Edge Computing Applications and Infrastructure
Connected- Services, Cars, Machines, MLOps Platforms and more...
Modern technology continues to exponentially progress as in the past decades, and our society has come to rely more heavily on compute capabilities of networks, with an unprecedented amount of data being stored in the cloud and shared digitally between users.
We truly believe future infrastructure will be a platform powered with programmable AI/Inference capabilities.
We discuss here, edge computing which will have a significant and positive impact on the workloads and apps by reducing network latency, improving data management, and optimizing cost and reliance on the cloud that comes with the common centralized computing method.
Future Applications are Hungry for Scale:
20X Peak Data Rate (Gb/S)
10X User Experience Data Rate (Mb/S)
3X Spectrum Efficiency
2X Mobility (Km/hr)
10X Less Latency (ms)
10X Connection Density (Devices/km2 )
100X Network Energy Efficiency
50X Traffic Capacity (Mbit/s/m2 )
This advancement, particularly the new trend of internet use towards “bandwidth- intensive content,” has led to the development of Edge Computing Infrastructure and need for an underlying Software Stack to Manage, Deploy and Monitor the workflows.
And in particular, edge computing is highly significant for Connected Cars, Automation in Manufacturing, AI Apps enabling connected services, End to End MLOps Platforms and many more workloads and applications requiring low latency and high throughput Inference and AI capabilities.
Edge Computing: Solutions and Beneficiaries
Edge Computing architecture indicates the method of pushing internet computing and online storage to the edge of networks, as a conduit or an alternative to centralized cloud method that is commonly seen.
Typical software orchestrated architecture of Edge computing -
Covering few key beneficiaries of this architecture would be:
Intelligent Video Analytics
A key application of edge computing is the ability to create advanced technologies to analyze videos, including a scalable and privacy-aware framework. More explicitly, this could mean creating an elastic urban video surveillance system with more capabilities than previously. Currently, video surveillance systems in public places are supported by the cloud, with all footage and data stored centrally. However, there are many limitations to this model that can be solved by incorporating edge computing to build new video analytics frameworks for public safety and emergency services.
One specific challenge of cloud-based frameworks is that the cloud is not able to support the sheer amount of data needed for the reliable connectivity that emergency services would require in surveillance systems. Additionally, an extremely large bandwidth is necessary and if it is not expansive enough latency delays can be common and potentially be very harmful in public safety scenarios.
Intelligent video analytics created by using edge computing techniques for data processing and storage are a solution to these aforementioned problems. Edge computing enables surveillance to be real-time with no network latency issues, which would provide the ability to track footage immediately and store it for later analysis. Particular applications of this method include face recognition, which would be useful in robberies or kidnappings, and traffic surveillance, which could be crucial in deployment of emergency services as well as managing traffic violations.
Location Services
Another useful application of edge computing is providing various types of location services. In general, this means gathering data regarding one particular location (or multiple locations) and relaying this information to the user. This type of service is also able to track location so changes in position can be quickly retrieved and outputted to the user.
Location services do not only track geolocation, meaning geographical coordinates, but also provide an online or logical location, which would be the Cell ID or a similar identification type. MEC can be utilized to bring cloud capabilities to the radio access network (RAN), which is significant to increase bandwidth and decrease network latency, as is needed for location services to be effective and an improvement on currently used technologies.
Specifically, for location services, User Equipment (UE) is a central element to understand when examining how edge computing allows this technology to function properly and for the data collected to be valuable. UE refers to an device that an end-user uses to communicate, and often comes in the form of mobile telephones or personal computers. Location services can provide a list of all of a certain category of UEs associated with the mobile edge host or a list of all UEs in a given location. By connecting to global navigation satellite systems, edge computing methods can be used to improve the flow of data related to positioning and make location services more accurate.
Connected Services
Mobile edge computing to be discussed in significance of Internet connected Applications. Convergence of connecting people, things, data and processes which will transform human life, business and everything revolving around it. From this definition, it is clear that the IoT will become an increasingly larger aspect of modern society and online networks.
Edge computing model can support a wide range of specific technologies as edge computing based apps can by deployed horizontally across subsystems to improve data management, distribute queries, enhance collaboration and communication, and allow for heightened levels of networking and data governance. Additionally, edge computing applied to connected things will improve network speeds and reliability, leading to further developments in online and mobile technology.
