, robust, low spectral efficiency).
  • 16QAM (4 bits/symbol).
  • 64QAM (6 bits/symbol).
  • 256QAM (8 bits/symbol, LTE-A Pro).
• Example: 12000 coded bits → 2000 QAM symbols (if 16QAM used).

Layer Mapping & Precoding (MIMO)

This stage is central to LTE’s Multiple-Input Multiple-Output (MIMO) functionality. Layer Mapping: The stream of complex symbols is demultiplexed into one or more parallel streams called layers. The number of layers is known as the transmission rank and is less than or equal to the number of transmit antennas. Precoding: Precoding is a form of transmit beamforming. The vector of layer symbols is multiplied by a precoding matrix W to produce the signals to be passed to the antenna ports. This operation maps the layers to the physical antenna ports. The matrix is chosen from a predefined codebook based on channel state information (CSI) fed back by the UE. The goal is to match the transmitted signal to the spatial characteristics of the radio channel, either to maximize signal power or to create spatially separated, orthogonal streams to improve data throughput.

• If MIMO is used:
  • Symbols mapped to layers (logical streams).
  • Precoder applies weights to map layers → physical antennas.
• Transmission modes:
  • TM2: Transmit diversity.
  • TM3/4: Spatial multiplexing (multiple layers).
  • TM9: up to 8 layers (LTE-A).
• Example: 2 layers, 2 antennas → spatial multiplexing doubles throughput.

Resource Element (RE) Mapping

The precoded complex symbols for each antenna port are mapped onto the time-frequency resource grid.

• A Resource Element (RE) = one subcarrier × one OFDM symbol.
• Resource Block (RB) = 12 subcarriers × 7 OFDM symbols (normal CP).
• Symbols mapped to REs according to allocation.
• Must also insert reference signals:
  • CRS (Cell-specific RS).
  • DMRS (UE-specific RS).
  • CSI-RS (for channel feedback).
• Control channels (PDCCH, PCFICH, PHICH) also occupy REs.

OFDMA Symbol Generation (IFFT)

This is the defining step of OFDMA. For each antenna port, the frequency-domain symbols mapped onto the resource grid are transformed into a time-domain signal using an Inverse Fast Fourier Transform (IFFT).
The symbols on the subcarriers act as the complex amplitudes for a set of orthogonal sinusoidal basis functions. The IFFT efficiently synthesizes the sum of all these sinusoids. For a 20 MHz channel, a 2048-point IFFT is used, where only 1200 central subcarriers are used for data, and the rest form guard bands.

Cyclic Prefix (CP) Insertion

To combat the effects of multipath propagation (Inter-Symbol Interference, ISI) and preserve subcarrier orthogonality, a Cyclic Prefix (CP) is added to the beginning of each time-domain OFDM symbol.

A copy of the tail end of the IFFT output waveform is prepended to the beginning.

Guard Interval: It provides a guard time against ISI. If the multipath delay spread is shorter than the CP duration, the energy from the previous symbol arrives during the CP of the current symbol and is discarded by the receiver, preventing interference.

Maintains Orthogonality: The CP makes the linear convolution of the channel appear as a circular convolution to the receiver’s FFT. This is critical because the FFT can only diagonalize a circulant channel matrix, thus preserving the orthogonality between subcarriers and allowing for simple single-tap frequency-domain equalization.

Types:

• Normal CP: Allows for 7 OFDM symbols per slot. Used in most cell environments.
• Extended CP: Allows for 6 OFDM symbols per slot. It is longer and used in environments with very large delay spreads, such as large single-frequency networks or high-speed scenarios.

Digital-to-Analog and RF Up-conversion

The final stage involves converting the digital time-domain baseband signal into an analog radio frequency (RF) signal.

• The parallel streams of digital samples (one for each antenna) are passed through Digital-to-Analog Converters (DACs).
• The resulting analog baseband signal is then mixed with a high-frequency carrier signal (up-conversion) to the desired licensed frequency band (e.g., 1.8 GHz).
• Finally, the RF signal is amplified by a Power Amplifier (PA) and transmitted through the antenna.

References:

• 3GPP TS 36.201 – LTE Physical Layer – General Description.
• 3GPP TS 36.211 – Physical Channels and Modulation.
• 3GPP TS 36.212 – Multiplexing and Channel Coding.
• 3GPP TS 36.213 – Physical Layer Procedures.
• 3GPP TS 36.300 – E-UTRA and E-UTRAN Overall Description.
• Dahlman, E., Parkvall, S., Sköld, J. – 4G: LTE/LTE-Advanced for Mobile Broadband, Academic Press, 2014.
• Sesia, S., Toufik, I., Baker, M. – LTE – The UMTS Long Term Evolution: From Theory to Practice, Wiley, 2011.
• Holma, H., Toskala, A. – LTE for UMTS: OFDMA and SC-FDMA Based Radio Access, Wiley, 2009.