Edge Computing

The “Edge” in this context can refer to the base station itself (eNodeB, RNC, etc.), but also data centers close to the radio network (e.g. at “aggregation points”). One of the terms that has arisen is “Cloudlet”, which refers to a micro datacenter that resides at the edge of the Internet.

Edge functionalities will pave the way to deploy innovative applications and services

Regardless of the terminology, or specific technologies involved, edge computing will usher into existence a wave of new solutions and services as well as enhance existing services. Existing latency dependent services such as Virtual Reality (VR) will benefit from improved accessibility as VR dependency upon a wired connection for required capacity and data transit time is diminished with edge computing.

New services such as real-time data on demand are anticipated to be a major benefit of edge computing. Contextual data associated with new and improved services such as gaming, information services, and other apps will become increasingly more valuable with edge computing.

Accordingly, data services providers are expected to be a major beneficiary of edge computing. The deployment and increased usage of edge computing is likely to reinforce another dominant known as Data as a Service (DaaS), which includes raw data as well as data analytics on demand and decision support services.

There are many groups involved with ICT standardization in general that support edge computing.

For example, the China based Edge Computing Consortium was launched in 2016 with the support of ARM, Huawei, Intel, and various Chinese organizations including the Shenyang Institute of Automation of Chinese Academy of Sciences, China Academy of Information and Communications Technology. One of the primary aims of the consortium is to facilitate collaboration between ICT and the world of Operational Technology (OT).

Established by ARM, Cisco, Dell, Intel, Microsoft, and Princeton University in 2015, the Open Fog Consortium, emphasizes the deployment of edge computing for IoT networks, systems, applications, and services. One of their focus areas is on the need for an ecosystem supporting a hierarchy of elements between centralized cloud infrastructure and endpoint devices.

The primary standards body for MEC standardization is the European Telecommunications Standards Institute (ETSI), which created a MEC Industry Specification Group (ISG) in 2014 to address the opportunity.

ETSI published a document entitled Executive Briefing – Mobile Edge Computing (MEC) Initiative in September of that year along with an introductory technical white paper entitled Mobile-Edge Computing with contributions from Huawei, IBM, Intel, Nokia Networks, NTT DoCoMo, and Vodafone.

The first Mobile Edge Computing Proof of Concept (PoC) was introduced at MEC World Congress in 2016. ETSI ISG MEC has continued to support PoCs as a vehicle to both highlight industry developments in MEC and a driver of standardization needs.

MEC terminology was initially limited to Mobile (cellular) networks – it is now “Multi-access”

In 2017, ETSI renamed MEC to Multi-access Edge Computing in order to accurately reflect a focus on multiple types of access technologies including LTE, 5G, WiFi, and fixed access.

Due in part to the strong industry emphasis on cellular based wireless, many commercial groups and individuals continue to refer to MEC synonymously with Multi-access Edge Computing and Mobile Edge Computing.

ETSI’s ongoing work with MEC includes collaboration with other organizations such as its plans to coordinate with the Virtual Reality/Augmented Reality Association in the area of interactive VR and AR technologies.

MEC is anticipated to improve ICT infrastructure in cellular networks as it is highly complementary to LTE through its various stages of evolution as well as a driver for adoption of 5G. This is because edge computing helps optimize the allocation of capacity.

Mobile Edge Computing helps optimize the allocation of network capacity for LTE and 5G. While Multi-access Edge Computing makes no assumptions on the underlying radio infrastructure, 5G will be a clear beneficiary to MEC deployment. New radio and other 5G innovations will provide significant capacity gains as well as minimize network latency.

However, there is always a need to optimize available capacity to ensure supply is appropriately deployed to meet localized demand. In addition, MEC will help the business case for certain 5G enabled applications and services such as anytime and anywhere Augmented Reality (AR) and VR apps. For example, MEC will ensure that localized quality of experience demands are met for AR and VR end-users without compromising overall network performance.

Just like any new technology, MEC deployment will not be without its own set of challenges. There are many functional network components involved in MEC deployment. A few of the different deployment options to consider are:

• Bump in the Wire: Deployment scenario in which MEC platform is located between base station and mobile core network
• Distributed EPC: Deployment scenario in which MEC logically includes all or part of EPC components
• Distributed S/PGW: Deployment scenario similar to Distributed EPC with exception of SGW and PGW portions being deployed at edge and control plane functions (MME and HSS) are at operator’s network core.
• Distributed SGW with Local Breakout: This scenario allows users to reach both MEC apps and operator’s core supported apps at same APN. In this scenario, both SGW-LBO and MEC apps may be hosted as virtual network functions in same MEC platform.
• Control/User Plane Separation (CUPS): All of the above options may involve distribution of EPC gateways at the network edge.

Each of the above deployment options has their own unique challenges, further amplifying the notion that simultaneous communications/computing planning and engineering associated with MEC will be a new and complex paradigm for carriers.

However, new 5G networks will seamlessly leverage MEC architectures originally deployed with LTE. There will be a relatively smooth transition from 4G to 5G resources including reuse of edge computing resources (servers, mini-datacenters, etc.) and transformation of LTE network functions to support of 5G MEC via software upgrades alone.

