Bandwidth Part (BWP) in 5G NR

Introduction to Bandwidth Part (BWP)

With the evolution from LTE to 5G NR, wireless networks have gained the ability to support extremely wide channel bandwidths—sometimes up to 100 MHz in sub-6 GHz bands (FR1) and up to 400 MHz in mmWave bands (FR2). While this flexibility unlocks unprecedented peak data rates, it also creates new challenges for both network operators and UE. Not every UE is capable, or needs, to always process the entire bandwidth.

This is where the concept of the Bandwidth Part (BWP) comes in. In 5G NR, a BWP is a mechanism that allows the network to split the overall carrier bandwidth into smaller, more manageable frequency segments. Each segment can be independently configured and assigned to the UE, enabling the network to optimize resource allocation, power consumption, and device complexity based on the requirements of different services.

Why Was BWP Introduced in 5G?

The introduction of BWPs is a direct response to the unique needs and challenges of 5G:

• Wider Bandwidths: 5G NR supports much larger channel bandwidths than LTE, but low-end UEs (such as IoT devices) don’t need or can’t handle these wide channels.

• Device Diversity: 5G must serve both high-end smartphones and low-power sensors on the same network.

• Service Flexibility: 5G offers eMBB, URLLC, and mMTC—each with different bandwidth and power requirements.

By configuring and switching between BWPs, the network can ensure:

• High-end devices use wide BWPs for peak throughput.

• Low-power devices use narrow BWPs to conserve energy.

• The same UE can switch between BWPs dynamically as its service or application changes.

Example:When a smartphone is idle or receiving small control messages, it can operate on a narrow BWP to save power. When streaming a 4K video, it can switch to a wide BWP for maximum throughput.

What is BWP (Bandwidth Part)

According to the specification TS 38.211, a Bandwidth Part is a “subset of contiguous common resource blocks (CRBs) for a given numerology (subcarrier spacing), defined within the carrier bandwidth.”

• Each BWP has a defined start (the CRB index where it begins) and a size (number of CRBs).

• BWPs can overlap in frequency, or be separate, and each can have different subcarrier spacings (numerologies).

Channel Bandwidth vs. Bandwidth Part

• Channel Bandwidth: This is the total frequency width allocated by the network for a 5G NR carrier. For example, the operator may deploy a 100 MHz wide NR carrier.

• Bandwidth Part (BWP): Each BWP is a subset of the total channel bandwidth. It is defined by its start position and size (both in terms of resource blocks). The BWP may span a much smaller portion than the entire channel.

Reference: A Primer on Bandwidth Parts in 5G New Radio, Xingqin Lin, Dongsheng Yu, Henning Wiemann

Points to be considered

• 5G NR divides the available spectrum into operating bands, categorized as FR1 (sub-6 GHz) and FR2 (mmWave).

• Each mobile operator is allocated a specific portion of the operating band for their use.

• Within the operator’s spectrum, a carrier operates over a defined channel bandwidth.

• A reference point (Point A) is established for all resource block numbering and bandwidth definitions.

• The cell-specific channel bandwidth is configured with an offset from Point A, determining the bandwidth available to each cell.

• Each UE can be assigned its own channel bandwidth, which may also be offset within the cell-specific allocation.

• Bandwidth Part (BWP) is a subset of the UE-specific bandwidth, defined by its starting resource block and size.

• The minimum size of a BWP must be at least as large as the SSB (Synchronization Signal Block) bandwidth, to ensure initial access and broadcast are always possible.

• The maximum BWP size is limited by the configured carrier bandwidth and the UE’s capability.

• The SSB defines the minimum BWP requirement—every BWP must fully contain the SSB region so the UE can receive system information.

• A UE can be configured with up to four BWPs per direction (downlink/uplink) on a given carrier, but only one BWP can be active at a time in each direction.

• In the downlink, the bandwidth of each BWP should be equal to or greater than the SSB bandwidth, but it may or may not actually contain the SSB.

• For downlink, the UE is not expected to receive PDSCH, PDCCH, CSI-RS, or TRS outside the active BWP.

• Each downlink BWP includes at least one CORESET with UE-Specific Search Space (USS).

