Traffic-Efficient Data Aggregation Scheme for Network Data Analytics in B5G Networks

To derive the accurate analytics in network data analytics function (NWDAF), lots of data from multiple sources (i.e., network functions (NFs)) should be collected, which can cause the network congestion. To mitigate this congestion, instead of delivering data whenever it generates, data can be aggregated. However, as a greater number of data is aggregated, NWDAF may have more delayed information (i.e., freshness of data may not be guaranteed). To balance the tradeoff between the network congestion and freshness of data, we propose a traffic efficient data aggregation scheme (TE-DAS) and investigate a data aggregation policy. Specifically, a data aggregation problem is formulated as constrained Markov decision problem (CMDP) with the objective to minimize the temporal network congestion while maintaining the freshness of data above a certain level. Evaluation results demonstrate that the derived data aggregation policy can reduce the temporal network congestion over 90% compared to no-aggregation policy while providing a sufficient freshness of data.