Augmented (AR) and Virtual (VR) Reality Programs
Augmented reality (AR) is the use of technology to create an enhanced version of reality in which the user can still see at least parts of the surrounding real-world environment, while virtual reality (VR) involves the creation of an entire simulated world. However, both AR and VR can be further developed by applying edge computing methods.
There are several aspects involved in transforming weak AR or VR systems into strong ones. These include “hand gesture recognition, eye movement tracking, touch feedback, and brain-computer interface.” For these technologies to be effective and reliable, there must be a reliable online framework and system of data storage. This is where edge computing methods come into play in improving both AR and VR.
This strong-interaction VR is often referred to as Cloud VR, and similar systems are also being developed for AR, although augmented reality technologies do not generally require as large of a data processing ability. In Cloud VR, image quality is improved, network latency is decreased, which allows for motion-to-picture (MTP) to be decreased and the dizzying effect on the user to be lessened, and motion-sensing capabilities are heightened. Finally, projection and coding techniques of AR and VR have also been improved by applying edge computing, which is highly significant for the creation of scientific models.
Edge Computing is valuable in improving the online experience of regular users by providing improved content delivery (via CDN) and quality of service (QoS), as well as increasing the efficiency of networks and applications. with an emphasis on virtual, cloud-based environments and large expanses of network infrastructure. These aspects are supported by edge computing, as it is a method that opens up new possibilities regarding highly dense and efficient online applications.
There are three main types of edge computing that are important to keep in mind when examining its various applications. These categories are local devices, localized data centers, and regional data centers, listed from smallest to largest in terms of data storage. Local devices generally fill a specialized role and are suited for home and office use, with cloud gateways and mobile phones being good examples of this type of edge computing use. Data centers, which can also be connected to the edge to support MEC, can be broken down into two separate classifications. Both provide significantly more data processing capabilities than local devices, but localized data centers only consist of one to ten racks, while regional data centers will always have more than ten racks, and thus will be substantially more powerful in terms of computing ability, data storage capacity, and data processing.
References/Inspiration:
White Paper of Edge Computing Consortium
http://en.ecconsortium.org/Uploads/file/20180328/1522232432850432.pdf
The Drivers and Benefits of Edge Computing
http://www.apc.com/salestools/VAVR-A4M867/VAVR-A4M867_R0_EN.pdf?sdirect=true
Edge Computing: Vision and Challenging
https://www.researchgate.net/publication/303890546_Edge_Computing_Vision_and_Challenges
Multi-Access Edge Computing (MEC) Applications
Performance Optimization in Mobile-Edge: Computing via Deep Reinforcement Learning
https://arxiv.org/pdf/1804.00514.pdf
Mobile Edge Computing Use Cases & Deployment Options
https://www.juniper.net/assets/us/en/local/pdf/whitepapers/2000642-en.pdf
Survey on Mobile Edge Computing: The Communication Perspective
https://arxiv.org/pdf/1701.01090.pdf
Understanding Information Centric Networking and Mobile Edge Computing
Location Service in Mobile Edge Computing
https://ieeexplore.ieee.org/document/7993865
Mobile Edge Computing; Location API
https://www.etsi.org/deliver/etsi_gs/MEC/001_099/013/01.01.01_60/gs_MEC013v010101p.pdf
MEC and Edge Computing: The Importance of Location
http://www.5gsummit.org/seattle/docs/slides/SenzaFili_IEEE_Seattle_5G.pdf
Enabling Live Video Analytics with a Scalable and Privacy-Aware Framework
https://www.privacyassistant.org/media/publications/tomm2018.pdf
Elastic Urban Video Surveillance System Using Edge Computing
http://www.cs.umsl.edu/~pan/papers/smartiot2017.pdf
Out of the Fog: Use Case Scenarios
https://www.openfogconsortium.org/wp-content/uploads/Video-Surveillance-Use-Case.pdf
Cloud VR Bearer Networks
Wireless VR/AR with Edge/Cloud Computing
http://5gsummit.ru/wp-content/uploads/2017/06/MEC_5G_Moscow_Rev_1.pdf
Introduction to Edge Computing in IIoT
https://www.iiconsortium.org/pdf/Introduction_to_Edge_Computing_in_IIoT_2018-06-18.pdf
IoT Future in Edge Computing
https://www.researchgate.net/publication/312068521_IOT_future_in_Edge_Computing
Smart Energy Efficient Sensing For Iot Edge Computing With Mobile Agents



Looking forward for more articles as such. Good info.