Browsing by Author "Kapoor, Shubham"

Challenges in managing cellular networks have grown in magnitude over the years. Upcoming 5G networks will further complicate the problem by introducing a larger number of scattered Network Elements (NEs) and functions than the prior cellular generations. Due to investment costs and user retention reasons, older technologies are not completely scraped out while new technologies are being introduced in the network. This results in a large, heterogeneous and complex network, which makes the data used to manage the network i.e. Management Plane (M-Plane) data, a big data candidate for the telcos. The conventional centralised Network Management Systems (NMS) will face fundamental scaling challenges in processing this big data, be it its collection, storage or quick analysis. In this thesis, we propose a novel concept of Quality of Monitoring (QoM) classes, which could be used for mobile edge compression of M-plane data. QoM classes specify the quality by which M- Plane data can be collected from the NEs. Quality here specifies the amount of relative information loss acceptable by network management applications consuming M-Plane data. The best QoM class with zero information loss could be assigned to monitor most critical NEs, while inferior QoM classes with some degree of information loss could be assigned to monitor auxiliary NEs. The proposed solution is based on the Publish/Subscribe paradigm for data delivery. The Publish/Subscribe paradigm makes the solution scalable and efficient by collecting data only once and providing flexibility to update QoM class subscription on the fly. The solution aggregates raw M-Plane data into a smaller set of the operator-defined performance metrics. This data is further compressed by removing portions of data with redundant or small information content. This helps the proposed solution to achieve reduction in the complexities of 3Vs (Volume, Variety and Velocity) of M-Plane data. The solution achieves quick processing of M-Plane data while saving the computational, bandwidth, storage and energy resources of the cellular operator.