PDSCH Slot Geometry
A scheduled PDSCH transmission occupies a time-frequency rectangle: one or more allocated PRBs in frequency × a contiguous symbol range in time (mapping Type A: from symbol 0 or 2; Type B: starts mid-slot for mini-slot scheduling). Within this rectangle, REs are categorized as PDSCH data, DMRS pilots, PT-RS phase-tracking, CSI-RS, or rate-matched-around (PDCCH, SSB, reserved).
DMRS · The First Pilots
DMRS (Demodulation Reference Signals) are inserted at fixed symbol positions within the slot. Type 1 DMRS uses 6 SCs per port per symbol; Type 2 uses 4 SCs. Configurations support 1 to 12 ports. Single-symbol or double-symbol; "front-loaded" first DMRS at symbol l_0; additional DMRS for high-mobility UEs. The UE estimates the channel from DMRS and applies it to the data REs on the same antenna port.
MCS Tables & Modulation Order Qm
Three MCS tables (TS 38.214 Tables 5.1.3.1-1/2/3): Table 1 = up to 64-QAM (Qm=6), Table 2 = up to 256-QAM (Qm=8), Table 3 = up to 64-QAM with low-SE entries for URLLC reliability. The MCS index from DCI selects (Qm, code rate) jointly. 1024-QAM (Qm=10) added in Rel-17 for FR2-2 high-SNR.
TBS Calculation
Transport Block Size determines the actual byte count for the slot. NR uses an explicit TBS-determination procedure (TS 38.214 §5.1.3.2): N_RE = N_RE_per_PRB × N_PRB; intermediate N'_info = N_RE × code_rate × Qm × layers; quantized to a TBS table entry. Larger TBs are split into multiple LDPC code-blocks (max 8448 bits per CB for BG1).
LDPC Encoding Chain
Information bits → CRC attachment → code-block segmentation (if K > 8448 BG1 / 3840 BG2) → per-CB CRC → LDPC encoding with selected base graph and lifting size Z → rate matching to the available REs → bit interleaving → scrambling → modulation mapping → layer mapping → precoding → RE mapping.
HARQ Process
Up to 16 parallel HARQ processes per UE. Each PDSCH carries an HARQ Process ID, NDI (New Data Indicator), and RV (Redundancy Version) in DCI. The UE decodes; on failure it stores the soft bits and waits for retransmission with a new RV. ACK/NACK is reported on PUCCH (k1 slots later) or piggy-backed on PUSCH.
Mapping Type A vs Type B
Type A (slot-based): PDSCH starts at symbol 0 or symbol 2-3, lasts the rest of the slot. Type B (non-slot, mini-slot): start can be any symbol, length 2/4/7 symbols — used for URLLC sub-ms latency, and for sharing slot space with control or other UEs.
PRB Bundling
A PRG (Precoding Resource-block Group) is a set of contiguous PRBs over which the gNB applies the same precoder, allowing the UE to channel-estimate jointly across the bundle. PRG sizes 2, 4, "wideband"; signaled by RRC and triggerable by DCI.
Frequently Asked Questions
What is PDSCH in 5G NR?
PDSCH (Physical Downlink Shared Channel) is the primary downlink data channel in 5G NR. It carries user-plane data from the gNB to the UE, scheduled by DCI on PDCCH. PDSCH uses LDPC coding, supports modulation up to 1024-QAM (Rel-17), and is governed by TS 38.211 §7.3 and TS 38.214 §5.1.
What is the difference between PDSCH mapping Type A and Type B?
Type A (slot-based) starts at symbol 0 or 2-3 and lasts most of the slot — used for normal eMBB scheduling. Type B (mini-slot) can start at any symbol and lasts 2/4/7 symbols — used for URLLC sub-ms latency and tight scheduling.
How is the PDSCH transport block size (TBS) calculated?
The TBS is determined by N_PRB × N_RE_per_PRB × code_rate × Qm × N_layers, then quantized to a TBS table entry per TS 38.214 §5.1.3.2. The MCS index from DCI selects (Qm, code_rate); the gNB determines N_PRB and N_layers from the scheduler.
What is PRB bundling in PDSCH?
PRB bundling means the gNB uses the same precoder across a group of contiguous PRBs (a PRG). This lets the UE pool DMRS estimates across the bundle, improving channel estimation accuracy. Bundle sizes: 2, 4, or wideband.