Logical, Transport & Physical Channels in LTE 4G

Introduction

LTE's radio interface never describes a flow of data just once. The same bits are named three times as they descend the protocol stack: as a logical channel (what kind of information it is), then a transport channel (how the physical layer formats and carries it), then a physical channel (the actual coded, modulated signal on the air). Channel mapping is the fixed set of rules that connects those three views — defined across TS 36.321 (MAC), TS 36.300 (architecture) and TS 36.211/36.212 (physical layer).

You meet this mapping in every LTE call flow. When a UE reads system information, is paged, sends its first RRCConnectionRequest, or streams a video, that data is walking one of these logical→transport→physical chains. The MAC performs the first hop (logical→transport) and the PHY performs the second (transport→physical), so knowing the chains tells you exactly which layer owns each transformation and where to look when a flow misbehaves.

It matters because almost every Layer-2 discussion — scheduling, multiplexing, HARQ, broadcast, paging, MBMS — is really a statement about one of these channels. Get the two mapping hops straight and the rest of the LTE user-plane and control-plane data path falls into place; it is also the single most common LTE layer-2 interview question.

Why channel mapping is needed

Why three descriptions instead of one? Because each abstraction lets a different part of the stack forget detail it does not need. RRC and the MAC scheduler reason about logical channels — priorities, buffer occupancy, which radio bearer a packet belongs to — without knowing modulation order or code rate. The physical layer reasons about transport blocks, redundancy versions and HARQ processes without knowing whether the bits are a SIB, a paging record or a slice of video. The mapping rules are the thin, standardised contracts that let those two worlds evolve independently — and they are exactly why one shared physical channel can carry wildly different traffic: the distinction lives in the logical channel, not in the waveform.

Three views of the same traffic

Nothing about a parcel changes as it moves through the delivery service — only the vocabulary each sorting office uses to reason about it. LTE works the same way: one flow of bits, described by three channel types and transformed at two hops. This section pins down each of the three views on its own terms before the mapping tables connect them, because keeping "what the data is", "how it is carried" and "which signal carries it" separate is what makes every later table read cleanly.

Why three layers instead of one? Because each abstraction lets a different part of the stack forget detail it does not care about. RRC and the MAC scheduler reason about logical channels — priorities, buffer occupancy, which radio bearer a packet belongs to — without knowing anything about modulation order or code rate. The physical layer, in turn, reasons about transport blocks, redundancy versions and HARQ processes without knowing whether the bits inside are a SIB, a paging record or a slice of video. The mapping rules are the thin, standardised contracts that let those two worlds evolve independently: you can change how PDSCH is precoded without touching what a DTCH means, and you can add a new radio bearer type without redesigning the physical layer. This decoupling is also exactly why one shared physical channel can carry wildly different traffic — the distinction lives in the logical channel, not in the waveform.

Logical channels — what kind of information

Logical channels are named by the RRC and MAC layers according to the type of information they carry, and split cleanly into control channels (signalling) and traffic channels (user data). LTE also defines two dedicated MBMS logical channels for point-to-multipoint broadcast, which is the LTE-specific twist covered in its own section below. The naming is mnemonic once you see the pattern.

Two distinctions on this table earn their keep in real call flows. First, CCCH versus DCCH: CCCH exists precisely for the window before the UE owns a dedicated identity, so it is how the very first RRC handshake and the RACH Msg3 travel; the instant the UE is connected and holds a C-RNTI, all dedicated signalling moves to DCCH. Second, BCCH is a single logical channel that nonetheless splits at the transport layer — the tiny, ultra-robust MIB takes one path while the larger, scheduled SIBs take another. That split is the first thing the mapping tables below make explicit.

Transport channels — how the PHY carries it

A transport channel describes how, and with what characteristics, data is transferred over the air — shared versus whole-cell broadcast, with or without HARQ, with or without dynamic scheduling, and with what kind of transport block (TB). The MAC hands data to the PHY as one (or, with spatial multiplexing, two) transport blocks per transmission time interval (TTI = 1 ms subframe) on these channels. The characteristics below are what actually distinguish the transport channels from one another; memorising them is what turns rote mapping into understanding.

Look at what the columns reveal. Only the two shared channels — DL-SCH and UL-SCH — combine dynamic transport-block sizing, HARQ and dynamic scheduling. That trio is exactly what lets the scheduler resize a grant subframe-by-subframe, chase the fading channel with link adaptation, and recover errors cheaply with soft-combining retransmissions. Every high-value unicast payload therefore funnels through the shared channels, which is precisely why the MAC needs multiplexing (packing several logical channels into one TB) and HARQ in the first place. The BCH is the opposite extreme: one fixed, tiny, whole-cell block whose format the UE already knows before it decodes anything — no scheduling, no HARQ, because there is nobody to send a grant to and nobody to send an ACK back. The PCH and MCH sit in between: both reach the whole cell, but PCH is folded onto DL-SCH physical resources so that paging can piggyback on the shared machinery and cooperate with UE DRX, while MCH keeps a semi-static schedule so that many cells can transmit identical waveforms for MBSFN combining. The RACH is the odd one out entirely: it carries a preamble sequence, not a transport block, so HARQ and scheduling simply do not apply.

Physical channels — the signals on the air

A physical channel is the coded, rate-matched, modulated set of resource elements actually transmitted on the grid. Some physical channels carry transport-channel data; others carry only physical-layer control that has no transport channel above them. Keeping those two groups apart is the second big source of confusion after the mapping chains themselves.