• In the primary carrier, at least one of the configured downlink BWPs must include a CORESET with Common Search Space (CSS).

• For uplink, the UE shall not transmit PUSCH or PUCCH outside the active BWP.

• If a UE is configured with a supplementary uplink, it can be configured with up to four BWPs in the supplementary uplink, but only one BWP can be active at a time.

• At any given moment, only one BWP per direction (UL/DL) is active.

• The network controls which BWP is active via signalling or configuration.

• Switching between BWPs can be triggered by control messages, timer expiration, or UE activity.

CRB and PRB relation

CRBs provide a common reference grid for the entire channel, while PRBs are UE/BWP-specific slices within that grid—linked by a simple offset.

• Common Resource Blocks (CRBs) are numbered from CRB0 (anchored at “Point A”) across the entire channel bandwidth for a given subcarrier spacing configuration.

• Physical Resource Blocks (PRBs) are defined within a specific Bandwidth Part (BWP) and are numbered from PRB0 up to the size of the BWP minus one.

• The mapping between a PRB inside a BWP and its location in the overall channel (as a CRB) is:

CRB index = PRB index + BWP start (N_{start_BWP,μ})

• This means PRB0 of a BWP always aligns with the starting CRB of that BWP, and each subsequent PRB aligns sequentially with the next CRB.

• This mapping enables the network to flexibly define which section of the full carrier bandwidth a UE will use, supporting spectrum efficiency, device capability adaptation, and dynamic resource allocation.

Types of BWPs and RRC Parameters for configurating BWP in 5G NR

• initialDownlinkBWP: DL BWP#0 configuration for initial access (ID=0)

• firstActiveDownlinkBWP-Id: 1: DL BWP#1 becomes active immediately after attach or reconfiguration

• defaultDownlinkBWP-Id: 0: DL BWP#0 is the default/fallback after inactivity

• bwp-InactivityTimer: ms40: Sets fallback timer to 40 ms

• downlinkBWP-ToAddModList: List of all DL BWPs configured for this cell/UE

• initialUplinkBWP: UL BWP#0 configuration for initial UL

• firstActiveUplinkBWP-Id: 1: UL BWP#1 is active after attach/reconfiguration

• uplinkBWP-ToAddModList: List of all UL BWPs

• Initial BWP: The bandwidth part configured for the UE during cell access or initial connection. It is typically used for system information, paging, and idle/inactive operation. The Initial BWP is always included in the RRC configuration as the default setup for the UE.

• FirstActiveBWP: The bandwidth part that becomes active for the UE immediately after the RRC connection setup or following a specific reconfiguration. This is where the UE starts its active communication on the carrier.

• Default BWP: The bandwidth part to which the UE reverts when no specific service or traffic condition requires another BWP. The Default BWP acts as a fallback or “baseline” operating mode for ongoing network activity.

• Regular BWP: Refers to any bandwidth part (other than Initial, FirstActive, or Default) that is configured for the UE. Regular BWPs are used to support specific services, higher throughput, QoS adaptations, or to optimize network resource allocation. Up to four regular BWPs can be configured in each direction (uplink and downlink).

• Configured BWPs: This is a collective term for all BWPs (Initial, FirstActive, Default, and Regular) that are set up for the UE in RRC signalling.

BWP Switching Mechanisms

A key innovation of BWPs is the ability to switch between them, allowing the network and the UE to optimize performance and energy consumption dynamically.

The selection (or switching) of the active BWP is controlled by several mechanisms as per 3GPP TS 38.321 Section 5.15:

How is BWP Switching Triggered?

There are ways to switch between active BWPs:

Key BWP Switching Mechanisms:

1. PDCCH (via DCI): The network may activate a specific BWP for the UE using the bandwidth part indicator field in DCI Format 0_1 (UL grant) or 1_1 (DL scheduling).

2. bwp-InactivityTimer: The network sets a timer (bwp-InactivityTimer); if the UE is inactive on the current BWP and the timer expires, the UE switches back to the configured default BWP.

3. RRC Signalling: The network can command a BWP switch at any time via an RRC Reconfiguration message.

4. MAC Trigger (Random Access): The MAC layer may switch BWPs during procedures such as random access, typically reverting to the initial BWP.

