This is the second article in a three-part series. The first article, which examined the driving forces for QoS in mobile networks, is available here. This article outlines the technical requirements of the LTE QoS framework, including the critical parameters and policy controls, and how these fit into the LTE architecture.
QoS involves a framework consisting of numerous standards that are widely used in communications networks to manage traffic in ways that differentiate subscribers and services to improve overall performance. QoS enables network operators to isolate traffic into flows based on attributes, such as traffic type (voice, video or control) or application needs (throughput, latency and/or jitter), and then to transport each flow accordingly. QoS also addresses subscriber SLAs by prioritizing traffic for critical users and applications, providing deterministic latency and jitter (required by real-time applications), and minimizing network congestion.
LTE Network Architecture Overview
Figure 3 illustrates a typical LTE network. Unlike other mobile broadband technologies, such as GSM that are circuit-switched, LTE is based on IP packets. But a packet-switched LTE network must also support a variety of legacy access technologies, including 2G GSM, 3G UMTS, WCDMA, CDMA2000 and Wi-Fi.
Figure 3: Simplified View of the LTE Network Architecture
Different services are carried over the radio interface to the evolved base station, or eNodeB (ENB) that connects with user equipment (UE) on one side, and on the other side with the core network, or evolved packet core (EPC). The EPC is then connected to the external IP networks, such as the IP multimedia subsystem (IMS). This LTE architecture is flat because the traditional control functions are collapsed into EPC and the radio network controller (RNC) functions (in a 3G network) are incorporated into the LTE eNodeB. This architecture simplifies the network infrastructure, reduces the number of network nodes by removing multiple hops and protocol translations, and makes the network faster and more cost-effective.
The LTE architecture supports peer-to-peer (eNodeB to eNodeB) connections, which lowers latency and minimizes round-trip delay, enabling LTE to offer throughput rates beyond 100Mbit/s and a latency of around 20ms. To support this complex functionality, however, more intelligence must be built into the eNodeB than in its predecessor (3G NodeB). In addition, comprehensive QoS can be achieved only by applying the techniques to both the user plane (user applications) and the control-plane (network control traffic).
LTE QoS Concept
The 3RD Generation Partnership Project (3GPP) is a standards body that develops the specifications needed to enable multivendor interoperability and multi-network roaming capabilities. Based on LTE’s use of IP over Ethernet transport, which leverages DiffServ for traffic management, 3GPP has defined the LTE QoS framework using the Evolved Packet System (EPS) bearer model.
EPS Bearer Model
A bearer is a traffic separation element that enables differentiated treatment of traffic based on its QoS requirements, and provides a logical path between UE and a gateway. All flows mapped to a single bearer receive the same packet-forwarding treatment (e.g. scheduling, queue management, rate shaping, link layer configuration, etc.) between the UE and the gateway.
A bearer can be classified based on its QoS requirements as either a default or a dedicated bearer, as shown in Figure 4. When a mobile device first attaches to an LTE network, it is assigned a default bearer, which is associated with the UE’s IP address. The default bearer does not have a bit rate guarantee and offers only best-effort service. A dedicated bearer acts as another bearer on top of the default bearer, and provides a dedicated tunnel to give appropriate treatment to specific services.
Figure 4: Default and Dedicated Bearers
A dedicated bearer is further classified as a guaranteed bit rate (GBR) EPS bearer or a non-GBR bearer. GBR has dedicated network resources and is needed for real-time voice and video applications. A non-GBR bearer does not have dedicated resources and is used for best-effort traffic, such as file downloads.
LTE QoS Parameters
Each bearer uses a set of QoS parameters to describe the properties of the transporting channel, such as bit rates, packet delay, packet loss, bit error rate and scheduling policy. The four key parameters are outlined here.
QoS class indicator (QCI): QCI specifies the forwarding treatment (e.g. scheduling weights, admission thresholds, queue management thresholds, link-layer protocol configuration, etc.) that the user-plane traffic receives between the UE and the gateway. The QCI specification with corresponding parameters and common applications are presented in Table 1. For example, a real-time gaming application with “3” in the QCI field will be given a priority of 3, and nodes along its path will need to guarantee a delay less than 50ms.
Table 1: QCI Values and Associated Parameters
Allocation and retention priority (ARP): ARP specifies the forwarding treatment for the control-plane traffic that the bearers receive. ARP enables bearer establishment and modification, as well as connection setup and release. For example, ARP can be used by the EPS to decide which bearer should be released during resource limitations or traffic congestion.
Maximum bit rate (MBR): MBR is applicable only for real-time services and is defined for GBR bearers. MBR is the bit rate that the traffic on the bearer may not exceed.
Guaranteed bit rate (GBR): GBR specifies the bit rate that the network guarantees (e.g. through the use of an admission control function) for that bearer. In 3GPP Release 8 and beyond, the MBR must be set equal to the GBR; that is, the guaranteed rate is also the maximum rate that is allowed by the system.
Policy Control and Assignment of QoS Rules to Flows
The network operator controls the mapping of packet flows onto a dedicated bearer and determines the QoS level of the dedicated bearer through policies. These policies are provisioned into the network policy and charging resource function (PCRF), also known as the policy controller, implemented in the PDN gateway.
The policy controller typically filters packet flows using five parameters, referred to as IP five-tuple: source IP address; destination IP address; source port number; destination port number; and protocol identification (TCP or UDP). PCRF determines the QoS requirements based on three criteria: application requirement; subscription information; and policy of the operator. Based on the decision made by PCRF, a dedicated EPS bearer may be established as shown in Figure 5.
Figure 5: QoS Assignment on EPS Bearers
Each packet entering the system is provided with a tunnel header, as shown in Figure 4, that contains the bearer identifier to enable the network nodes to treat it with appropriate QoS. These QoS parameters for user-plane and control-plane are preserved in the LTE transport network using appropriate transport layer protocols. Figure 6 illustrates the protocol stack implemented in the LTE transport network. For example, the GTP protocol between eNodeB and S-GW is transferred over UDP/IP.
Figure 6: LTE User-Plane and Control-Plane Protocol Stacks
For more informaiton visit www.lsi.com.