Four physical channels — PDCCH, PCFICH, PHICH and PUCCH — have no transport channel above them at all. They exist to make the shared channels work: PDCCH tells the UE where and how to decode PDSCH or transmit on PUSCH; PCFICH tells the UE how large the control region is so it knows where data begins; PHICH acknowledges uplink transport blocks; and PUCCH carries the UE's feedback (ACK/NACK, scheduling requests, channel-quality reports) whenever it has no PUSCH to piggyback on. This physical-layer control is generated and consumed inside the PHY itself, which is why it never appears in the MAC's logical-to-transport mapping. Everything else in the table is a carrier for a transport channel — and those carriers are what the mapping chains connect.

The DL and UL mapping chains

Here is the whole point of the page: how each logical channel walks down to a physical channel. Read every row as logical → transport → physical. The downlink is richer because it must also carry broadcast, paging and MBMS; the uplink is lean.

Downlink mapping:

Uplink mapping:

Read across the downlink table and the pattern is stark: five of the seven logical rows converge on DL-SCH → PDSCH. Only the MIB (which needs a fixed, ultra-robust path), paging (which needs whole-cell reach and DRX) and MBMS (which needs MBSFN) branch off. The uplink is even simpler — every logical channel a UE can send maps to UL-SCH → PUSCH, and the only other uplink physical channels (PRACH, PUCCH) carry things that never were logical-channel data: a preamble and physical-layer feedback. The figure below draws both directions at once, including the MBMS branch.

The MBMS chain — the LTE-specific branch

Everything so far is standard unicast plumbing. The one branch unique to LTE’s channel model is MBMS (Multimedia Broadcast/Multicast Service) — the machinery for sending the same content to many UEs at once, used for things like mobile TV, live events and mission-critical group communication. It has its own logical, transport and physical channels, and they behave quite differently from the shared-channel world.

Because there is no feedback path, the MCH has no HARQ and only a semi-static schedule — the resource allocation is signalled through configuration rather than a per-subframe PDCCH grant, and reliability comes from robust, conservative coding plus the MBSFN combining gain instead of retransmissions. MCCH carries the control information that tells UEs which MBMS sessions are running and how they are scheduled; MTCH carries the actual content. Both ride MCH → PMCH in the special MBSFN subframes, entirely separate from the PDSCH that serves unicast traffic in ordinary subframes. This is the clearest example on the whole page of why the three-layer model exists: the same "traffic vs. control" logical distinction is reused, but the transport and physical layers underneath are completely different because the delivery requirements — one-to-many, no feedback, multi-cell combining — are completely different.

Content vs. format vs. carrier — the mental model

If you remember one framing from this page, make it this triad. Each layer answers a different question about the same bits, and every confusion below collapses once you keep the questions separate.

Content is what RRC and the scheduler care about — priority, radio bearer, buffer status. Format is the contract at the MAC–PHY boundary — transport-block size, whether HARQ applies, whether scheduling is dynamic. Carrier is the physical reality — a specific block of resource elements, coded and modulated. The reason a single logical channel like BCCH can take two different chains (the MIB path and the SIB path) is that identical content type can be assigned different formats for good engineering reasons: the MIB must be decodable with zero prior knowledge, so it gets the fixed, whole-cell BCH/PBCH format; the SIBs can afford the scheduled shared-channel format because by the time a UE reads them it already has the MIB. Conversely, many different content types (CCCH, DCCH, DTCH) share one format (DL-SCH/PDSCH) because the shared channel is flexible enough to carry all of them — the MAC just multiplexes them into the same transport block and lets the logical-channel identity, not the waveform, keep them straight.

Common confusions, cleared up

A handful of pairings trip people up constantly, so pin them down explicitly. PDCCH, PCFICH, PHICH and PUCCH have no transport channel above them — they carry DCI, CFI, HARQ indicators and UCI, which are physical-layer control, not MAC data, so they never appear in a logical-to-transport mapping. The MIB has its own dedicated path (BCCH → BCH → PBCH), while the SIBs take the ordinary shared-channel path (BCCH → DL-SCH → PDSCH) despite sharing the same logical channel. PCH is a distinct transport channel but is carried on PDSCH, not on any "paging physical channel" — there is no such thing. And RACH carries a preamble on PRACH with no transport block at all, so asking about its HARQ or scheduling is a category error. Keep the "content / format / carrier" triad in hand and each of these resolves itself.

Summary

LTE describes the same traffic at three levels, and channel mapping is the two-hop chain that connects them: the MAC maps logical channels (what the information is) to transport channels (how it is formatted — shared vs. whole-cell, HARQ or not, scheduled or not), and the PHY maps transport channels to physical channels (the coded, modulated signal). Name the owner of each hop and the mapping usually falls out on its own.

Two patterns do most of the work. First, the shared channels — DL-SCH→PDSCH and UL-SCH→PUSCH — are where almost all unicast content converges (SIBs, paging, CCCH/DCCH/DTCH), because only they combine dynamic transport-block sizing, HARQ and dynamic scheduling. Second, the exceptions each exist for a concrete reason: the MIB takes the fixed, whole-cell BCCH→BCH→PBCH path so it is decodable with zero prior knowledge; paging rides PCH for whole-cell reach and DRX; and MBMS uses the LTE-only MCCH/MTCH→MCH→PMCH MBSFN chain with no HARQ.

Finally, remember which physical channels have no transport channel above them — PDCCH, PCFICH, PHICH and PUCCH carry pure L1 control (DCI, CFI, HARQ indication, UCI) generated inside the PHY, so they never appear in a logical-to-transport mapping. With the shared-channel rule, the broadcast exceptions, and the L1-control set in hand, you can trace any LTE flow through all three layers in either direction.