A scheduler dynamically allocates resources, including time, frequency, and power, to serve user equipment (UEs) connected to a base station (gNB) on a 5G network. Delivering high-quality service (QoS) to diverse 5G use cases, such as ultra-reliable Low-Latency Communication (uRLLC), Massive Machine-Type Communication (mMTC), and Enhanced Mobile Broadband (eMBB), is critical.
Scheduling is not explicitly detailed in the 3GPP 5G standards. The standards define processes like measurement collection that can be used in scheduling algorithms. It’s up to the gNB maker to define and implement the scheduling algorithm. Some basic scheduling considerations include (Figure 1):
- Measurements like changing channel conditions can be used to make real-time dynamic adjustments as user demands and locations change. Power headroom reports that measure the difference between the transmit power requirements support power-aware packet scheduling.
- The Buffer status report lets the gNB know how much data is currently queued for transmission by a user.
- QoS requirements, including priority, delay, and throughput, are based on use cases and data types, such as real-time video versus noncritical data packets.
- Is the associated radio bearer a default or a dedicated radio bearer established to support a specific QoS level or flow requirements?
Figure 1. A scheduler uses a range of data inputs to arrive at a resource allocation scheme to maximize overall QoS. (Image: ShareTechnote)
Downlink factors
The downlink can dynamically allocate resources to various UEs. For example, the gNB may stop transmission to a UE if latency-critical data is in the buffer and immediately switch the schedule to send it.
The scheduler must ensure that each UE receives its guaranteed bit rate (GBR). It must deliver at least the minimum GBR required to support the needed level of service but not exceed the maximum GBR, as exceeding the maximum GBR can result in network congestion or exceed the UE’s processing capabilities.
Uplink factors
Uplink resources are also dynamically allocated to UEs. The most significant difference is that the scheduling algorithm must rely on reports sent by the UEs for the uplink. As in the case of downlinks, the gNB may stop transmission from a UE if latency-critical data is in the buffer and immediately schedule the latency-critical data.
The UE has an uplink rate control function that manages the sharing of uplink resources between logical channels. The radio resource control (RRC) gives each logical channel a prioritized bit rate (PBR) and a buffer size duration (BSD). For example, low latency applications typically have a small BSD since data is sent almost immediately. In a typical scheduler algorithm, the UE serves the logical channels in decreasing order based on their PBRs. That ensures the most critical PBRs are served first, and then the leftover capacity is shared between the remaining logical channels in their priority order.
Three-bit rate classifications
The required bit rate for different applications is an important factor in scheduling. Delay Critical GBR is a new concept in 5G. It joins the concepts of non-GBR and GBR used in 4G LTE networks. Some characteristics of these three types of bit rates include (Table 1):
- Non-GBR has the lowest scheduling priority and doesn’t produce a guaranteed flow bit rate (GFBR) or level of performance. Typical applications include web surfing, texting, and other non-time-sensitive uses.
- GBR has a higher priority and is used for time-sensitive applications like voice and video calls and vehicle-to-anything (V2X) systems. It provides a GFBR to the application for a higher QoS.
- Delay Critical GBR is the highest priority and provides significantly lower latencies for critical and time-sensitive applications like intelligent transportation and Industry 4.0 automation systems. While there’s no required level of performance, a scheduler algorithm might be programmed to deliver a 100 ms delay budget to GBR applications and a 10 ms delay budget for Dealy Critical GBR applications.
Table 1. The non-GBR and GBR classifications are common to 4G LTE and 5G schedulers, while the Delay Critical GBR is available only in 5G. (Table: LinkedIn)
Summary
Scheduling radio traffic is a critical function to ensure the required QoS in 5G networks. The scheduling algorithm is implemented in the gNB but is not explicitly defined in the 3GPP standards. The standards describe how a scheduler should operate and provide definitions and guidance, but it’s up to individual gNB makers to implement scheduling.
References
Intelligent scheduling for 5G user plane function placement and chaining reconfiguration, Computer Networks
Behind the new SI scheduling enhancement: why it’s needed and how it works, Ericsson
Flexible Reinforcement Learning Scheduler for 5G Networks, Universidade de Vigo
Quality of Service Based Radio Resources Scheduling for 5G eMBB Use Case, MDPI symmetry
Scheduling, 3GPP
Scheduling Algorithms for 5G Networks and Beyond: Classification and Survey, IEEE Access
What is Scheduling, Huawei
EEWorld Online related links
DSS lets 5G and LTE share spectrum
Private 5G: What is it? How does it work?
What are 5G’s, the different types, and how are they used?
What is precoding and what are the benefits?
What to expect from 5G-Advanced