Computing at the edge of networks can take many shapes and forms depending on a variety of factors including the network types involved, computing resources, devices, applications, and services.

For example, Cisco coined the term “Fog Computing” to refer to cloud capabilities at the edge of networks. Releasing their vision for a decentralized computing infrastructure in January 2014, the company emphasized the importance of edge computing. More specifically, they focus on how distributed computing can complement centralized cloud resources as well as provide certain unique benefits such as localized data processing.

While much of the emphasis on Fog Computing is directed towards benefits for Internet of Things (IoT) networks, it supports other areas as well such as smart cities. Another edge computing term, Multi-access Edge Computing (MEC), also refers to decentralized computing as MEC offers app developers and content providers cloud computing capabilities and an IT service environment at the edge of the network.

One downside to purely centralized cloud architecture is that it introduces latency into the system, as machines have to wait for the data to be collected, sent, backhauled to the cloud, processed, sent back to the end-device, and acted upon.

While edge computing is clearly a dominant trend in ICT, centralized cloud infrastructure does not go away or become diminished in importance. Quite the contrary, we see a centralized topology as necessary for many ICT use cases such as the many “as a service” models.

However, edge computing will become increasingly important for many wireless applications as a means of gaining greater efficiency. Applications will include both human-oriented services as well as those autonomous M2M instances associated with IoT.

Edge computing will provide significant improvements to virtually any application that requires substantive computing resources and/or are latency sensitive. This includes various immersive technologies such as Augmented Reality (AR), Virtual Reality (VR), and Haptic Internet-related apps and services.

Edge computing in mobile/cellular networks brings an entirely new dimension with mobility. The “sweet spot” for Mobile Edge Computing (MEC) will be those applications and services that are computationally heavy, require a high degree of mobility, and are highly latency-dependent.

The MEC Sweet Spot: Applications and services that are computationally heavy, require a high degree of mobility, and are highly latency-dependent.

Some immersive technologies, such as AR, will be used extensively anywhere, anytime by consumers. However, it is a tougher argument to make for VR on the go for consumers, except for perhaps VR while in an autonomous vehicle, which is also another beneficiary of MEC.

However, there are a few key enterprise and industrial applications that will benefit greatly from MEC. For example, Cloud Robotics, Teleoperation, and Telerobotics all require very low latency in order to operate without errors.

Even a one-tenth of a second lag would cause many of these applications to behave erratically with potentially catastrophic consequences resulting if/when the robots involved are handling enterprise assets and/or in the vicinity of people.

An especially good example of the mobility aspect, Cloud Robotics entails robotics as a service in a cloud services model. This model will allow remote control of robots in an on-demand basis, regardless of the robot’s location, so long as there is good MEC coverage. On the other side of the communications, the person controlling the robot could theoretically be located anywhere there is good MEC coverage.

The person controlling the robot could even be wearing a haptic suit and causing the robot to move in a natural, human-like way. Assuming the robot has a high degree of articulated movement, and there is ample data speed available (via 5G), robotic movements on one side could mirror human movements on the other side.

As mentioned previously, self-driving automobiles are anticipated to be another major beneficiary of MEC as vehicles are inherently mobile and autonomous cars require a major amount of data processing for operations.

With edge computing, there are opportunities to capture and process real-time data at the edge. Some data may remain static and/or archived in the centralized Cloud, whereas data-in-motion may be dynamic and processed in real-time with corresponding real-time decision making.

Some data is passed to a centralized storage and processing area while other data is processed locally via edge computing

As mentioned earlier, MEC will facilitate opportunities for data service providers to offer DaaS offerings including raw data, data analytics, and decisions as a service. Unlike scenarios involving data services based on centralized data (often stored as static data in “data lakes”), many use cases will involve real-time data that is streaming through communications facilities at the edge of networks and locally processes by MEC platforms.

There will be a need for distributed data processing software including intelligence necessary to determine what data should be processed locally as opposed to data that should be transported to centralized platforms or even discarded. This will require software programming to identify data triggers to determine actions to take as well as programs to act on data captured including predictive analytics algorithms.

Longer term, there will also be a need for inclusion of AI algorithms to assist in the determination of local data behavior. This will be helpful both for both data management (determining what actions to take and assisting with big data analytics) as well as various aspects of data security.

The aforementioned opportunities (new and improved apps, network improvements, and data services) will not come without challenges.

Mind Commerce sees many opportunities for MEC driven economic expansion

Short-term challenges are largely aligning communications planning with computation planning. This simultaneous computing and communications planning and engineering is not an area that carriers are accustomed to handling. Accordingly, telecom network operators will need help from the IT world to determine a proper blending of ICT and operational technologies.

Longer term, there will be additional challenges, such as the ability to identify, authenticate, and authorize computing on a distributed basis with many nodes. One solution we see is the use of Blockchain and Artificial Intelligence to facilitate a trust environment among MEC servers and mini-datacenters.

Regardless of the challenges, edge computing is anticipated to be a key component of ICT infrastructure, driving many network improvements, opportunities for enhanced applications, and new opportunities to exploit data at the edge of networks.