Page 58 - Kaleidoscope Academic Conference Proceedings 2022

2022 ITU Kaleidoscope Academic Conference

SRv6 protocol to deliver the overlay anycast packets to the 2.1 Service resource discovery
appropriate computing service gateway. By establishing a
network address identification mechanism, the readability Since computing services deployed on highly distributed
and manageability of the control plane are improved when heterogeneous infrastructures such as edge computing nodes
configuring the forwarding policy and mobility policy of and cloud computing nodes are often dynamically
each computing service. In addition, the control plane will instantiated and de-instantiated according to indicators such
also obtain the information of computing service instance as the number of users, occupancy cost, energy consumption,
and network status to support for the load balancing decision and estimated service quality, service load balancing through
of anycast service. In summary, our contributions are as the network control plane assumes the network control plane
follows: can efficiently and dynamically discover the currently stand-
by service instances.
1) Identification mechanism of network address: The
globally unique anycast IP address assigned to the 2.2 QoE-oriented service instance selection
computing service instance will identify its service ID
information, so that the control plane can maintain the For users, the QoE of some computing services should be
mapping relationship between computing service and fully guaranteed, and the traditional best-effort policy is
anycast IP address. The computing service gateway will be unacceptable. For example, high latency and packet loss will
assigned multiple Segment Identifiers (SIDs), and each SID likely cause dizziness and nausea to users using XR
will also identify its corresponding computing service applications [9]. Therefore, the network control plane should
information. The SID directly reflects which computing be able to understand the demand information of computing
services are deployed behind the computing service gateway, services for network quality as well as node status, and filter
which enables the control plane to discover the computing and route user service requests to service instances with
services, and forward the computing service traffic to the better quality of experience.
appropriate computing service gateway. The user identity
information is identified in the IP address of the terminal to 2.3 Seamless mobility and service continuity
facilitate the control plane to provide users with seamless
service mobility and continuity. Since the network and node status is highly time-varying, the
service instances selected by the network control plane for
2) Enhanced control plane: To support the load balancing users will also change frequently. In addition, due to the
decision among computing service instances, the status of movement of the user, the gateway attached may also change.
each computing service instance is obtained through the Since frequent service interruptions are less acceptable than
interface between the control plane and the computing QoE degradation, the network control plane should ensure
infrastructure controller. that users do not switch to other service instances when they
use session-based computing services.
3) QoE-oriented load balancing decision: The control plane
can select one or more computing service instances for user 3. RELATED WORKS
terminals to provide high-quality services based on the
computing and network KPI requirements of the computing 3.1 Edge computing support in 3GPP system
service, the available area limitation of the computing
service, the status information of each computing service 5G New Radio (NR) provides support for enhanced Mobile
(eMBB),
Broadband
Ultra-Reliable
Low-Latency
instance, and the current connection status between user
terminal and each available computing service instance. Communications (URLLC), and enhanced Machine-Type
Communication (eMTC) services, and exposes various
The remainder of this paper will be organized as follows. In network capabilities to provide service quality assurance
support for edge computing applications. Uplink Classifier
Section 2, the load balancing requirements of the computing (UL CL) and IPv6 multi-homing mechanisms can easily help
and network convergence for ubiquitous computing services edge computing application traffic offload to specific edge
is summarized and in Section 3, some existing related work computing nodes [10]. A 3GPP system also helps User
is reviewed and a potential solution is elaborated in Section Equipment (UE) discover edge computing services by
4. The content of Section 5 is analyzing the relationship introducing an edge computing enablement layer on UEs and
between our research content and existing standards. Section edge computing servers [11], which utilizes the capabilities
6 is a conclusion of this paper.
exposed by the core network to provide mobility
management and service continuity for edge computing
2. REQUIREMENTS services. However, it only provides well-established
enabling tools for 3GPP systems and edge computing
Based on the background introduction above, and with applications and has not yet provided support for
reference to [7] and [8], the requirements of computing and heterogeneous networks and cloud-edge-terminal
network convergence infrastructure platform used to support collaboration scenarios. Besides, edge computing
large-scale distributed applications and services such as XR applications need to be built based on edge enabling layers,
and Vehicle to X (V2X) are summarized. which also brings a lot of extra work.

– 12 –