Intent preparation

The customer's requirements could be extracted through dialogue with INTEND tools (inChat and inSwitch) which will make suggestions for specific network configurations and workload placements based on QoS estimations. The customer objectives (needs) will be specified as high level intents formatted according to TM Forum intent ontology definition.​

Below, we illustrate the decisions to be made by the tools to fulfill a customer's (or an application's) need for low latency. The available options are to a) configure the network to decrease the latency (setting up a slice); b) move the application closer to the user, or; do both. To decide on which would be a viable option there is a need to estimate (or ideally measure) current latencies in different parts of the communication path as outlined in the sketch below. 1Total latency; 2Radio access network latency; 3Transport network Latency; 4Internet latency, and; 5Local edge data center latency. Further an estimation of possible latency reductions tied to the different parts needs to be done to make an informed decision on which option to choose.

To the lower left we have provided a simplified latency estimation model for this purpose. Fill in values for the different fields and push the Decide intent(s) needed button in the form below to get a calculation of current latency and suggested action(s) / intent(s) displayed.​ Note that this model is a simplification. Still, for this MVS it suits the purpose of selecting among the available options.​​

Latency Legs Diagram

Latency estimation model

The maximum acceptable total latency for your application, including all latency elements.
Latency from User Equipment (UE) to the gNodeB (base station). This is usually low since the closest gNodeB is probably close.
Latency from the base station (gNodeB) to the breakout point.
Latency from the breakout point to the application server.
Time taken by the application server to process data and send results back to the client application.
Expected reduction in latency from using network slicing.
Expected reduction in latency by using a local data center.

About This Tool

This tool will decide which of the three possible actions need to be taken to meet the required total latency. The possible actions are:

  1. It is only needed to configure a network slice.
  2. It is only needed to move the application to a datacenter that is closer to the handsets (UE).
  3. It is needed to do both a) and b).

For b), the local datacenter that the application must be moved to may be drawn from a set of local datacenters that would all make it possible to meet the latency requirement specified.

Hopefully, the tool owners will be able to calculate the asked input for this model based on the offered synthetic data. Calculations could be done by AI agents in combination with common knowledge that resides in an LLM, or algorithmically (for simplicity in the MVS). This is something the tool owners need to dwell over. If some of the numbers can't be retrieved/deduced, we need to add something to the synthetic data or make some simplifying assumptions allowing algorithmic calculations.​

Next Step