Semi-Persistent Scheduling

Description

Semi-Persistent Scheduling (SPS) is a radio resource management mechanism defined in 3GPP specifications for LTE and NR that strikes a balance between fully dynamic scheduling and fully static scheduling. In dynamic scheduling, the network sends a downlink control information (DCI) grant on the PDCCH for every single uplink transmission or downlink reception, which provides maximum flexibility but incurs significant control channel overhead. SPS, in contrast, configures a UE with a recurring pattern of radio resources (specific subframes/slots and resource blocks) for a particular logical channel, typically associated with a periodic service like Voice over IP (VoIP). Once activated via a single DCI message, the UE automatically uses the pre-defined resources at the configured interval without needing further grants, dramatically reducing PDCCH load.

The operation of SPS involves configuration, activation, and deactivation phases. First, the network configures the UE with SPS parameters via RRC signaling. These parameters include the SPS interval (e.g., 20 ms for VoIP), the specific time-domain offset, and the frequency resources. This configuration is tied to a specific Cell Radio Network Temporary Identifier (C-RNTI) and a configured grant configuration index. Activation is performed dynamically via a special DCI (format 0_0/0_1 for UL, 1_0/1_1 for DL in NR) scrambled with the CS-RNTI (Configured Scheduling RNTI, which evolved from the SPS C-RNTI concept). This activation DCI points to the pre-configured SPS configuration and effectively 'starts the timer.' Upon receiving it, the UE begins using the periodic resources according to the configured pattern.

For uplink SPS, the UE transmits on the pre-allocated resources without waiting for a UL grant. For downlink SPS, the UE monitors the PDSCH in the pre-allocated resources. The network can also send dynamic grants that override the SPS resources for a particular occasion, providing flexibility. SPS is deactivated either explicitly by a DCI scrambled with CS-RNTI with a specific field set to indicate deactivation, implicitly after a configured number of empty transmissions (in UL), or upon RRC reconfiguration. Key components include the CS-RNTI (which uniquely identifies UEs with active SPS configurations), the SPS configuration parameters in RRC, and the specific DCI formats and scrambling rules used for activation/deactivation.

In 5G NR, SPS concepts were enhanced and generalized under the umbrella of 'Configured Grants' for uplink, which includes two types. Type 1 configured grant is similar to traditional SPS where all parameters are provided by RRC. Type 2 configured grant involves RRC providing some parameters and a DCI (using CS-RNTI) activating the periodic resources. This provides more dynamic control. SPS remains crucial for NR services requiring ultra-reliable low-latency communications (URLLC) and periodic industrial IoT traffic, where minimizing control latency and guaranteeing resource availability are paramount. The mechanism offloads the PDCCH, reduces scheduling latency for periodic packets, and conserves UE battery life by reducing the need for continuous blind decoding of control channels for every transmission.

Purpose & Motivation

SPS was developed primarily to efficiently support services with predictable, periodic traffic patterns, most notably Voice over IP (VoIP). In early LTE deployments, it was observed that using fully dynamic scheduling for VoIP—where a small packet (e.g., a voice frame) is generated every 20 ms—was highly inefficient. Each 40-byte voice packet would require a separate DCI grant on the PDCCH, which could be 40-50 bits in size. This meant the control signaling overhead could approach or even exceed the data payload itself, severely limiting VoIP capacity on the cell.

The problem SPS solves is this control channel overhead bottleneck. By pre-allocating resources, SPS eliminates the need for a grant for every single packet transmission. This dramatically increases the number of VoIP users a single cell can support, as the limiting factor becomes the available PDSCH/PUSCH resources rather than the PDCCH capacity. It also reduces scheduling latency because the UE does not need to wait for a grant to arrive; it already knows when and where to transmit or receive, which is beneficial for maintaining consistent low latency for real-time services.

Historically introduced in LTE Release 8, SPS addressed a key challenge for making LTE a competitive technology for voice services. As networks evolved to support more IoT and M2M applications with periodic reporting (e.g., smart meters, sensor data), the utility of SPS expanded beyond VoIP. In 5G NR, the principles of SPS were refined into the more flexible configured grant framework to support a wider range of URLLC and periodic traffic scenarios with even stricter latency and reliability requirements. SPS represents a fundamental optimization in cellular design: trading a small amount of scheduling flexibility for massive gains in control efficiency and latency performance for predictable data flows.