As you have read in previous posts, 5G will bring vast innovations to the industry enabled by various improvements to the next generation of mobile networks. One of these improvements is something known as Multi-access Edge Computing (aka edge computing), but what is edge computing and how does it affect the end-users experience?
If you had that question in mind, this article will hopefully clear some confusion you may have.
Before we define edge computing, let’s visualise our current situation when we are navigating the web, over a mobile connection:
In a current typical scenario we would see something similar to above; where the data network (e.g. requested web application such as YouTube) is situated far away geographically speaking, resulting in higher latencies and increased strain on bandwidth.
Most of the applications we use commonly will be placed in distant data-centres with 100s of miles between the end-user and the application being used — meaning we are not getting the optimal connection quality which we could achieve. Limiting the use-cases and types of applications which can be possibly addressed.
The current method of deploying applications is suitable for current use-cases and consumer habits, however this trend is quickly changing and rapidly advancing use-cases/requirements mean we need to start considering deploying applications in a different way.
So now you have seen the current situation we have; applications hosted far away from the end-user; resulting in high-latency and under-strain bandwidth.
How can we fix this?
The natural progression of today’s application deployment is to bring the application being used closer to the source of data (the end-user). This means that the data has less distance to travel and doesn’t need to traverse many transit networks.
In the diagram above you can now see that the application before hosted in the Data Network (DN); is now being hosted in the Local Area Data Network (LADN) — closer to the user by being situated in close proximity of the point of connection to the network, in this case the mobile antenna.
The 5G core network (5GC) can intelligently be aware of the presence of applications which are hosted at the edge. This means that the user’s data (UPF of the 5GC) will automatically be routed to the closest hosting of the applications which they want to use. The user will start to see much higher speeds and most importantly ultra low latency communications.
In the 5G architecture user packets are sent via the User Plane Function (UPF) which is a function of the 5GC. The UPF must be able to identify the LADN vs DN and determine which is best for the end-user’s requirements (such as ultra low-latency). To do this, the UPF can communicate with an Application Function (AF) of the 5GC and the AF will start to influence the UPF to steer the packet traffic towards the LADN.
Below you can see a simplified reference 5G architecture; to visualise these different components in relation to the UE (user equipment).
The MEC deployment (managing the LADN) will communicate with the Policy Control Function (PCF) of the 5GC to request traffic steering by identifying the user traffic which needs to be steered. The PCF will then will then transform that request into policies that will affect the user’s PDU (Protocol Data Unit) session and provides the routing rules to the Session Management Function (SMF). The SMF will then identify a UPF which is present in the PDU of the user.
The UPF will then be able to route the user’s data to the relevant LADN.
But how can we identify if a user should be consuming an application hosted in a LADN vs DN?
We can identify applied traffic steering rules by many different metrics; some examples of this are UE identification (is it a mobile phone user or a connection vehicle), where is the UE geographically located (what LADNs are close to them?) or S-NSSAI (used to select and identify available slices in the network).
In this article you have seen what edge computing is and why we need it for new innovative use-cases. In addition, how the 5G network allows edge computing applications to be deployed; some architectural details have been omitted for simplicity and can be seen even lower level in future articles.You have also seen some of the use-cases which are enabled by this innovation; which may have not been possible in previous methods to deploy our applications.