SON function performance prediction in a cognitive SON management system

Abstract

As a reply to the increasing demand for fast mobile network connections the concept of Self-Organising Networks (SONs) has been developed, reducing the need for humans to execute Operation, Administration and Maintenance (OAM) tasks for mobile networks. However, a SON contains functions which are provided by different vendors as black boxes, making it hard to predict the performance of the network, especially under untested configurations. Since Mobile Network Operators (MNOs) have to fulfil rising mobile network performance demands while reducing costs at the same time, it is crucial to gain a better understanding of the network behaviour to allow a costneutral performance improvement while simultaneously reducing the risk of network misconfiguration and service disturbance. In this paper an approach is introduced to enhance SON Management models with cognitive Machine Learning (ML) methods. Therefore, the simulated behaviour of three different SON Functions is analysed and described by a Linear Regression (LR) Model. In a second step, performance data of network cells are analysed for similarities using k-Means Clustering. The findings of these two steps are then combined by fitting the models onto smaller clusters of cells. Finally, the utility of these models for predicting the performance of the network is evaluated and the different stages of refinement are compared with each other.

Correlation-Based Feature Mapping of Crowdsourced LTE Data

“…Most of the contributions include SON function experiments performed with an internal LTE simulator made by Nokia. It has been widely used in SON research [54,145,144,84,137]. The simulator setup used in this thesis comprises 20 LTE base stations with 32 LTE macro cells covering an urban area with a radius of 5 km and with 1000-5000 users depending on the scenario.…”

“…Lohmüller et al [85] have presented an adaptive SON function management mechanism that analyzes context-specific measurements in order to define the effect of a configuration set on the network. Moreover, Lohmüller et al [84] have studied whether the unknown configuration sets of SON functions can be estimated reliably.…”

“…For this purpose, they have proposed methods such as association mining, dimensionality analysis, and manifold learning. Lohmüller et al [84] have clustered similarly behaving network cells with respect to a set of KPIs while specific SON functions has been active. Ciocarlie et al [29] have presented a topic modelling approach to post-action analysis.…”

“…Characterization of SON functions has been studied in view of the dynamic mapping of operator goals for SON function configurations [55,85]. Also, some approaches have enriched the predefined SON function models by gathering experience of their actual performance and impact on the network [85,84]. Other SON-related studies have specifically focused on the verification and coordination of the SON functions, either to avoid degradations [137,145,28] or to analyze joint effects [59,68,14,54] of the functions.…”

“…The time series-based metadata creation method of SON functions is a combination of event detection and association rule learning among detected events. With respect to measurement-based pattern mining, Lohmüller et al [84] presented a method in order to characterize cells while single SON functions were active. Ciocarlie et al [29] stored measurement-based patterns in order to characterize topology changes (the addition of new cells) in the network [29].…”