Reference: A Primer on Bandwidth Parts in 5G New Radio, Xingqin Lin, Dongsheng Yu, Henning Wiemann

The above diagram illustrates the dynamic switching of Bandwidth Parts (BWPs) in 5G NR, where only one downlink and one uplink BWP can be active at any moment.Switching between initial, active, and default BWPs is triggered by events such as RRC configuration, DCI signalling, inactivity timers, or random-access procedures.

Idle Mode:

• The UE first synchronizes using SSB and acquires MIB/SIB1.

• The initial BWP (DL/UL BWP#0) is configured and used for random access and early signalling (e.g., paging).

• SSB and CORESET#0 are mapped to the smallest region required (20/24 RBs).

UE Enters Connected Mode:

• After random access and RRC setup, the first active BWPs (DL BWP#1 / UL BWP#1) are switched on.

• These BWPs have much wider resource allocations (e.g., 270 RBs), supporting high-throughput data sessions.

• BWP switching at this point is typically triggered by RRC signalling (firstActiveBWP).

Data Transfer Phase:

• Only one BWP (DL and UL) is active for actual data transfer.

• The network may switch to another BWP (e.g., for QoS, energy savings, or traffic demand) by sending a DCI with the bandwidth part indicator.

BWP Inactivity or Timer Expiry:

• If no traffic is detected for the configured inactivity timer period (e.g., bwp-InactivityTimer), the UE automatically switches to the default BWP (DL BWP#2), which typically uses fewer resources (e.g., 52 RBs).

• This switch can also occur because of other inactivity or idle triggers.

Repeat as Needed:

• BWP switching can occur multiple times during a session, based on DCI commands, RRC messages, inactivity, or MAC triggers.

The Connection Between CORESET and BWP in 5G NR

CORESET is a fundamental concept in 5G NR, defining a configurable region in the time-frequency grid where the Physical Downlink Control Channel (PDCCH) is transmitted. Essentially, a CORESET specifies which resource blocks and OFDM symbols are used to carry control information such as scheduling grants, uplink/downlink assignments, and other critical control signalling for the UE.

• A CORESET consists of a set of resource blocks (in frequency) and OFDM symbols (in time), allowing flexible placement and sizing according to deployment needs.

• Multiple CORESETs can be configured per cell, each with its own parameters (frequency location, duration, mapping, and interleaving).

• The PDCCH is only transmitted within a CORESET; therefore, UEs monitor PDCCH candidates only in the CORESET(s) relevant to them.

Relevance to BWP:

• Each Bandwidth Part (BWP) is associated with at least one CORESET in the downlink.

  • In fact, a BWP must be configured with at least one UE-specific CORESET (with UE-Specific Search Space), so the UE knows where to monitor PDCCH within that BWP.

• In the Primary Downlink BWP, there must be at least one CORESET with a Common Search Space (CSS), used for broadcasted control messages such as paging or system information.

• When a BWP is activated, the UE only needs to search for control information within the CORESET(s) defined for that BWP—this minimizes power consumption and complexity, especially in wide-band 5G channels.

• CORESET configuration within a BWP allows network operators to:

  • Tailor control signalling to device or service needs,

  • Enable device-specific scheduling,

  • Support efficient resource allocation for diverse services (eMBB, URLLC, mMTC).

CORESET acts as the “control region” for a BWP. It ensures that the UE only searches for and decodes downlink control information within the part of the spectrum defined by its current active BWP, supporting both flexible control signalling and efficient UE operation in 5G NR.

Reference

• 3GPP TS 38.211, Section 7.3.2: Defines CORESET and its association with BWP in 5G NR.

• 3GPP TS 38.213, Section 12: Describes BWP operation and CORESET mapping.

• 3GPP TS 38.331, Section 6.3.2: Covers RRC signalling for BWP and CORESET configuration.

• Dahlman et al., “5G NR: The Next Generation Wireless Access Technology,” Academic Press: Explains CORESET and BWP in downlink control.

• A Primer on Bandwidth Parts in 5G New Radio - Xingqin Lin, Dongsheng Yu, Henning Wiemann