Architecture of 3GPP LTE Transport Block Size (TBS) Determination
In the 3GPP Long Term Evolution (LTE) and LTE-Advanced specifications (3GPP TS 36.213, TS 36.211, and TS 36.306), peak physical layer throughput is derived not from an empirical approximation, but through deterministic Transport Block Size (TBS) matrices. Unlike 5G New Radio (NR) which calculates TBS using closed-form algebraic approximations of LDPC code rates and active subcarriers, 4G LTE standardized explicit, look-up table matrices to eliminate receiver rounding ambiguities.
1. The Mechanics of 3GPP TS 36.213 Table 7.1.7.2.1-1
Every 1 ms subframe (Transmission Time Interval, or TTI), the eNodeB MAC packet scheduler allocates radio resources based on reported User Equipment (UE) Channel Quality Indicators (CQI 1–15). This process follows a rigorous three-stage protocol chain:
- MCS Selection: The eNodeB selects a Modulation and Coding Scheme index ($I_{\text{MCS}} \in [0, 28]$). In Release 8, $I_{\text{MCS}}$ mapped to QPSK, 16-QAM, or 64-QAM. In Release 12, higher-order 256-QAM was introduced (Table 7.1.7.2.4-1), and Release 15 added localized 1024-QAM for indoor small cells.
- TBS Index Mapping: $I_{\text{MCS}}$ translates directly to a Transport Block Size index ($I_{\text{TBS}}$). For standard 64-QAM, $I_{\text{TBS}}$ spans 0 through 26. For 256-QAM, $I_{\text{TBS}}$ extends up to 33.
-
Matrix Cross-Referencing: The scheduler cross-references the assigned $I_{\text{TBS}}$ with the number of allocated Physical Resource Blocks ($N_{\text{PRB}} \in [1, 100]$) to extract the precise Transport Block Size (in bits) delivered to the physical layer:
Physical Downlink Shared Channel (PDSCH) Single-Layer RateR_{\text{SISO}} = \frac{TBS(I_{\text{TBS}}, N_{\text{PRB}})}{1\text{ ms}} = TBS(I_{\text{TBS}}, N_{\text{PRB}}) \times 10^{-3}\text{ [Mbps]}
For a standardized 20 MHz carrier ($N_{\text{PRB}} = 100$):
- 64-QAM (Release 8): At $I_{\text{TBS}} = 26$, the maximum single-layer Transport Block Size is exactly $TBS = 75,376\text{ bits}$. Over a 1 ms TTI, this delivers $75.376\text{ Mbps}$ per spatial layer.
- 256-QAM (Release 12): At $I_{\text{TBS}} = 33$, the maximum single-layer Transport Block Size rises to $TBS = 97,896\text{ bits}$, yielding $97.896\text{ Mbps}$ per spatial layer—a 29.88% throughput gain over 64-QAM.
- 1024-QAM (Release 15): In ultra-dense small cells, theoretical single-layer payload reaches approximately $122,370\text{ bits/ms}$ ($122.37\text{ Mbps}$).
2. MIMO Spatial Multiplexing and Code Word (CW) Architecture
To surpass the Shannon capacity of a single antenna, LTE employs Multiple-Input Multiple-Output (MIMO) spatial multiplexing. In the downlink, 3GPP standards support up to two independent Transport Blocks (termed Code Words, CW0 and CW1) per subframe, which are mapped across up to 8 spatial layers ($v \in [1, 8]$):
- Rank 1 (SISO): Single code word mapped to 1 layer ($v = 1$).
- Rank 2 (2×2 MIMO): CW0 maps to Layer 0, CW1 maps to Layer 1 ($v = 2$). Throughput doubles: $R_{\text{2x2}} = 2 \times R_{\text{SISO}}$.
- Rank 3 (3×3 MIMO): CW0 maps to Layer 0, CW1 maps across Layers 1 and 2 ($v = 3$).
- Rank 4 (4×4 MIMO): CW0 maps across Layers 0 and 1; CW1 maps across Layers 2 and 3 ($v = 4$). In decorrelated multipath environments, Rank 4 quadruples throughput: $R_{\text{4x4}} = 4 \times R_{\text{SISO}}$.
- Rank 8 (8×8 MIMO - LTE-A Pro): Defined in 3GPP Release 10 for up to 8 spatial layers ($v = 8$), scaling peak theoretical bit rates up to 8× single-layer capacity.
3. TDD Uplink-Downlink Subframe Framing & Special Subframe Mechanics
In Frequency Division Duplex (FDD), paired spectrum provides dedicated 100% time occupancy for downlink transmission. In Time Division Duplex (TDD), transmission occurs on unpaired spectrum, requiring the single RF channel to time-share between Downlink (D), Uplink (U), and Special (S) subframes within every 10 ms radio frame:
The Special subframe is divided into three distinct functional fields:
- Downlink Pilot Time Slot (DwPTS): Used for downlink synchronization, CRS transmission, and PDSCH data payload. Under Special Subframe Pattern 7 (SSP 7) with Normal CP, DwPTS spans 10 OFDM symbols, contributing an effective $\alpha_{\text{DwPTS}} \approx 0.60$ of a full downlink subframe.
- Guard Period (GP): Silent period providing round-trip propagation time buffer to prevent downlink-to-uplink cell interference.
- Uplink Pilot Time Slot (UpPTS): Used for Sounding Reference Signals (SRS) and PRACH preamble transmission.
In standard commercial TDD deployments worldwide (e.g., Band 40 and Band 41), operators universally configure TDD Subframe Configuration 2 ($\text{DSUDDDSUDD}$), which provides 6 full DL subframes and 2 Special subframes per 10 ms frame. With SSP 7:
Consequently, a 20 MHz 4×4 256-QAM carrier yielding $391.58\text{ Mbps}$ in FDD yields exactly $391.584 \times 0.720 = 281.94\text{ Mbps}$ under standard TDD Config 2.
4. Carrier Aggregation (CA) & UE Category Hardware Enclosures
To achieve Gigabit speeds, LTE-Advanced introduces Carrier Aggregation (CA), bonding up to five 20 MHz component carriers (CCs) into a unified 100 MHz pipe (Release 10 through Release 13):
However, real-world user throughput is strictly gated by the terminal's UE Category (3GPP TS 36.306). Regardless of the network cell's capabilities, the device's internal baseband processing architecture, soft buffer memory size, and RF receiver chains impose a rigid ceiling:
- UE Category 4 (Baseline 4G): Maximum physical downlink limit of $150.0\text{ Mbps}$ (2×2 MIMO, 64-QAM, 1 CC). Connecting to a 4×4 256-QAM cell ($391.6\text{ Mbps}$) will be strictly clamped to $150.0\text{ Mbps}$.
- UE Category 6 (LTE-A): Maximum limit of $300.0\text{ Mbps}$ (2CC CA, 64-QAM, 2×2).
- UE Category 11: Maximum limit of $600.0\text{ Mbps}$ (3CC CA, 256-QAM, 2×2).
- UE Category 16 (Gigabit LTE): Maximum limit of $1,050.0\text{ Mbps}$ ($1.05\text{ Gbps}$), requiring 4×4 MIMO, 256-QAM, and 3CC–4CC aggregation.
- UE Category 20 (LTE-A Pro Peak): Maximum ceiling of $2,000.0\text{ Mbps}$ ($2.0\text{ Gbps}$), combining up to 5CC CA with 4×4 / 8×8 MIMO and 256-QAM.